Hi Creatives! 👋

This week: AI tools are moving closer to the places where creative work actually happens.

Magnific is bringing its tools into software like Premiere Pro, After Effects, Photoshop, Figma, Final Cut Pro, and DaVinci Resolve. Midjourney V8.1 is now the default. Luma is turning saved prompts and references into reusable workflows. And AI-assisted films are showing up in more serious production and storytelling spaces.

The bigger shift is not just better outputs. It is AI becoming part of the creative process itself.

For creatives, that means the work is becoming less about testing one tool in isolation and more about building a clear workflow: what you generate, what you review, what you keep, what you reject, and how you make sure the final piece still feels intentional.

This Week’s Highlights:

• Magnific Brings Its AI Tools Into Creative Software

• Midjourney V8.1 Is Now the Default

• Luma Skills: AI Creative Workflows Are Becoming Reusable Systems

• Runway, UNHCR, and AI-Assisted Storytelling

Magnific Brings Its AI Tools Into Creative Software

Earlier this month, Magnific introduced its MCP integration at Upscale Conf SF, alongside Agents and Flows. The announcement was about bringing Magnific into the AI clients creatives already use, including Claude, ChatGPT, Cursor, VS Code, Gemini, and others. 

You can read our full breakdown of the three Magnific updates here

But the more interesting question is: what does that actually look like in a real creative workflow?

Magnific gave one answer with Candela, the first short film from Magnific Studios. The project was created using Claude connected to the Magnific MCP, moving through the full production process — from characters and locations to still frames, video, audio, and editing.

So instead of using AI for one-off image generation, the team used it more like a creative production assistant. Claude helped guide the process, while Magnific handled the visual and media generation side. Every asset was kept organized inside Spaces, which made the workflow easier to track from the first idea to the final cut.

For creatives, this is where it gets worth paying attention. Have you tried using AI this way yet — not just to make one image, but to help structure a whole project?

That is the bigger shift Magnific seems to be pointing toward: AI as part of the creative system, not just the output machine.

📹️ Magnific

And now, Magnific is expanding beyond its web platform with a new set of plugins for creative software, including After Effects, Premiere Pro, Final Cut Pro, DaVinci Resolve, Photoshop, and Figma.

The update means creators can now use Magnific’s AI tools more directly inside the apps where they already edit, design, and produce content. Depending on the plugin, users can generate assets, upscale images, remove backgrounds, reframe visuals, relight scenes, and pull from Magnific’s feed or stock library without leaving their workspace.

For creative teams, this is less about one new feature and more about where AI tools are heading. The workflow is becoming more integrated. Instead of jumping between browser tabs, downloads, and editing timelines, AI generation is being placed closer to the actual production process.

This could be especially useful for video editors, designers, social content teams, and brand creatives who need to test ideas quickly while still keeping their projects organized.

The thing to watch is how well these plugins hold up in real production. Speed is helpful, but creatives still need clean outputs, flexible controls, predictable credit usage, and enough quality for client-facing work.

The bigger takeaway: AI tools are becoming less like separate destinations and more like layers inside the software creatives already use.

Know more about the plug-in here.

Midjourney V8.1 Is Now the Default

🖼️ Midjourney

Midjourney has officially made V8.1 the default model, following its earlier V8.1 update that improved sharpness, image quality, and support across both the web app and Discord.

For creatives, this update is less about a brand-new look and more about making AI image workflows faster, cleaner, and easier to repeat.

What changed

• V8.1 is now the default model
  If you do not choose a specific version, Midjourney will now use V8.1 automatically.

• Faster generations
  This makes it easier to test more ideas, compare art directions, and move through visual concepts with less waiting.

• Better prompt details
  V8.1 is designed to follow more specific prompts, which helps when working with product details, styling notes, lighting, props, and composition.

• Sharper image quality
  The previous V8.1 update already focused on improving image sharpness, especially for style references, moodboards, and HD images.

• HD images are more useful now
  Higher-resolution outputs can help with pitch decks, social assets, thumbnails, and early campaign visuals.

What to watch for

• Old prompts may behave differently, so test them before using them in client work.

• V7 may still be useful for certain features, especially workflows that rely on stronger reference control.

• HD is helpful for polished outputs, but standard generations may still be enough for early exploration.

• Document your model version, references, and settings if you need to recreate a look later.

Our insight 💡 

Read full details here.

Luma Skills: AI Creative Workflows Are Becoming Reusable Systems

Luma’s new Skills feature points to where AI creative tools are heading: less one-off generation, more repeatable workflows.

Instead of rebuilding the same prompt, references, settings, and creative rules each time, Skills let users save a process and run it again with new assets. Think of it like saving a mini production system, not just a prompt.

Why it matters

For creatives, the useful part is consistency.

A Skill can help with:

• Repeating a visual style across campaigns

• Creating product lifestyle images with the same art direction

• Keeping characters, materials, or scenes more consistent

• Handing off approved workflows to a team

• Reducing the “wait, how did I make that?” problem

This is especially helpful for brand teams, social creators, agencies, and small businesses that need to create variations without starting from scratch every time.

What to watch

• AI workflows may become part of brand systems

• Prompts and references may become creative documentation

• Teams may build libraries of approved AI processes

• Consistency will still need human review

• Creative direction becomes more about setting clear rules, not just generating outputs

Insight

Read the full details here.

Runway, UNHCR, and AI-Assisted Storytelling

Runway recently shared how director upperfast and production company ground6 used AI to help create UNHCR’s FIFA World Cup film, “More than a Game.”

The film features eleven footballers with refugee or displacement backgrounds, reimagined as their younger selves. The goal was to show how safety, opportunity, and belonging can change the future of a child.

What makes this case interesting is not just that AI was used. It is that AI helped make a story possible that would have been difficult to produce through a traditional shoot.

Coordinating players across different countries, recreating childhood likenesses, managing travel, crews, and budget would have been a huge production challenge. Instead, the team used archival research, creative direction, AI-generated visuals, and Runway’s video tools to build the film.

What to watch for

• AI as part of the production pipeline
  This was not a one-prompt video. The team still needed research, direction, editing, post-production, and final finishing.

• The rise of the AI artist role
  AI artists helped with likeness, visual consistency, jersey colors, and shot execution. This points to a new hybrid creative role between researcher, art director, compositor, and prompt strategist.

• Human texture still matters
  The final film was printed onto 16mm and scanned back to digital, adding grain and a more grounded visual feeling.

• Ethics and disclosure are key
  UNHCR clearly noted that the film and images were created with the assistance of AI tools. For a campaign involving real people, childhood likeness, and displacement, transparency matters.

Insight

Read the full article here.

💡 Insight

Tools like Magnific, Midjourney, Luma, and Runway are not just asking creatives to generate more. They are starting to fit into editing timelines, design files, saved systems, production pipelines, and storytelling workflows.

That can be very useful for small teams, agencies, brand creatives, filmmakers, and social teams who need to move faster without losing the thread of the project.

But the human part still matters.

The more AI becomes part of the workflow, the more creatives need to protect the things that make the work good: taste, context, references, review, ethics, and point of view.

So maybe the real question this week is not “Which tool is the most impressive?”

It is: “Can this help us make better creative decisions?”

That’s it for this week, friends.

Hi Creatives! 👋

Something we keep thinking about:

AI is not just making creative work faster. It is making the creative process more visible.

• The references we choose.

• The prompts we keep.

• The outputs we reject.

• The client notes we document.

• The rules we create around likeness, IP, and brand safety.

All of these used to live in scattered folders, moodboards, and messy Slack threads. Now, they are becoming part of the workflow itself.

And that is where creatives need to pay attention.

The next advantage is not just knowing which AI tool is trending. It is knowing how to build a system around the tool that is repeatable, reviewable, and reliable.

• For designers and art directors, that means keeping organized references, approved palettes, character sheets, product angles, client-approved prompts, and “do not use” rules.

• For video editors and filmmakers, it means choosing tools that support revision, not just generation. The future of AI video is less “make me a clip” and more “help me control this shot, revise the scene, and keep the story consistent.”

• For social creators, tools like Meta’s Creator Assistant could make content planning more analytics-driven. Useful, yes. But the risk is that everyone starts following the same platform signals. Use the data, but keep your own voice.

• For agencies and creative teams, AI needs to be treated like a production system. Track the tool used, prompts, references, approvals, permissions, likeness review, and what is cleared for publishing.

• For storyboard and concept artists, the value is not just making images. The value is visual judgment: story clarity, camera language, continuity, emotional framing, and knowing what actually serves the idea.

AI can create more options. But knowing which option actually works is still a human creative skill.

This Week’s Highlights:

• Runway x Lionsgate: AI Film Is Getting More Serious

• Luma’s Ray3.2 Shows Where AI Video Is Heading

• Meta Adds AI Support for Creator Strategy

• ChatGPT Memory Is Getting Smarter

• Pixel Studio Is Moving Into Gemini

• Martin Scorsese, AI, and the New Storyboard Room

Runway x Lionsgate: AI Film Is Getting More Serious

Lionsgate

Runway and Lionsgate are expanding their partnership, and this feels like one of those updates creatives should keep an eye on.

Lionsgate has taken an equity interest in Runway, and the two companies are now launching a joint development program to create new IP together. Their first planned project is a short-form episodic series using parts of Lionsgate’s existing IP and Runway’s generative AI models.

In plain English: AI video is moving beyond “cool tool demo” territory and getting closer to actual studio production.

What to watch for

• Short-form as the testing ground
  Studios may use shorter content first to test AI-assisted storytelling, visual styles, and audience response before applying it to bigger productions.

• AI inside the production workflow
  Runway’s tools are already being used across pre-visualization, storyboarding, and final-frame production. This partnership suggests AI may become part of how content is developed from the beginning.

• Existing IP + AI gets complicated
  When a studio uses generative AI with its own library of characters, worlds, and franchises, it raises bigger questions around ownership, credit, consent, and creative labor.

• More pressure on creative roles
  AI can help make ideas faster and cheaper to test, but that also means creatives need to be clearer about their value: taste, story, direction, editing, and authorship.

This means that for independent filmmakers, designers, art directors, and content creators, this update is both useful and a little uncomfortable.

The useful part: AI video tools can help smaller teams create pitch films, mood videos, campaign concepts, and visual tests that used to require a much bigger budget.

The uncomfortable part: big studios also have the advantage of large IP libraries, legal teams, and distribution. So while AI gives more people access to production tools, it may also help large companies move even faster.

Our insight 💡 

Read full details here.

Luma’s Ray3.2 Shows Where AI Video Is Heading

Luma had two updates this week that are worth reading together.

First, the company joined Human After All, Webedia-Elephant’s new AI Creator Studio in Paris, alongside partners like Google and ElevenLabs. Then Luma introduced Ray3.2, its latest AI video model focused on more control, longer clips, and production-ready output.

Together, these updates point to a bigger shift: AI video is moving from simple generation toward a more directable creative workflow.

GPT Image Generated

Luma just joined Human After All, a new AI Creator Studio by Webedia-Elephant in Paris.

What makes this interesting is the approach: start with a real human performance, then use AI to transform the world around it.

At the launch, Luma showed live talent being transformed in real time with Ray3.14. The actor stayed in the frame. The environment, style, and visual world shifted around them.

Why it matters

This feels like a useful signal for where AI video is heading.

Less “type a prompt and hope it works,” more:

• Start with real footage

• Keep the performance and motion

• Use AI to change the world, wardrobe, style, or scene

• Build a more flexible production workflow

For creatives, that could be helpful for pitch films, music videos, campaign concepts, previsuals, and social-first storytelling.

What to watch

A few things stand out:

• AI creator studios are becoming physical spaces, with training, open office hours, and production support.

• Video-to-video workflows are getting more practical, especially for keeping human performance while changing the visual direction.

• Creator programs are becoming creative R&D labs, where tools are tested inside real projects instead of polished demos.

• Rights and consent still need attention, especially when AI transforms someone’s face, voice, body movement, or likeness.

Read the full details here.

Credits to Luma

Ray 3.2 gives creators more ways to shape a video across time instead of relying on one prompt to carry the entire scene.

Key updates include:

Read more here.

Meta Adds AI Support for Creator Strategy

📷️ Meta

Meta is bringing more AI into the Facebook creator dashboard with its new Creator Assistant.

Instead of only checking charts and metrics, creators can now ask questions about their content performance, audience behavior, and what to try next.

Meta is also expanding AI translations for Reels into more languages, including Arabic, Bahasa Indonesian, French, Thai, and Vietnamese. The tool can preserve voice tone and offer lip-sync, making it easier for creators to reach global audiences.

What to Watch

• AI is moving into content strategy, not just content generation.

• Analytics are becoming easier to understand through plain-language questions.

• Localization is becoming more accessible for creators and small teams.

• Creative judgment still matters, especially when AI starts shaping what platforms suggest creators should make.

Insight

Read the full details here.

ChatGPT Memory Is Getting Smarter

GPT Image Generated

OpenAI introduced Dreaming, a new update to ChatGPT’s memory system designed to make the assistant better at remembering useful context across conversations.

The main idea: less repeating yourself, more continuity.

For creatives, this could be useful when working on:

• Newsletter tone and recurring formats

• Brand voice and visual direction

• Ongoing campaigns or client projects

• Repeated creative constraints and preferences

Instead of treating every chat like a fresh start, ChatGPT may become better at carrying relevant context forward.

What to watch for

• More personalized responses based on past conversations and preferences

• Less manual context-setting when working on repeat projects

• Better support for long-term workflows, like content calendars, campaigns, and client work

• More control over memory, including the ability to review, correct, or delete what ChatGPT remembers

• Privacy considerations, especially when working with client materials or sensitive project details

Why this matters for creatives

Creative work depends on context: taste, tone, audience, references, feedback, and past decisions.

A stronger memory system could make AI tools more useful for repeatable workflows, especially for solo creators and small teams who use ChatGPT across content, strategy, scripts, prompts, and client communication.

But the caution is real too. Better memory means creatives need better memory hygiene: reviewing what the tool remembers, removing outdated context, and using Temporary Chat when needed.

The bigger shift here is that prompting is becoming less about one-off instructions and more about managing context over time.

For creative communities, that means the next practical skill is not just knowing what to ask AI.

It is knowing what AI should remember.

Read more here

Pixel Studio Is Moving Into Gemini

GPT Image Generated

Google is ending the image generation feature inside Pixel Studio and redirecting users to Gemini instead.

Pixel Studio first launched with the Pixel 9 as a fun, built-in tool for making quick images, stickers, and visual ideas directly from your phone. Now, Google seems to be folding that experience into Gemini, where its newer image models like Nano Banana are becoming the main focus.

What to Watch For

• AI tools are getting consolidated

  Instead of keeping separate apps for every AI feature, companies are moving more tools into bigger platforms like Gemini.

• Standalone AI features may not last long

  A tool can be interesting, but if people do not use it often, it may eventually get absorbed into something larger.

• Image generation is becoming more workflow-based

  Gemini is not just about generating one image. It is moving toward editing, refining, and continuing visual ideas through conversation.

Insight

Read the full article here.

Martin Scorsese, AI, and the New Storyboard Room

Martin Scorsese Credit: TheStewartofNY/FilmMagic

Martin Scorsese is one of the most influential filmmakers in modern cinema, known for films like Goodfellas, Taxi Driver, The Irishman, and Killers of the Flower Moon. His work has shaped how generations of filmmakers think about visual storytelling, editing, character, and cinematic language.

So when Scorsese starts experimenting with AI in filmmaking, people pay attention.

According to the report, Scorsese is working with Black Forest Labs and using its FLUX image model as part of the storyboarding and pre-production process. The important detail here is that AI is not being framed as the final film. It is being used as a way to help visualize ideas earlier in the creative process.

For creatives, this is where the conversation gets practical.

AI can help directors, art directors, and production teams quickly explore:

• Scene references
• Mood boards
• Storyboard directions
• Visual treatments
• Early pitch materials

But it also raises fair concerns. If AI storyboards become a replacement for storyboard artists, concept artists, or production designers, then the tool starts shifting from support to displacement.

The healthier way to look at this is simple:

AI can be useful for testing and communicating ideas faster, but it still needs human taste, craft, and judgment to shape the final creative direction.

For our creative community, the key takeaway is not that AI is “taking over film.” It is that AI is becoming another layer in the creative workflow. The question is how we use it responsibly, without flattening the value of the artists who make visual storytelling worth watching.

Read more here.

💡 Insight

The bigger theme this week is workflow maturity.

AI image, video, voice, memory, analytics, and API tools are all moving in the same direction: closer to repeatable creative systems. That can be useful for small teams, solo creators, agencies, and production teams who need to move faster without losing context.

But the creative responsibility also gets bigger.

When AI remembers your preferences, supports your content strategy, edits your images, transforms video footage, or plugs into a production pipeline, the question becomes less about whether the tool is impressive and more about whether the process is clear.

Can you review what happened?
Can you control the output?
Can you protect client work, likeness, and brand safety?
Can you still make choices based on taste, not just tool suggestions?

The tools are becoming more capable, but the creative point of view still needs to come from people.

That is where the real work is.

That’s it for this week, folks.

Upscale Conf Recap: AI Is Moving Into the Creative Workflow

I spent some time at Upscale Conf in San Francisco, and the biggest takeaway was pretty clear:

AI creative tools are moving beyond one-off outputs.

The conversation is becoming less about “look what this model can generate” and more about how AI can fit into an actual creative workflow — the kind with references, revisions, team notes, consistency checks, client feedback, and final polish.

Basically: less slot machine, more production system.

The Big Theme

The next phase of creative AI feels like it’s going to be about workflow design.

Not just prompting.

Not just generating.

But knowing how to build a repeatable process that helps you create faster without losing your taste, voice, or creative direction.

That was also the thread behind Magnific’s three big announcements at the event.

Credits to Magnific

3 Magnific Releases to Know

1. Magnific Agents

Magnific Agents are designed to carry the context of a project and team decisions through the creative process.

What stood out to me here is that the agent is not positioned as something replacing the creative. It is more like a production companion that understands the direction, follows the workflow, and helps keep things moving.

For creative teams, this could become useful when you need to manage multiple rounds of assets, versions, references, client notes, and “please don’t use that version again” moments.

You know, the glamorous side of creative work.

2. Magnific MCP

MCP lets Magnific connect with tools like ChatGPT, Claude, Cursor, and other MCP-compatible assistants.

In plain English: your AI assistant could become a way to access Magnific’s creative tools without constantly jumping between platforms.

Why creatives should care:
This points to a future where your assistant can help generate, upscale, organize, or reference creative assets inside a more connected workflow. Less app-hopping. More continuity.

3. Magnific Flows

Flows might be the most practical announcement for teams.

The idea is simple: take a repeatable creative workflow and turn it into something reusable.

That could be especially useful for brand teams, agencies, production teams, social teams, and anyone who constantly needs to create variations without losing consistency.

Think:

• Product campaign assets
• Character consistency workflows
• Social content templates
• Editorial image systems
• AI video storyboard workflows
• Client-approved production pipelines

The value here is not just speed. It is making the process more repeatable, reviewable, and reliable.

That is where AI becomes more useful for professional creative teams.

Not just “Can we make something cool?”
But “Can we make something consistent enough to use again?”

Sessions Worth Watching

Since the sessions were livestreamed on YouTube, these are the ones I’d start with:

Martin LeBlanc’s Product Demo

This is the main one to watch for Magnific’s product direction, especially around Agents, MCP, and Flows.

Make it Magnific

This is a good place to understand Magnific’s bigger product direction and how they are thinking about creative AI as a workflow system, not just a generation tool.

Ctrl + Create: Turning AI Into Your Creative Advantage — Nik Kleverov

Nik Kleverov’s session explores one of the biggest creative questions surrounding AI today.

When everyone can generate content, the real edge becomes storytelling, taste, judgment, and knowing what is actually worth making.

The session looks at AI through the lens of filmmaking, advertising, and original IP — and why technology may lower the cost of production while increasing the value of creative direction.

Real AI Film Workflows

This panel is worth watching for anyone interested in filmmaking, production, and how AI is entering real creative pipelines.

The conversation around AI film is getting more useful when it moves beyond “Can AI make a film?” and into:

• What does the workflow actually look like?
• Who owns the creative direction?
• How do teams collaborate?
• What new creator models become possible?
• What still needs human taste, judgment, and craft?

Finding and Holding Onto Your Voice in Latent Space

This topic feels especially relevant for artists and creators.

As tools get more powerful, the creative question becomes less about whether the tool can make something polished and more about whether the work still feels like you.

That is going to be one of the biggest creative tensions in the next phase of AI.

Key Takeaways for Creatives

Here’s what I’d pay attention to:

• Workflow design is becoming a creative skill.
The people who know how to build repeatable systems will have an edge.

• Consistency is the new bottleneck.
One good image is easy. A full campaign, character, world, or brand system is harder.

• AI assistants may become the front door to your creative stack.
With things like MCP, tools may start connecting more directly into ChatGPT, Claude, and other assistants.

• Taste still matters. A lot.
The tools are getting faster, but speed without direction can still make generic work.

• Creative teams need rules.
As tools become more automated, teams need clearer standards around brand safety, review, usage rights, and what should stay human-approved.

Insight 💡 

Hi Creatives! 👋

This week’s highlights show AI tools becoming more embedded in everyday creative work.

From Canva adding Veo 3 for quick AI video clips, to YouTube making AI labels more visible, to AI cinema entering the festival space, the conversation is moving beyond what AI can generate.

The bigger question now is how creatives use these tools with taste, context, and responsibility.

This Week’s Highlights:

• Canva Brings Veo 3 Into AI Video

• YouTube Is Making AI Labels More Visible

• From Still Image to Moving Shot

• AI Cinema Enters the Festival Scene

• Claude Opus 4.8: Less Flashy, More Useful

• Runway’s API Is Turning Into a Creative Workflow Layer

A few days left before Upscale Conf lands in SF

A tiny reminder for my AI creative friends:

Upscale Conf is happening in just a few days, and if you have been curious about where creative AI is heading next, this is one of those rooms worth paying attention to.

On June 3 to 4, Upscale Conf is bringing together creatives, filmmakers, designers, technologists, founders, and AI tool builders in San Francisco for two days of talks, workshops, panels, and real conversations around AI, design, and creativity.

Not just “look what this tool can make” energy.

More like:

• How are creatives actually using AI in their workflow?

• What does production look like when tools become more visual, fast, and collaborative?

• How do we keep taste, direction, and human judgment in the process?

• What should brands, creators, and studios be watching next?

A little treat for our community

See you there?

Canva Brings Veo 3 Into AI Video

Canva

Canva is adding Google’s Veo 3 into Canva AI through a new feature called Create a Video Clip, letting users generate short video clips with sound from a text prompt.

The feature creates 8-second clips with audio, including ambient sound, dialogue, and sound effects. From there, users can bring the clip into Canva’s Video Editor to add text, music, brand assets, and resize it for social posts, presentations, or campaign materials.

What to watch for

• AI video is moving into everyday design tools
  This makes video generation more accessible for creators, marketers, and small teams who already use Canva.

• Sound is now part of the creative prompt
  With Veo 3, creators need to think beyond visuals and include mood, pacing, dialogue, and audio direction.

• It is still best for concepting
  With an initial limit of 5 video generations per month, this feels more useful for testing ideas, pitch visuals, and quick mockups than full-scale production.

• Editing still matters
  The generated clip is only the starting point. The final quality still depends on taste, brand direction, and how the creator refines it.

Insight

Read the full details here.

YouTube Is Making AI Labels More Visible

Youtube Blog

YouTube is updating how AI-generated and AI-altered content is labeled, making disclosures easier for viewers to see.

For long-form videos, AI labels may now appear below the video player, above the description. For Shorts, labels may appear directly on the video as an overlay.

This mainly applies to realistic AI content, especially when AI makes someone appear to say or do something they did not, changes footage of a real event or place, or creates realistic scenes that did not actually happen.

What To Watch For

• Creators are still expected to disclose meaningful AI use during upload.

• YouTube may also add labels automatically if its systems detect realistic AI-generated or altered content.

• Not all AI use needs a label, such as brainstorming, captions, minor edits, thumbnails, or general production support.

• Labels alone do not affect monetization or recommendations, according to YouTube.

Why This Matters For Creatives

This update is less about punishing AI use and more about viewer trust.

For creatives, the key question is simple: does the AI use change what the audience believes is real?

If AI is helping behind the scenes, the disclosure need may be limited. But if AI creates or alters realistic people, places, or events, it is better to be transparent.

As AI becomes part of everyday production, disclosure is becoming part of the creative workflow too.

Read more here

From Still Image to Moving Shot

LTX

Image to video tools are no longer just for turning a still image into a moving clip. They are becoming part of the creative workflow.

A product shot can become a short ad.
A character concept can become a motion test.
A campaign image can become a few social variations.

The real question now is not just which tool looks the best, but which tool gives you the right amount of control for the work you are creating.

What to Watch For

• Better motion control
  The strongest tools are giving creators more say over camera movement, subject motion, and scene consistency.

• Reference images matter more
  Image prompts help keep product details, characters, lighting, and style more consistent.

• AI video is becoming more editable
  The workflow is shifting from “generate once” to generate, adjust, extend, and refine.

• Different tools serve different needs
  Some are better for quick social clips. Others are better for cinematic tests, product visuals, or campaign development.

Why Creatives Should Care

For creators and small teams, image to video can stretch one strong visual into more formats without needing a full production setup every time.

But motion still needs intention. Just because an image can move does not mean it should. The best results still come from clear direction, pacing, and a reason for the visual to become video.

Our Take

Read full details here.

AI Cinema Enters the Festival Scene

A fully AI-generated feature film is heading to Tribeca, and it is already opening up a bigger conversation around AI, filmmaking, and creative responsibility.

Dreams of Violets is a 75-minute AI-generated docudrama by Iranian-born brothers Ash and Pooya Koosha. The film reportedly cost around $2,000 to make and is being positioned as the first feature-length, live-action film completely generated by AI to be accepted by a major film festival.

The film is based on political violence and civilian resistance in Iran, which makes this more than a technical milestone. It also raises questions around how AI should be used when telling stories connected to real people, real history, and real trauma.

What to Watch For

• AI filmmaking is moving beyond short demos
  A full-length AI-generated film at Tribeca signals that these tools are starting to enter serious creative and cultural spaces.

• Production costs are shifting
  A reported $2,000 budget shows how AI could make ambitious visual storytelling more accessible for independent creators.

• Ethics will matter more
  When AI is used to recreate real-world events, transparency, consent, and context become part of the creative process.

• Human judgment is still the differentiator
  The tools can generate the visuals, but the story, taste, research, and responsibility still sit with the creator.

Right now, this is not just about whether AI can make a movie.

The more useful question is: what kind of stories should be told with AI, and how should they be handled?

AI can lower the barrier for filmmakers who do not have access to traditional budgets, locations, or production teams. But lower cost does not mean lower responsibility.

As AI-generated films become more common, the strongest work will likely come from creators who are clear about their process, thoughtful about their subject matter, and intentional about when AI actually serves the story.

Insight

Read the full article here.

Claude Opus 4.8: Less Flashy, More Useful

Claude

Anthropic released Claude Opus 4.8, and while it is not positioned as a dramatic jump, it points to where AI workflows are heading: more control, better reasoning, and stronger support for complex tasks.

For creatives, the interesting part is not just that the model is “better.” It is that tools like Claude are becoming more useful for the work around the work: research, planning, creative direction, workflow building, campaign thinking, and reviewing ideas before they become final outputs.

What’s New

• Same regular pricing as Opus 4.7

• Better reasoning and coding performance

• Effort control, so users can choose between faster responses or deeper thinking

• Improved honesty around uncertainty, which helps reduce overconfident answers

• Dynamic workflows in Claude Code, where larger tasks can be broken into smaller parts and handled by subagents

What To Watch For

• More control over speed versus quality

• AI moving from generator to workflow partner

• Human review still matters

Insight

Read more here.

Runway’s API Is Turning Into a Creative Workflow Layer

Runway’s latest API changelog points to a bigger shift in AI production: creative tools are becoming more connected, more modular, and easier to build into repeatable workflows.

One update worth noting is HappyHorse 1.0, a newer AI video generation model designed for text-to-video and image-to-video workflows. It supports short-form video clips, with options for 720p or 1080p output and durations from 3 to 15 seconds.

For creatives, this makes it useful for quick concept videos, product motion tests, social-first content, and early-stage visual storytelling.

What To Watch For

• More model options in one place
  Runway’s API is starting to feel less like a single tool and more like a hub for video, image, audio, upscaling, and third-party models.

• More reference-based control
  Features like reference images, videos, and keyframes make AI video more useful for branded content, product storytelling, and visual consistency.

• Workflow matters more than one-off outputs
  The real value is not just generating more assets. It is building systems that are repeatable, reviewable, and reliable.

• Cost and moderation still need attention
  As AI video becomes easier to scale, teams will need clearer rules around prompts, references, budget, brand safety, and human review.

Insight

Check the full article here.

💡 Insight

AI tools are becoming faster, more accessible, and more connected to real production workflows.

That is useful for testing ideas, creating variations, and building content with smaller teams. But the creative edge still comes from human direction: knowing what to make, why it matters, and when the output is actually worth sharing.

The tools are improving. The point of view still has to come from the creator.

That’s it for this week, folks.

Hi Creatives! 👋

This week’s updates point to one clear shift: AI tools are becoming less about one-off outputs and more about full creative workflows.

From Upscale Conf returning to San Francisco, to Google’s agentic AI updates, Figma’s Design Agent, Midjourney V8.1, and Claude coming into ComfyUI workflows, the focus is moving toward tools that help creators plan, generate, edit, and refine in a more connected way.

For creatives, the real question is not just what these tools can make, but how well they fit into actual production.

This Week’s Highlights:

• Upscale Conf Returns to San Francisco

• AI Moves From Assistant to Creative Operator

• Figma’s AI Workspace Is Becoming More Agentic

• Midjourney V8.1 Gets Sharper

• Claude Comes to ComfyUI Workflows

Upscale Conf Returns to San Francisco

Upscale Conf

After the success of last year’s San Francisco edition, Upscale Conf is coming back to SF on June 3 to 4, 2026, this time under Magnific.

Last year, Upscale brought together more than 600 attendees and featured voices across AI, design, storytelling, and creative technology, including Chad Nelson from OpenAI, Julie Zhuo from Sundial and Meta, John Maeda from Microsoft, and Jason Zada from Secret Level.

This year, the conversation continues with a new lineup of creatives, technologists, filmmakers, journalists, designers, and AI builders gathering at The Midway in San Francisco for two days of talks, workshops, panels, networking, and hands-on sessions.

Upscale Conf is not just about showing what AI tools can generate. It is more about where creative work is heading next.

This year’s speaker lineup brings together people working across different parts of the creative AI ecosystem:

• Casey Newton and Kevin Roose bring the media and technology perspective

• Paul Trillo, Noah Wagner, Eliza McNitt, Kavan Cardoza, and Christina Lee Storm bring the filmmaking and AI storytelling lens

• Niceaunties, Curt Cameruci, Paige Piskin, and Julien Durand bring creator-led experimentation

• Joaquín Cuenca, Martin LeBlanc, Elena Berjillos, and Javi Lopez bring the Magnific product and creative workflow perspective

• Nik Kleverov, Billy Bohan Chinique, Jassica Bouvier, Thomas Squire, and Joanna Popper bring brand, agency, studio, and immersive media experience

Compared to last year’s lineup, which leaned heavily into big-picture conversations around AI, design, visual storytelling, and the future of creative tools, this year feels more focused on production workflows, narrative systems, brand applications, and real-time creative generation.

What to Watch For

A few sessions feel especially relevant for creatives and creative teams:

• AI for e-commerce and brand visuals
  A workshop focused on building product photography, paid ads, UGC-style assets, and brand visuals with AI.

• Character-to-narrative workflows
  A Magnific Spaces session showing how to go from one source image to character sheets, storyboards, video clips, and voiceover.

• Real-time AI generation
  A session exploring what happens when AI generation becomes fast enough for interactive creative tools.

• AI storytelling and production
  This year’s panel features Joanna Popper, Kavan Cardoza, Javi Lopez, and Christina Lee Storm, pointing toward a stronger conversation around AI-native content, studios, and cinematic workflows.

Want a free 2-day pass to Upscale Conf?

Here is a sneak peek of the 2-day agenda: Click here

Brands Are Becoming World Builders

Magnefic

I’ve been thinking about how AI is shifting brand storytelling from one-off campaigns into something closer to world-building.

Not just faster content creation, but richer visual systems, recurring characters, product-led stories, and creative workflows that feel more like entertainment than traditional advertising.

For creatives, the real question is not just what can we generate?
It’s how we build workflows that are repeatable, reviewable, and reliable while still leaving room for taste, direction, and storytelling.

I shared more thoughts on this. Check it here

AI Moves From Assistant to Creative Operator

Gemini era

Google’s I/O 2026 keynote makes one thing very clear: Google is no longer positioning Gemini as just a chatbot. The bigger story is the shift toward agentic AI, where models can reason, plan, edit, build, search, and take action across different products.

For creatives, this matters because the workflow is moving from “type a prompt, get an output” to something closer to directing a creative system. You are not just asking for an image, video, deck, or app. You are shaping a process where AI can help brainstorm, organize, revise, generate variations, and connect ideas across formats.

Gemini Omni Flash

One of the biggest updates is Gemini Omni Flash, Google’s new multimodal model starting with video.

The main idea: creators can use text, images, audio, and video as input, then generate or edit video through conversation.

Why it matters:

• It supports more natural video editing through prompts

• It can help maintain characters, scenes, and visual consistency

• It works with reference materials, which is closer to how creative projects actually begin

• It is rolling into Gemini, Google Flow, and YouTube Shorts

• API access is expected for developers and enterprise users

For creatives, this means AI video is becoming less about one-off clips and more about directing and revising a visual idea.

A Few Things to Keep an Eye On

♦ AI is becoming a workflow partner

Google Flow, Gemini Omni, and Flow Agent point to AI moving beyond one-off generation into planning, editing, organizing, and refining creative work.

For creatives, this means better support for character consistency, conversational video editing, batch edits, story development, and custom tools.

♦ AI video is getting more production-ready

Gemini Omni works across text, image, audio, and video, making it more useful for campaigns, social content, pitch decks, previsualization, and product storytelling.

The watch-out: stronger AI video also means stronger responsibility around authenticity, consent, disclosure, and avatar use.

♦ Brand systems are becoming AI-native

Google’s updates around Pics, Stitch, and Pomelli show AI moving into broader brand workflows, from image creation to websites and prototypes.

For small teams, this reduces production friction, but makes taste, originality, and brand point of view even more important.

♦ Provenance is part of delivery

With SynthID and Content Credentials expanding, verification is becoming part of the creative process.

Read full details here.

Figma’s AI Workspace Is Becoming More Agentic

Figma

Figma’s latest update is not the start of its AI workspace story. It is the next layer.

Figma has already been building AI into its ecosystem through FigJam, Figma AI, Slides, Make, Draw, Buzz, and Sites. The newer shift is that AI is becoming more active inside the design file itself.

With the new Figma Design Agent, AI is moving from one-off assistance to more agent-driven workflows. Instead of only helping with tasks like renaming layers or generating a first draft, the agent can work directly on the canvas, understand design system context, apply feedback, remix screens, and support repetitive production work.

For creative teams, the bigger story is not “Figma has AI now.” It is that Figma is making AI part of the workspace where design decisions already happen.

What To Watch For

• Figma Design Agent is now in beta
  It can help generate directions, make bulk edits, remix screens, and apply feedback inside Figma Design.

• Design systems become more important
  The agent can use components, variables, tokens, and libraries as context. A messy system may lead to messy AI-assisted output.

• Figma Make is getting more structured
  Custom Skills, voice-to-text, question cards, version history, and Zapier connectors make it easier to create repeatable workflows.

• Access and AI credits will matter
  The agent is free during beta, but standard AI credits will apply once it becomes generally available.

Watch their demo here 👇️ 

Insight

Read the full article here.

Midjourney V8.1 Gets Sharper

CTTO: Midjourney

Midjourney’s V8.1 update is a practical quality improvement for creators working with visual consistency. The update brings V8.1 to both Discord and the web, with sharper image quality, better HD output, and stronger results when using SREFs and Moodboards.

For creatives, this matters because Midjourney is becoming less about one-off image generation and more useful for building repeatable visual systems, from campaign concepts to product worlds and editorial moodboards.

What to Watch For

• Sharper HD results
  V8.1 improves image clarity, especially for high-resolution generations.

• Better SREF and Moodboard performance
  Useful for creators trying to keep a consistent style across multiple images.

• Discord access is back in the workflow
  V8.1 is now available on both Discord and midjourney.com.

• Community ranking is shaping V8.2
  Midjourney is using user feedback to improve the next version.

Insight

Read more here.

Claude Comes to ComfyUI Workflows

ComfyUI shared that Anthropic’s Claude is now available as a partner node inside ComfyUI, giving creators a way to add reasoning directly into image and video pipelines.

For creatives building more complex workflows, this means Claude can help with:

• Turning a rough idea into a more detailed prompt before generation

• Reading an image and giving descriptions, critique, or edit suggestions

• Creating captions, alt text, scene breakdowns, or workflow logic that other nodes can use

Key tools used: Claude by Anthropic, ComfyUI, image/video generation checkpoints

Check full post here.

💡 Insight

We are seeing more support for design systems, character workflows, real-time creative generation, image quality, and more flexible production pipelines. That is helpful for creators and small teams who need faster ways to test ideas and build content.

But faster does not always mean better. Taste, direction, consistency, and review still matter.

The creative edge is not just knowing the newest tool. It is knowing how to use these tools in a workflow that is repeatable, reviewable, and reliable, while still keeping the human point of view at the center.

That’s it for this week, folks.

Hi Creatives! 👋

This week’s highlights point to a more settled phase in AI tools for creative work. Across voice, video, image workflows, and production systems, the shift is less about novelty and more about how these tools perform in actual use.

Some of these updates make creative processes easier to manage. Others are a reminder that speed, access, and convenience do not always mean long-term reliability. For creatives, the question is becoming less about what a tool can do once and more about whether it fits into a workflow that needs consistency over time.

This Week’s Highlights:

• Freepik’s new 3D Scenes brings camera control into AI image workflows

• Gemini 3.1 Flash Live makes voice AI more practical

• Sora’s shutdown is a reminder that AI video platforms can shift fast

• ElevenLabs Flows brings node-based creative pipelines into one canvas

• Midjourney’s Version system is becoming a workflow choice, not just a settings toggle

• A Sheep’s Big Ambition

• From Prompting to Production

Freepik’s new 3D Scenes brings camera control into AI image workflows

Twitter tweet

Freepik’s new 3D Scenes feels less like another image generator and more like a virtual set builder. You can start from a preset scene, your own image, a Street View panorama, a 3D model, or even a blank canvas, then place objects, move the camera, and capture multiple shots from different angles without rebuilding the environment each time. Freepik’s official docs focus on that workflow, while Freepik’s own social post describes the feature as “Powered by World Labs spatial AI.”

What it actually does

• Turns an uploaded image into an explorable 3D environment.

• Lets you add subjects from a built-in 3D library, upload your own GLB model, or generate a 3D object from an image on the fly.

• Gives you Compose mode for building the scene and Shoot mode for framing shots with field of view, zoom, aspect ratio, and cinematic settings.

• Keeps the environment persistent between captures, so lighting, objects, and scene structure stay consistent across angles.

• Offers a free Standard preview mode and a credit-based High Quality render mode for final images.

Insight

Read full details here.

Gemini 3.1 Flash Live makes voice AI more practical

🖼️ Gemini

Gemini 3.1 Flash Live is built for real-time voice interaction, with the focus on faster responses, more natural rhythm, and stronger performance in longer or messier conversations. It is available in preview through the Live API and is also being rolled into voice-facing products and enterprise use cases.

What to watch for

• It is designed for live dialogue, not just speech output
  The model is optimized for low-latency, audio-to-audio interaction and voice-first applications.

• The bigger pitch is reliability in real use
  It is positioned for multi-step task execution, stronger tonal understanding, and better handling of noisy environments and interruptions.

• It supports more than simple chat
  The preview model supports function calling, Live API access, and Google Search grounding, which makes it more useful for assistants and interactive tools.

• This is meant to show up across the stack
  The model is being pushed into developer tooling, enterprise customer experience, and consumer-facing voice products.

Insight

Read more here.

Sora’s shutdown is a reminder that AI video platforms can shift fast

🖼️ CTTO

OpenAI’s reported decision to discontinue the Sora video platform points to a bigger issue in AI video right now: platform stability. What makes this notable is not only the product decision itself, but how quickly the direction appears to have changed. Just days earlier, OpenAI’s Sora release notes were still rolling out new editing features, while its developer docs now show a formal deprecation path for the Videos API and Sora 2 model aliases, with removal scheduled for September 24, 2026.

For anyone following AI video, this looks less like a small product cleanup and more like a reminder that even high-profile tools can shift course fast. Sora helped set the tone for the current wave of cinematic text-to-video, but the latest reporting suggests OpenAI is redirecting focus toward areas it sees as more commercially important, including coding and enterprise products.

What to watch for

• This looks bigger than just an app change
  The move appears tied to a broader pullback from Sora as a standalone video product, not only its front-end experience.

• The direction changed quickly
  Sora had still been receiving product updates, which makes the reported shutdown feel abrupt.

• Workflow flexibility matters more than ever
  If access, APIs, or core tools can change this fast, teams need systems that can adapt without breaking production.

• Exports and asset management matter
  When platforms change direction, having organized files, references, and backups becomes just as important as the tool itself.

Insight

Read the full article here.

ElevenLabs Flows brings node-based creative pipelines into one canvas

📷️ Credits to ElevenLabs

ElevenLabs has introduced Flows inside ElevenCreative, a node-based creative canvas designed to connect image generation, video, text to speech, lip-sync, sound effects, and music in one workspace. The pitch is straightforward: instead of bouncing between separate tools, exporting files, and rebuilding the same sequence over and over, teams can map a pipeline once and reuse it with different inputs. ElevenLabs says Flows is available now in ElevenCreative, with API access coming soon.

For creatives, the more interesting part is not just that another platform added more models. It is that ElevenLabs is pushing a workflow-first idea. A Flow can be reused, rerun, and adjusted at the step level, which makes it easier to test variations without regenerating an entire project from scratch

What to watch for

• Flows is built for end-to-end pipelines
  You can chain visuals, voice, lip-sync, and music in one canvas, then rework a single step without regenerating everything.

• Reuse and batch testing are a big part of the value
  ElevenLabs says one Flow can be reused with different products, hooks, avatars, voices, or soundtracks, which makes it relevant for ad testing and repeatable content production.

• It sits inside a larger multimodal setup
  ElevenLabs recently expanded its Image & Video offering with models such as Veo, Sora, Kling, Wan, Seedance, Nanobanana, Flux Kontext, GPT Image, and Seedream, plus its own audio tools.

• API access is not here yet
  Flows is available now in ElevenCreative, while API access is listed as coming soon.

Insight

For the creative community, the value here is not simply “more AI in one place.” Plenty of platforms are moving toward all-in-one positioning. What stands out is the shift toward repeatable creative infrastructure.

That matters because a lot of production time is lost in the handoff layer:

• exporting and re-uploading assets

• rebuilding the same prompt chain

• changing one step and having to redo the rest

• managing version drift across tools

Flows is trying to reduce that friction by turning a creative process into something more modular and reusable. The upside is speed, testing, and more consistency across outputs. The trade-off is that a more automated pipeline can also make it easier to produce a large volume of average work. For creatives, the difference will still come from direction, taste, and knowing where a workflow should stay flexible instead of becoming too templated.

Read it full here.

Midjourney’s Version system is becoming a workflow choice, not just a settings toggle

🖼️ Credits to Midjourney

Midjourney’s Version control lets users switch between model generations with --version or --v, and the current default is V7. The official docs now position versioning less as a technical detail and more as a practical creative choice, since each model handles prompts, style, speed, and feature compatibility differently.

The bigger shift is that Midjourney now has a clearer split between stable production use and early testing. V7 remains the default model, while V8 Alpha launched on March 17, 2026 on the alpha site only, not on the main site or in Discord. Midjourney says V8 Alpha is its fastest model so far, with standard jobs rendering about 4 to 5 times faster than earlier versions, and adds optional 2K HD generation through --hd.

What to watch for

• V7 is still the safer default
  It remains the most practical choice for stable workflows and broader feature support.

• V8 Alpha is faster, but still early
  It is meant more for testing than for fully established production use right now.

• Feature support varies by version
  Some tools and prompt controls do not work the same way across models, so version choice can affect your workflow as much as the output.

• Speed and consistency tools matter
  Features like Draft Mode and Omni Reference make Midjourney more useful for ideation and repeatable visual systems.

Read the full details here.

A Sheep’s Big Ambition

Edmond Yang and the Nerve Lab team just released a new promotional film for Rednote Redgalas, inspired by the story of a sheep determined to become a sheepdog. It is a strong example of how AI can support a larger production pipeline when paired with clear creative direction, collaboration, and craft across story, art direction, sound, and post. The project was built entirely in Freepik, with Spaces used for collaboration, iteration, and finalization.

Read more here

From Prompting to Production

Henry Duabrez shared JUNKYARD KING — Episode 0 is an early look at how AI video tools are moving from short-form generation into more structured storytelling. Built over roughly four weeks, the project explores how a small creative setup can develop an animation prototype with more of a studio mindset, using prepared assets, consistent visual references, and a more directed workflow. What stands out is not just the final piece, but the process behind it: using tools like Seedance 2.0, Google Flow, and Gemini to build a more intentional pipeline for series development.

Check full details here.

💡 Insight

Taken together, this week’s updates suggest that creative work is becoming more shaped by workflow decisions, not just model quality. The more useful question now is not only what a tool can generate, but whether it supports the way a team actually works, adapts when platforms change, and helps maintain quality from draft to delivery.

For the creative community, that makes judgment, flexibility, and process design just as important as the tools themselves. The technology keeps moving, but the value still comes from knowing what to use, when to use it, and where to stay cautious.

That’s it for this week, folks.

Hi Creatives! 👋

This week’s highlights reflect a more practical phase for AI in creative work. Across image editing, filmmaking, animation, and commercial production, the focus is shifting away from one-off outputs and toward how these tools actually support real workflows.

What stands out is how many of these updates are about connection: keeping ideation, editing, revisions, and production closer together in the same process. For creatives, that can mean faster iteration and less friction across a project. At the same time, it puts more pressure on maintaining quality, consistency, and clear creative judgment as tools take on a bigger role in the workflow.

This Week’s Highlights:

• Luma’s agent is less like a chatbot and more like a creative operator

• Netflix’s InterPositive deal shows where AI filmmaking is actually heading

• Adobe is making AI editing more useful inside Photoshop and Firefly

• Runway’s Under Armour case study shows where AI fits in commercial work

• Turbulence shows the value of a more connected animation pipeline

• AI film is getting easier to make. Rights clearance is not.

Luma’s agent is less like a chatbot and more like a creative operator

Luma’s app is centered on creative agents rather than a single prompt box. On its product page, Luma describes these agents as systems that can plan, generate, iterate, and refine creative work while keeping the full project context intact across the workflow. The main idea is simple: instead of restarting every time you move from text to image to video to editing, the agent carries the project forward in one connected workspace.

What Luma means by “agent”

• Not just a chat assistant
  Luma’s agent is positioned as a system that does more than respond to prompts. It is built to help carry a project forward.

• A context-aware creative operator
  The agent keeps track of the brief, references, previous outputs, and project direction, so you do not have to restart each step from scratch.

• Built to work across formats
  It can move between text, image, video, and audio while keeping the same creative context connected.

• Designed for iteration
  Instead of producing one output at a time, the agent is meant to help teams explore, refine, and compare multiple directions inside the same workspace.

CTTO

It can :

• Create across formats
  The agent works across text, image, video, audio, and voice, making it easier to move between formats without losing context.

• Carry a project from idea to output
  It is built to support the full process, from early concepting to final asset development.

• Make iteration easier
  Editing, refinement, and revisions happen within the same workspace, which can reduce handoff friction.

• Help keep campaigns consistent
  Because the agent holds onto project context, it can support stronger continuity across assets, versions, and deliverables.

• Support commercial creative use cases
  Luma highlights:

  • multi-asset campaigns

  • product visuals from different angles

  • social video ads

  • presentation-ready slide decks

  • storyboards

  • localized videos

  • infographics

  • podcast clips turned into video

💡 Insight

Read the full details here.

Netflix’s InterPositive deal shows where AI filmmaking is actually heading

CTTO

Netflix’s acquisition of InterPositive, the filmmaking tech company founded by Ben Affleck, is another sign that AI in film is moving past hype and closer to real production use.

What makes this interesting is that InterPositive is not being framed as a “type a prompt, get a movie” company. The focus is more practical: tools that can support filmmakers with things like cleanup, reframing, background adjustments, color work, and VFX-related tasks. For creatives, that is a much more useful signal. It suggests the next wave of AI in entertainment may be less about replacing the process and more about supporting it.

What to watch for

1. AI is becoming part of the production pipeline
This deal points to a more grounded use of AI in film. Instead of chasing full automation, the value seems to be in helping teams move faster on technical tasks that already exist in post and production.

2. Control and consistency matter more than spectacle
Tools tied to a project’s actual footage are often more useful than broad prompt-based generation. For creatives, that could mean better continuity, cleaner revisions, and fewer outputs that need to be rebuilt from scratch.

3. The standard is getting higher
As more companies build AI into production tools, the expectation will shift. It is no longer enough for a tool to generate something interesting. It has to fit into real workflows, protect quality, and support creative intent.

Read the full details here.

Here is what we think about it.

Netflix Acquires InterPositive, Bringing AI to Entertainment Production

The inevitable next phrase of AI in entertainment isn’t “type a prompt, get a movie.”

It’s enterprise-grade tooling that quietly rewires the production pipeline.
Netflix just acquired InterPositive (founded by Ben Affleck) and is bringing over the full ~16-person team. Affleck is joining as a Senior Advisor; deal terms weren’t disclosed.

Read it full here.

Adobe is making AI editing more useful inside Photoshop and Firefly

CTTO

Adobe’s latest update is less about adding another AI feature and more about making editing faster inside the tools creatives already use. The main update is AI Assistant in Photoshop, along with a more capable Firefly Image Editor for editing, cleanup, and image refinement.

What to watch for

• AI Assistant in Photoshop
  Users can describe edits in plain language and let Photoshop apply them or guide them through the steps.

• More precise editing
  With AI Markup, users can point to a specific area of an image and direct the edit more clearly.

• Firefly is becoming more of an editing workspace
  It now brings together tools like:

  • Generative Fill

  • Generative Remove

  • Generative Expand

  • Generative Upscale

  • Remove Background

• Model choice is part of the strategy
  Adobe is also giving users access to multiple AI models inside Firefly, not just its own.

Why it matters for creatives

• Faster revisions

• Less switching between tools

• Easier cleanup and refinement

• More control during early-stage visual development

Read the full details here.

Runway’s Under Armour case study shows where AI fits in commercial work

CTTO

Runway’s Under Armour case study is a good example of AI being used in a practical way inside advertising. The project was not about replacing the shoot. It was about helping a small team move faster in pitching, post-production, and revisions.

What to watch for

• AI helped before and after production
  The team used Runway to visualize ideas during the pitch stage, then used it again in post to build out the final fire effects.

• This was still a real production
  The campaign included a live shoot, a full crew, and athletes on set. AI supported the workflow rather than replacing it.

• Speed was the real advantage
  One of the strongest takeaways was how quickly the team could handle client feedback and turn around revisions.

Read full details here.

Turbulence shows the value of a more connected animation pipeline

Turbulence is an animated short created by Tumblehead Animation Studio and Christopher Rutledge in collaboration with SideFX. What makes it worth noting is not just the film itself, but the fact that the team used Houdini across the production pipeline rather than treating it as a tool for only one stage.

What to watch for

• This was built as an end-to-end Houdini project
  SideFX says the film used Houdini for all aspects of production, and the surrounding talks point to tools like KineFX, APEX, Vellum, crowds, and Copernicus as part of the workflow.

• The real advantage was pipeline structure
  The team says they moved from Maya and Redshift because they wanted a better way to organize shots, assets, and pipeline flow, and found Houdini’s node-based approach and Python API a better fit.

• Animation inside Houdini is becoming more practical
  One of the animators said the workflow felt natural after the adjustment period, and highlighted features like selection sets and inbetweening tools as useful for day-to-day animation work.

• Style still mattered as much as software
  Outside coverage describes the short as having a fluid, caricatured, clay-like look, which is a good reminder that the technical setup was supporting a very specific visual direction.

Insight 💡 

The useful takeaway here is not that every team should switch tools. It is that creative workflows get stronger when animation, FX, lookdev, and comp stay more connected. That usually means fewer handoffs, less context loss, and better continuity from first pass to final frame.

Read full details here.

AI film is getting easier to make. Rights clearance is not.

As AI film tools get better at generating voices, songs, dialogue, and synthetic performances, the harder part is often no longer the output. It is whether the work is cleared, attributable, and safe to publish or distribute. The U.S. Copyright Office now separates these issues into different tracks, including digital replicas and copyrightability, which is a useful reminder that voice, likeness, and authorship are not the same thing.

What to watch for

• Voice rights and copyright are different
  A cloned or AI-simulated voice can raise digital replica and consent issues, even if the final file is newly generated.

• Music still needs permission
  Using copyrighted music, vocals, or artist-style imitation without authorization remains a major risk area for AI-assisted film and content.

• Human contribution still matters
  If the final work is heavily AI-generated, the human creative role in writing, editing, arranging, or directing becomes more important when thinking about ownership and protection.

• Disclosure is becoming part of publishing
  YouTube requires creators to disclose realistic, meaningfully altered or synthetic content, including examples like synthetically generated music or cloning someone else’s voice. Repeated failure to disclose can lead to labels, content removal, or suspension from the YouTube Partner Program.

• Provenance tools matter more now
  As synthetic media becomes easier to make, tools and standards that show how content was created are becoming more relevant for client work and distribution.

Read full details here.

AI Craft with Strong Visual Direction

We’ve been enjoying the work of Alexandre (Arlo) Garnier, whose latest piece caught our attention for its strong visual atmosphere and polished execution. It’s the kind of work that reminds you how much craft still matters: lighting, pacing, and mood all doing the heavy lifting. Worth a look if you’re into AI work that feels thoughtful, cinematic, and well-directed.

Check the work here.

A Reality Show Built for the AI Era

The first AI reality show is here! 🎬

This concept from OpenArt Studios reimagines the reality TV format using AI influencers as the contestants. In Bot House, six AI personalities enter a shared space filled with challenges and social drama, with one clear rule: stay relevant or get eliminated. It’s an interesting example of how generative tools are starting to shape not just visuals, but entire entertainment formats.

💡 Insight

Looking across this week’s updates, the pattern is fairly clear: the most useful AI tools are becoming the ones that fit into the creative process more naturally, not the ones that simply produce something fast.

That matters because speed on its own is not the goal. What still shapes strong work is direction, taste, and the ability to make thoughtful decisions from draft to final. As AI becomes more embedded in production, the creative edge still comes from the people guiding the work, refining it, and knowing where the tool helps and where it should stop.

That’s it for this week, folks.

Hi Creatives! 👋

This week’s highlights all point to a practical shift: generative tools are being judged less on “wow” outputs and more on whether they hold up inside real creative pipelines. You can see it across image gen (where typography and repeatability matter), video (where “directing” beats rerolling), and music (where fast drafts help you move from concept to cut sooner). The upside is speed and iteration. The trade-off is higher expectations around consistency, rights, and verification.

This Week’s Highlights:

• Nano Banana 2: fast image gen that’s actually built for real workflows

• Gemini’s AI music maker (Lyria 3 beta)

• Hollywood vs. ByteDance (Seedance 2.0)

• Paul Trillo x Asteria’s “no slot machine” pipeline

Nano Banana 2: fast image gen that’s actually built for real workflows

🖼️ Credit to Google Deepmind

Nano Banana 2 (also referred to as Gemini 3.1 Flash Image) is Google DeepMind’s updated image generation and editing model. The focus is less on “one pretty image” and more on outputs that fit real creative workflows: readable text, consistency across a series, and practical sizes for delivery.

What it can do

• Cleaner text in images
  Better at generating legible headlines, labels, and signage. It also supports translating text inside an image for localization.

• More consistent series outputs
  Designed to keep the same characters and objects stable across multiple images, useful for storyboards, campaign sets, and recurring characters.

• Web grounded generation
  Can pull context from web search to help create infographics and diagrams that reference real world information.

• Production friendly specs
  Supports common aspect ratios and higher resolution outputs, up to 4K.

What we noticed:

• Nano Banana: still had some hallucinated details, and the text didn’t land in the typography style we wanted.

• Nano Banana 2: text looks cleaner and more legible, and the overall style is closer to what we prompted.

Recraft v4 vs Seedream 5 Lite vs Nano Banana:

For the full comparison, check our previous post on Recraft vs Nano Banana vs Seedream for generating images with text here.

What to watch for

• Web grounded does not mean always correct
  Infographics and “data visuals” still need human verification.

• Brand and likeness risk rises with realism
  Avoid using real people, logos, or trademarked characters unless you have permission and cleared assets.

• Provenance is improving, but not universal
  Google highlights SynthID watermarking and C2PA style credentials. Helpful for transparency, but many platforms and viewers still do not check.

Quick examples you can try

1) Poster mockup with readable type

Prompt: Create a minimalist product poster for a skincare serum. Add a bold headline: “Hydration, minus the heaviness.” Add a subhead: “Lightweight gel serum for daily use.” Keep the text perfectly legible and aligned. Output 4:5.

2) Localization pass

Prompt: Translate the headline and CTA into Spanish while keeping the typography style consistent. Update any small details so it feels natural for Spanish speaking audiences.

3) 3 frame storyboard with the same character

Prompt: Create three frames featuring the same character with identical face, hair, and outfit. Frame 1: entering a studio. Frame 2: arranging props. Frame 3: presenting the final setup. Keep the character consistent across all frames. Output 16:9.

Read the full details here.

Gemini’s AI music maker (Lyria 3 beta)

Google is adding AI music generation directly inside the Gemini app, powered by Lyria 3 (beta). The promise is simple: you describe a mood, a genre, or a moment, and Gemini generates a 30 second track that you can use as a quick soundtrack, plus optional cover art for sharing.

For creators, this is less about replacing music production and more about speeding up the “I need something that fits this edit” step, especially for social videos, drafts, and concept pitches.

What you can do with Lyria 3 in Gemini

• Generate music from text prompts
  Describe genre, instruments, tempo, mood, and structure.

• Use images and video as inspiration
  Upload a photo or short clip, then ask for a track that matches the vibe.

• Instrumental or lyrical tracks
  Gemini can also generate lyrics, depending on the prompt and settings.

• Shareable output
  Google positions this as a lightweight creation flow, made to live inside chat.

What to watch for (so it doesn’t backfire)

• Rights + “style prompts”: Even if a tool claims it avoids copying, anything that references a specific artist or recognizable track can raise legal and brand risk. Keep it genre/mood-based for client work.

• Edit control: 30 seconds is great for drafts, but you may still need clean structure (intro/build/hit) for cutting to picture, which usually means multiple generations.

• Training transparency: Coverage notes ongoing debate about what music models train on and how clearly companies disclose that. For commercial use, assume scrutiny can happen.

💡 Insight

For creators, this is most useful as a draft soundtrack generator for pitches, animatics, and early edits — when you need a vibe fast and you’re not ready to commission music. For anything commercial, the practical move is to keep prompts generic (mood/genre/instrumentation), track your prompt versions, and have a backup plan with licensed libraries if a project needs cleaner rights posture.

Read the full details here.

Hollywood vs. ByteDance (Seedance 2.0)

🖼️ (Jonathan Raa / Getty Images)

A new Los Angeles Times report says the Motion Picture Association (MPA) has sent a cease-and-desist letter to ByteDance over Seedance 2.0, calling the situation “pervasive copyright infringement.” This is a notable escalation because it moves from individual studio pushback to an industry trade group stepping in with a unified position.

The flashpoint is familiar: viral AI clips that look cinematic and realistic, but appear to use actors’ likenesses and studio owned IP without permission.

Here’s the core sequence, in plain terms:

• Seedance 2.0 generates high fidelity AI video that has been used to create viral clips featuring recognizable actors and well known franchises.

• Major studios have already issued legal threats, and the LA Times reports the MPA has now added its own cease-and-desist.

• The MPA’s argument goes beyond “users misused the tool.” It argues the model was trained on studio works without consent and that the infringement is built into how the system performs.

• ByteDance has said it respects IP and is working on stronger safeguards.

This means if you’re a creator or a brand team, don’t use Seedance 2.0 for anything publishable or commercial when the output involves any of the following:

• Real people’s likenesses (celebs, influencers, employees, clients) without explicit permission.

• Recognizable IP: characters, franchises, scenes, distinctive costumes, or anything that reads like a specific studio property.

• “In the style of” prompts tied to a known film / show / creator if the goal is to mimic a recognizable look closely enough that viewers could confuse it for the original.

Read the full details here.

Paul Trillo x Asteria’s “no slot machine” pipeline

High-End Production with Aston Martin F1 Team

We’ve been following Paul Trillo, and this LinkedIn post is a good look at how Asteria is using generative video in a real production pipeline, not just text prompts.

For an Aston Martin F1 project (with CoreWeave), they combined live action, miniatures, 3D, VFX, and generative rendering—but the key is control: they rebuilt the workflow so they can edit specific frames on a timeline and refine shots repeatedly (instead of rerolling random outputs). Some shots used 10+ keyframes, art-directed in Cinema 4D + Photoshop, then placed back onto a 3D animatic—making it possible to accurately track logos and brand details, which is usually where AI workflows break.

Why it matters for creatives

• More predictable iterations: refine a shot with intention, not roulette.

• Brand-safe direction: they mention training fully licensed custom models on Marey (no scraped data).

• Signals the shift: generative video as an extension of 3D pipelines, not a replacement.

Trade-off

• This is still high-effort, high-craft production. The control comes with complexity.

Check their work here.

💡 Insight

If there’s one theme to keep in mind, it’s this: tools are getting better at generating assets, but the value still comes from your decisions. Use these updates to buy time back for what actually differentiates your work: clearer creative intent, stronger brand consistency, and cleaner risk posture.

A simple way to apply this week’s highlights:

• Treat “web-grounded” and AI-infographics as drafts: verify facts before you ship.

• Keep anything client-facing IP-safe: use cleared assets, avoid recognizable likenesses/logos unless you have permission.

• Design for repeatability: build small prompt templates and series rules (type style, layout grid, character consistency) so your outputs don’t drift from post to post.

That’s it for this week, folks.

Hi Creatives! 👋

This week’s highlights all point to the same shift: generative tools are moving from “cool demo” territory into actual production pressure. You’re seeing it in three places at once:

• Video that reads like coverage and editing choices, not a single awkward shot (the Spaghetti Benchmark with Seedance 2.0).

• Image generation that thinks like a designer, with layout, hierarchy, and negative space in mind (Recraft v4 on Freepik).

• Workflow tools that reduce blank-timeline pain, like rough cut assembly and last mile polish (Adobe Quick Cut, Magnific Video Upscaler).

The upside for creators: faster iterations, clearer pitches, and more room to explore. The catch: expectations rise just as quickly, especially around rights, provenance, and consistency.

This Week’s Highlights:

• The Spaghetti Benchmark: from “AI slop” to “wait… that’s a real cut?”

• Recraft v4 on Freepik: “design-ready” images, not just pretty ones

• Adobe Firefly Video Editor “Quick Cut”

• State of Generative Media (Vol. 1): 2025 was the year “production” caught up to “play”

• Amazon MGM is testing in-house AI tools for film and TV production

• AI Fan Edit: Rewriting the Stranger Things Finale Battle

The Spaghetti Benchmark: from “AI slop” to “wait… that’s a real cut?”

Remember when “Will Smith eating spaghetti” was the unofficial stress test for how broken AI video still was?

CTTO

Now we’ve got the next chapter: “Will Smith fighting a spaghetti monster… epic action film scene… different cuts… 80s movie scene.” The point isn’t the celebrity prompt (more on that in a sec) — it’s that the output reads like edited coverage, not one awkward continuous shot.

Made with ByteDance’s Seedance 2.0

Made with ByteDance’s Seedance 2.0 (a new multimodal video model).

Seedance 2.0’s headline flex: it’s designed for multi-shot, cinematic-ish generation and can take text + image + video + audio as inputs, with “reference” controls (so you can guide it more like a director than a lottery player).

Why creatives should care (the practical version)

If you do any kind of pitching, pre-vis, concepting, or brand work, the interesting shift is:

• Story rhythm is getting easier to prototype: “different cuts” is a big deal because timing and pacing are where ideas usually live or die.

• Previs is starting to look like a real creative decision (camera language, beats, mood), instead of a proof-of-tech.

• Audio-supported clips change how clients judge “rough drafts” — even simple sound makes something feel closer to a real spot or trailer.

The caution (please don’t build a client workflow on celebrity likeness)

A lot of viral tests use famous faces because it’s an easy yardstick. It’s also where legal + brand risk spikes.

If you’re testing the model:

• Use original characters, your own footage, or licensed references

• Treat it as concept / previs, then rebuild with cleared assets for anything commercial.

Read the full details here.

Recraft v4 on Freepik: “design-ready” images, not just pretty ones

AI Generated Image inside Freepik ( Recraft )

Recraft v4 is Freepik’s new image model aimed at a very specific pain point: images that look good and leave room for typography.

Instead of generating a scene that you have to fight in Photoshop just to place a headline, Recraft v4 is tuned to produce visuals with clear structure, balanced composition, and intentional negative space—the stuff designers care about when something needs to ship.

What’s different (in plain language)

• Layout-aware outputs: more grid-friendly compositions and cleaner hierarchy, so your text overlays don’t feel like an afterthought.

• More natural photorealism: better lighting/material feel, so results are closer to “usable” without heavy cleanup.

• Two resolution options: v4 (1K) for day-to-day work, and v4 Pro (2K) when you need sharper detail.

When it’s actually useful

This is most helpful for:

• Ads + social posts where you need space for copy

• Posters / editorial covers that need hierarchy

• Brand systems where you’re generating a set of visuals, not a one-off

Quick “try it without overthinking” workflow

1. Open Freepik’s AI Image Generator

2. Pick Recraft v4

3. Prompt like a designer: format + layout need + subject

   • Example: “Hero banner, clean negative space top third for headline, centered product shot, soft studio lighting…”

💡 Insight

Recraft v4 is part of a broader shift: image models are starting to compete on design utility, not just aesthetics. For creatives, that’s a real productivity gain, less time fixing compositions so they can hold typography.

The tradeoff: when tools deliver “layout-ready” faster, teams often expect more variations, faster approvals, and tighter brand consistency. The winners won’t just be people who can generate images—they’ll be the ones who can art direct with constraints (grid, hierarchy, brand rules, and usage rights) and keep the process from turning into endless options.

Magnific Video Upscaler

Credit to Freepik

Here’s what it brings to the table for creators polishing AI or real footage:

• Upscale up to 4K

• 3 presets + a custom mode (so you can dial in the look, not just “enhance everything”)

• Turbo mode when you need speed

• FPS Boost for smoother motion

• 1-frame preview to sanity-check results before spending credits

Why it matters: this is one of those “last-mile” tools that can make test renders, social edits, and AI clips feel more production-ready—without committing to a full post pipeline every time.

Adobe Firefly Video Editor “Quick Cut”

Credit to Adobe

If you’ve ever had a folder full of clips and that immediate “ugh, I have to find the story first” feeling, Adobe’s Quick Cut in the Firefly video editor (beta) is aimed at that exact moment.

Instead of starting from a blank timeline, Quick Cut automatically assembles an initial rough cut from the footage you upload, so you’re reacting to a draft rather than organizing chaos from scratch. It uses scene detection, shot selection, and audio analysis to pick cut points and pull “important” moments into a first pass, then you refine from there.

What you can do with it

• Generate a first draft faster: Upload footage, get an initial cut, then trim, reorder, and polish.

• Steer the edit with intent: Quick Cut can be guided by a text prompt plus adjustable settings.

• Shape deliverables earlier: You can tweak preferences like aspect ratio, video duration, and even add a B-roll track during generation.

Creator pros (what’s in it for you) ✅

• Less grunt work upfront: You spend less time scrubbing and more time editing decisions.

• Better momentum: A rough cut helps you spot what’s missing early, like B-roll gaps or pacing issues.

• Format-ready faster: Aspect ratio and duration controls matter if you’re cutting for socials.

Creator watch-outs (what to keep an eye on) ⚠️

• “Best shot” is subjective: Auto selection can favor technical clarity over emotional timing, comedic beats, or intentional awkwardness.

• Audio driven choices can skew the story: If your strongest moments are visual, audio analysis might underweight them.

• Risk of sameness: If teams rely on first drafts too heavily, edits can start to look algorithmically “tidy” rather than creatively distinct.

💡Our Take

Read the full details here.

State of Generative Media (Vol. 1): 2025 was the year “production” caught up to “play”

The State of Generative Media

fal’s State of Generative Media: Volume 1 is a snapshot of how generative media shifted in 2025. The main point: it’s moving from “fun experiments” to real production workflows across image, video, audio, and 3D.

Key takeaways

• Scale is the story. There isn’t one “winner model.” Teams are using a growing mix of tools, and they’re updating that stack often.

• Image is slightly ahead of video in production use. Video adoption is growing fast, but it’s still harder to control and standardize at scale.

• Organizations care about integration. Creatives may use apps, but companies increasingly want tools that fit into pipelines (often via APIs).

• ROI shows up when the use case is specific. Strong results tend to come from clear goals like faster iteration, versioning for campaigns, and content variations for testing.

• IP and ownership concerns are still the biggest brakes. Many teams want speed, but legal and brand risk questions can slow adoption.

What’s in it for creatives

• More shots on goal: faster iterations, more concepts, more variations without blowing up budgets.

• Better pre-production: quicker storyboards, pre-viz, pitch visuals, and campaign exploration.

• New expectations: clients may want more options faster, plus clearer answers on rights and provenance.

Read the full details here.

Amazon MGM is testing in house AI tools for film and TV production

Amazon MGM Studios is preparing to test in house AI tools designed to cut costs and speed up production workflows. A closed beta is planned for March 2026 with select industry partners, and the team expects to share early results by May 2026.

What they’re building

This effort is being run through a dedicated AI Studio led by Albert Cheng, set up like a small internal startup (engineers and scientists, with creative and business support). The tools are meant to slot into real production pipelines, focusing on practical bottlenecks like:

• Character and continuity consistency across shots and scenes

• Pre production + post production support (the parts that tend to eat time and budget)

• Editing assistance to speed up iteration, while keeping humans in control

Amazon’s framing is that these tools are meant to support creators, not replace them—while acknowledging the broader industry anxiety around roles, labor, and expectations.

One example they highlighted

Amazon pointed to its series House of David, where AI enhanced footage supported the creation of large scale battle sequences in a more cost effective way.

Why this matters for creators

This is a shift from “AI as a novelty” to AI as pipeline infrastructure. If it works, it likely means:

• More pressure for faster turnarounds on versions, edits, and deliverables

• More value on creators who can run workflows + quality control (not just generate assets)

• More urgency around credit, authorship, and accountability when AI touches editorial steps

Read full details here.

AI Fan Edit: Rewriting the Stranger Things Finale Battle

Chenran Ning shared a short experiment titled “Trying to Fix Stranger Things Finale Battle” — Episode 2, imagining Will, Eleven, and Eight vs. Vecna, made with Seedance 2.0.

The post is basically a “what-if” fan rewrite: using generative video to remix a well-known story beat, then asking the audience if they should keep going and finish the arc.

Why this is interesting for creatives

• It’s a clean example of AI as a rapid “scene prototyping” tool (test a narrative idea fast, then iterate based on feedback).

• The format doubles as community building: you’re not just posting a clip, you’re inviting viewers into the creative decision.

Watch-outs

• Fan edits in popular IP can be a gray zone—fine for experimentation, but be careful if you plan to monetize or use it commercially.

• “Made with X model” is helpful context, but it can also distract from story choices—so the edit still needs a clear creative intent.

Check the work here.

💡 Insight

The throughline this week is pretty blunt: making more content is getting easier. Standing out still comes from the decisions.

Here’s what that means in practice, based on this week’s mix of editing tools, studio pipelines, and fan-driven experiments:

• Direction becomes the differentiator. Tools can assemble a first pass (Quick Cut) or generate multi shot sequences (Seedance 2.0), but the “why this cut, why this beat, why this frame” is still on you.

• Design utility is the new competitive arena. Models like Recraft v4 aren’t only chasing prettier outputs, they’re chasing outputs that hold typography and ship in real layouts. That’s a real time saver, and it also raises the bar for brand consistency.

• IP and likeness risk stays loud. Celebrity prompts and fan edits are handy benchmarks, but they’re also where legal and brand risk spikes. Treat them as experiments, rebuild with cleared assets for anything commercial.

That’s the week. Same tools arms race, same opportunity: use speed to buy more thinking time, not more chaos.

That’s it for this week, folks.

Hi Creatives! 👋 

This week’s theme is simple: AI video is graduating from “one cool shot” to “can you actually finish a sequence?”

Kling 3.0 is clearly aiming at the hardest problem in AI video — keeping things consistent across shots so you can build a sequence, not just a single clip. Freepik’s AI Editing is coming at it from the other end: the finishing room stuff (cleanup, polish, quick fixes) that usually eats hours. Then Google’s Project Genie is a reminder that image-to-“world” is creeping into mainstream workflows, turning a single photo into something you can explore, not just reference.

On the content side, The Chronicles of Bone becoming a Freepik Original feels like a signal that AI work is moving toward repeatable formats (series, styles, pipelines) instead of one-off experiments. And the last two highlights — hybrid filmmaking andArtlist’s “3 Big Game ads in 5 days” experiment are basically the same lesson in different outfits: tools help, but the real differentiator is direction and taste when the timeline is tight.

In other words: making is easier. Finishing and choosing well still isn’t.

This Week’s Highlights:

• Kling 3.0 update: it’s trying to fix the hardest part of AI video

• Freepik’s AI Editing is basically “the finishing room” for your videos

• Google’s “Project Genie” turns a photo into a playable mini-world

• The Chronicles of Bone (now a Freepik Original)

• Hybrid filmmaking, directed with taste (and a faster pipeline)

• Artlist’s “3 Big Game ads in 5 days” experiment

🎥 Kling 3.0 update: it’s trying to fix the hardest part of AI video

Most AI video tools can spit out one impressive shot. The struggle starts when you need shot 2, shot 3, and they all feel like the same world.

That’s what Kling 3.0 is aiming at: multi shot flow, stronger consistency, and built in audio, so you can move from “cool clip” to “usable sequence” with less patchwork. 🧩🎬

What’s new (and why creatives should care)

• Multi shot storyboarding 🗂️
  Instead of praying one long prompt holds together, Kling 3.0 is being positioned around scene based generation, where a single video can be structured into multiple shots with defined beats and timing.

• Native audio inside the generation 🔊
  The pitch is “audio and video together” so you are not always exporting and rebuilding sound elsewhere. Some partner pages also mention voice integration and more character specific audio.

• Consistency tools (“Elements” system) 🧍‍♀️🧍‍♂️
  Better ability to keep a character or key visual element consistent across shots, which is basically the difference between a demo and a real workflow.

• Up to ~15s clips (commonly cited) ⏱️
  Still short, but short plus multi shot structure can be more useful than longer clips that drift.

What this means for creative workflows

If Kling 3.0 delivers on consistency, it’s a big deal for:

• Short form ads where you need 3 to 6 shots that match

• Narrative tests for music videos, branded stories, film previz

• Style locked series where you need repeatability across episodes

Still, the real job stays the same: taste and editing. You will likely spend less time regenerating from scratch, and more time choosing what to keep.

Here’s a quick comparison from the previous version Kling 2.6

The Way of Kings by Brandon Sanderson.

To make the Kling 3.0 update feel less abstract, here’s a concrete example from PJ Accetturo (AI filmmaker)

He says he used Kling 3.0’s Multi-Shot (Multi-cam) workflow to build a short opening sequence in two days, and calls out two things as the big leap:

• Multi-shot continuity: one image → one continuous output → multiple angles/shot lengths you define

• Speaking / dialogue realism: he says the “speaking is greatly improved,” and when paired with multi-cam it starts to feel close to live-action quality

He also shares the rough workflow he used: upload a single image, choose Multi-cam, set shot lengths, then describe how each shot differs (he even mentions using start frame + end frame for character clarity)

Check his work here.

Freepik’s AI Editing is basically “the finishing room” for your videos

You know when a video is almost good… but it still feels a bit flat? Like the pacing is right, the shot is decent, but it’s missing that last 10% polish that makes people actually stop scrolling.

That’s the vibe of Freepik’s latest post: AI editing inside the Clip Editor is built for finishing. Not replacing your footage, just helping you tighten motion, clean audio, and give the whole piece a more intentional look.

What Freepik’s Clip Editor is trying to solve

Instead of bouncing between 3–5 tools for little fixes, the Clip Editor is positioning itself as a one place “final pass” workflow:

• Motion + audio + look handled in the same editor

• Fast adjustments right on the timeline, so polishing doesn’t feel like a separate painful step

• Automation, but still flexible, so you’re not stuck with one default style

The featured AI tools (simple breakdown)

Here are the main ones called out in the article:

• Motion Shake 
  Adds controlled camera movement to a clip so shots feel less stiff. Useful for action beats, transitions, or moments that need energy.

• Audio Isolation 
  Helps separate voices (or other sounds) from messy background noise. Great for street interviews, event clips, or anything recorded in chaotic environments.

• Video Effects 
  Quick “finish” looks you can apply to unify a video without hours of tweaking. Think: a fast way to make a set of clips feel like they belong together.

And the article also points to other tools in the ecosystem that can help with finishing touches:

• Color Grading for mood consistency

• Lipsync for matching speech and mouth movement

• Topaz Upscaler for higher quality exports (when your source footage needs help)

Read full details here.

Google’s “Project Genie” turns a photo into a playable mini-world

Google’s experimenting with Project Genie (powered by DeepMind’s Genie 3): you give it one image or a prompt, and it generates a tiny interactive world you can move around in. Think “walkable moodboard,” not a full game (yet).

What’s in it for creatives

• Moodboards become explorable: you can “walk” a vibe instead of describing it.

• Faster iteration on environment design: generate many options, then pull the best ideas into real production (3D, set design, comp).

• Bridging film + game language: directors and DPs can test spatial rhythm and camera feel earlier.

The catch (and it’s a real one)

• The current experience is short + limited control, so it’s not replacing Unreal/Blender pipelines.

• As this improves, IP similarity + authorship questions get sharper: who owns a “world” generated from references, and how close is too close?

The Chronicles of Bone (now a Freepik Original)

The Chronicles of Bone is now officially a Freepik Original series.

We’ve been following Kavan Cardoza’s fantasy series The Chronicles of Bone, and Episode 4 just dropped—with a big update: it’s now officially a Freepik Original series. 🎬✨

This chapter, “Prologue Part Four: The Outlaw,” kicks off a darker Robin Hood origin story: the fall of Loxelan, a childhood marked by loss, and the rise of an outlaw who becomes a symbol of hope while fighting a vampiric empire. 🏹🩸

There are two more prologue episodes coming to complete the Prologue Season—then Season One begins.

Watch the full episode here.

Hybrid filmmaking, directed with taste (and a faster pipeline)

Watch the full video here

We’ve been watching filmmaker Noah Miller explore what “hybrid filmmaking” looks like when you start with real actors—then bring in generative tools to finish the job. Using fal, he shoots live-action first, and uses AI to test styles, create inserts, and solve the tricky in-between shots that usually eat up time (or budget). 🧩🎥

What we like about this approach: it frames AI as a production partner, not a replacement. The story still starts with performance and direction—AI just helps expand the range of what’s possible for smaller teams and creators who haven’t had access to big pipelines. ⚡

This video is part of Beyond the Loop, a storytelling series by Wonder Studios in partnership with fal.

Check his work here.

Artlist’s “3 Big Game ads in 5 days” experiment

We didn’t make just one Big Game ad.

We’re spotlighting a fun (and very telling) experiment from Artlist: they didn’t produce just one Big Game style ad — they built three different versions in five days, using the Artlist AI Toolkit.

Each version explored a different angle of Big Game advertising:

• The past: a nod to iconic ads that shaped the category

• The present: a rapid response to what’s culturally relevant right now

• The future: a peek at where ad making might be headed

They ultimately shipped the “present” version — and the interesting part isn’t just speed. It’s the strategy: make multiple creative directions quickly, then choose based on what’s resonating in culture this week, not two months ago.

For creative teams, this is a good reminder that AI’s biggest advantage isn’t “one perfect spot.” It’s the ability to run more creative reps and commit with more confidence.

Check the work here.

💡 Insight

Here’s the slightly uncomfortable (but kinda helpful) truth: as the tools get better, “good enough” gets crowded. When everyone can make a decent clip, the work that stands out is the work with clear choices behind it — continuity, taste, and intent.

This week’s highlights all point to the same shift:

• Kling 3.0 is basically trying to solve the hard part of AI video: keeping things consistent across multiple shots, so you can build a sequence and not just a single “wow” moment.

• Freepik’s AI Editing is leaning into the part of the process nobody posts about: the finishing room. Cleanup and polish matter more when “first draft” is easy.

• Google’s Project Genie turns a photo into a playable mini-world, which feels like a nudge toward pre-vis and worldbuilding becoming more everyday, not just for game studios.

• The Chronicles of Bone as a Freepik Original is a signal that AI work is moving toward repeatable formats (series, pipelines, a consistent look) instead of one-off experiments.

• Hybrid filmmaking + Artlist’s “3 Big Game ads in 5 days” experiment both underline the same lesson: speed helps, but when the clock is brutal, the edge is still direction and editing.

So if you’re thinking “what’s my edge now,” here’s a simple try-this-next-week:

• Think in sequences, not shots. Even rough story beats help.

• Pick a few rules and stick to them. Style bible, camera logic, what stays fixed.

• Use the tools for options, not your identity. Your taste is the part that doesn’t scale.

If one of these highlights actually maps to your workflow, reply with the one you’re most curious about — and I’ll do a creator-focused breakdown next edition.

That’s it for this week, folks.

Hi Creatives! 👋 

This week has a very specific vibe: fewer flashy demos, more workflow upgrades. The kind that help you keep a look consistent, generate options that belong together, and collaborate without turning your process into a mess.

If your work lives in sets, rounds, and reviews, these updates will feel… oddly practical.

Big signal across the list: the workflow is becoming the product. Tools aren’t just adding models. They’re trying to become the room where decisions happen.

This Week’s Highlights:

• Freepik Variations: one image in, a whole set of options out

• Dear Upstairs Neighbors, a Case Study in AI-Assisted Animation

• Claude now opens Slack, Figma, and Canva inside the chat

• FLORA raises $42M to build a “unified creative environment.”

• Rethink Reality: “World models” are getting weirdly practical

• Google Photos: “photo to video” just got text prompts

Freepik Variations: one image in, a whole set of options out

In our last newsletter, we briefly introduced Freepik’s Variations. Here’s the practical “what you can do with it” version.

Variations is designed for the moment when a hero image gets approved and the next request is, “Great. Now we need the close up, the vertical, and a few alternates for different audiences.”

What Variations is (in plain terms)

Freepik Variations lets you take one image (uploaded or generated) and spin out multiple consistent options in a grid so you can compare, pick, and move on without constantly restarting from scratch. 🗂️

The main modes (aka: where the value is) 🎛️

• Reframe

AI-Generated Image

Generate different camera angles, framing, and compositions while keeping the same scene direction.
Best for: turning one hero into wide, close-up, and vertical placements.

• Storyboard

AI-Generated Image

Expand one image into a connected sequence of shots that feel like the same world.
Best for: planning beats and continuity before you move into motion.

• People

  

  AI-Generated Image

  
  Explore human subject changes while keeping the concept consistent, including:

  • Demographics 

  • Expressions 

  • Age 

• Custom
  Prompt guided changes like mood, lighting, environment, or concept shifts.

Why creatives should care

• Faster review cycles: grids make feedback easier and less chaotic

  More output from one direction: less “same thing, slightly different” rework

• Better prep for motion: storyboard frames can act like key moments before video

Quick watch outs ⚠️

• Consistency helps, but art direction still matters. Don’t let the grid decide for you.

• People variations are powerful, but use them thoughtfully for brand fit and cultural nuance.

Read full details here

Dear Upstairs Neighbors, a Case Study in AI Assisted Animation

Dear Upstairs Neighbors is previewing at the Sundance Film Festival, and it’s one of the better examples of AI showing up as a production helper, not the author.

Led by Pixar alumni Connie He and Márcia Mayer alongside veteran creatives and AI researchers, the team translated Connie’s hand painted canvases, storyboards, and 2D line animation into the final short. Their workflow included custom Veo and Imagen fine tuning on the artwork, using rough animation as visual direction, and targeted region edits to iterate without restarting whole shots.

Read full details: https://goo.gle/45uM34K

Claude now opens Slack, Figma, and Canva inside the chat

AI-Generated Image

Claude now opens Slack, Figma, and Canva inside the chatClaude can now run interactive “apps” inside the chat (via MCP Apps), so instead of only describing what to do in Slack/Figma/Canva, you can work with those tools in-panel while Claude helps.

Now, with an MCP extension called MCP Apps, Claude can open supported tools as interactive apps inside the chat so you can:

• Draft and format Slack messages and preview them before sending 💬

• Work with Canva decks in real time, as an actual editable interface 🖼️

• Use Figma or FigJam for diagramming and planning flows 🧩

• Manage project boards with tools like Asana or monday.com ✅

• Pull files from Box, depending on what your org has enabled 📁

• Explore charts with tools like Hex or Amplitude 📊

The key detail: it’s no longer just “Claude told me what to do.” It’s “Claude is operating alongside the interface where the work lives.”

3 ways to use it right away

• Slack approvals: summarize a feedback thread → draft a clean “here’s what we’re changing” update 💬

• Figma / FigJam: turn messy notes into a structured plan or flow map

• Canva decks: outline → tighten story → refine slide structure in one session

Quick do’s and don’ts ✅🚫

✅ Do

• Start with low risk tasks: drafting, formatting, planning, summarizing feedback

• Use clear review steps: “draft only,” “do not send,” “propose changes first”

• Keep a human checkpoint for anything client facing

🚫 Don’t

• Connect sensitive workspaces without understanding permissions

• Treat outputs as final, especially for brand voice, legal claims, or client commitments

• Let tools sprawl into a new mess. Fewer surfaces is the goal 🙂

Read full details here

FLORA raises $42M to build a “unified creative environment.”

FLORA has raised a $42M Series A, led by Redpoint, to build a unified creative environment for the generative era.

FLORA just raised a $42M Series A (led by Redpoint) to keep pushing toward a single workspace where creative teams can mix text, image, and video models in one workflow. 🧩

Key insights

• Solves a real pain: AI tooling is still fragmented, and “best models” often means platform hopping

• Their focus is team-first: workspaces, real-time collaboration, unlimited seats

• Use cases they call out: brand marketing, product visualization, film/VFX, with faster iteration cycles

What’s in it for creatives

• Less “one cool output,” more repeatable workflow: iteration, versions, reviews, handoffs in one place

• Watch-outs: “unified” still depends on model access, cost, and workflow depth as teams scale

Rethink Reality: “World models” are getting weirdly practical

This episode is basically a philosophical jam session that keeps bumping into real product details. The guest (World Labs) argues that “seeing” and “doing” are tightly linked, and that the next leap in AI is less about prettier pixels and more about spatial, navigable worlds that can support creators, game devs, VFX teams, and even robotics.

World models are starting to feel like a workflow, not a demo

This podcast convo hits a point the creative community is already feeling: AI isn’t just “make me an image.” It’s moving toward build me a space I can direct.

They talk about how intelligence is basically a loop between sensing and action. And that matters because “world models” only get interesting when they stop being flat outputs and start behaving like something you can work inside.

The part creators should pay attention to

World Labs’ model, Marble, is framed like a material for building worlds, not a one off generator.

What that signals for creative work:

• True 3D matters. “Behind,” “under,” “move the camera 20 feet” becomes a real instruction, not a vibe.

• Control > surprise. The goal is less slot machine generation, more director style camera blocking and continuity.

• Pipeline thinking is showing up. Stuff like collision meshes and layout tools are small details that scream “this is heading toward gaming, VFX, previs, and simulation workflows.”

They also mention “Chisel” (layout guided generation), which is basically: block out a rough 3D structure, then generate within constraints. That’s the kind of feature that turns AI into a teammate for set building instead of a mood board machine.

The take that’s worth holding with two hands

They’re pretty honest about the double edge: world models can empower creators, but they can also level up surveillance and misuse. The episode’s stance is basically: tech always spreads, so the real question is whether we build the social rules, norms, and ethics fast enough to keep creators from being the collateral.

What’s in it for creatives

• Faster previs and shot planning with less guesswork

• More consistent camera language and space continuity

• A path toward interactive scenes (and eventually moving worlds)

Watch the full episode here.

Google Photos: “photo to video” just got text prompts

AI-Generated Image

If you’ve tried Google Photos’ photo to video feature before, you know the vibe: you got “Subtle movement” or “I’m feeling lucky”, and you mostly just… hoped. Now Google Photos is adding text prompts, so you can describe how you want the image to move, what style to lean into, or what effect you’re after.

Why creatives should care

• Your camera roll becomes quick b-roll: slow push-ins, parallax, gentle motion for Reels/Shorts.

• Better for content variations: one hero image can turn into a few motion options fast.

• This is another sign that consumer apps are absorbing creator controls that used to live in niche tools.

Quick prompt ideas

• “Slow handheld push in, shallow depth of field.”

• “Gentle wind movement in hair and clothing.”

• “Cinematic parallax, subtle pan left.”

• “Dreamy slow zoom, warm light leaks.”

The tradeoffs

Pros

• Faster iteration than exporting to a dedicated video tool for simple motion.

• More intentional results than “I’m feeling lucky.”

• Audio by default reduces finishing steps for casual posts.

Cons / watch-outs

• Creative control is still bounded (you’re not getting full timeline editing or shot continuity).

• Safety and misuse risk exists with any tool that animates real people, and The Verge notes how similar prompting tools have been abused in other apps.

• If you’re working with client assets, treat this like any generative step: get internal approval, confirm rights, avoid using identifiable faces without consent.

Read full details here.

💡 Insight

This week’s highlights feel like the industry quietly agreeing on a new workflow:

You can see the pattern across the list. Freepik Variations is basically campaigning built into a button: one input, a usable set of options. Claude inside Slack/Figma/Canva is the assistant moving into the room where decisions actually happen, not just the playground. FLORA’s $42M is a bet that pros want a unified environment, not five tools stitched together by exports. And Dear Upstairs Neighbors is the case study version of the same idea: human taste leads, AI helps carry the look through the pipeline.

Even the “weird” stuff is getting practical. Rethink Reality’s world models points at a future where you’re not just generating shots, you’re maintaining a consistent “world” you can iterate on. And Google Photos adding text prompts to photo to video is the reminder that pro-ish controls are sliding into consumer tools fast, which will raise expectations for everyone.

What’s changing (from my seat)

• The unit of work is becoming the set, not the single image or clip.

• The real advantage is less context switching and better versioning.

• Taste gets louder. When options are cheap, your filter is the craft.

The upside for creatives

• Faster exploration without losing continuity

• More client-ready options per round

• A clearer path to production workflows for small teams

The risk

• You can ship faster while saying less. “Smooth” can become forgettable.

• Workflow templates can quietly turn into style templates if you don’t define what makes it yours.

That’s all folks.

Hi Creatives! 👋 

This week’s highlights all point to one thing: AI is getting absorbed into the parts of the workflow that actually ship. Not just generating clips, but helping with the in-between problems we fight every day, like patching shots, matching continuity, and keeping a process repeatable.

If you’re building content for campaigns, clients, or a team review process, this week is basically less “generate another one,” more “make it fit the cut.”

And the clearest signal is the World API going live. When “world building” becomes something you can call, iterate, and version like any other creative primitive, it’s not a demo moment, it’s infrastructure.

This Week’s Highlights:

• Editing gets faster (the “boring” parts of polish are getting real tooling)

• The World API Is Live: “world creation on demand.”

• 🎬 Veo 3.1 becomes more creator-practical: vertical-first improvements, reference images, upscaling

• 🎬 AI-first film signal: Bal Tanhaji prequel (Bal Tanhaji) first look

• “Gym Rats”: a reproducible AI film pipeline that treats workflow like the product

The New Bar for GenAI in 2026: Reliability

Production-ready means repeatable, reviewable, and accountable.
Enterprise teams aren’t buying AI. They’re buying reliability.

If you’re in Park City for Sundance this weekend, I’m hosting a small, invite-only happy hour for Freepik in partnership with Asteria. We’ll kick off with a short panel on what “production-ready” GenAI actually looks like in 2026, from creative craft to enterprise adoption.

Panelists: Paul Trillo, Henry Daubrez, Paula Vivas (moderated by me)

We’ll cover:

• Consistency at scale: keeping style, characters, and continuity intact when multiple people touch the workflow

• Operational reality: approvals, handoffs, and version control across creative, post, and marketing teams

• Governance and risk: rights, provenance, brand safety, and how teams set guardrails

• The real definition of production-ready: real deadlines, many deliverables, and failure that costs time and money

Invite-only. RSVP: https://luma.com/im18bfuq
If you’re a fit and want details, DM me.

The part of editing we all “love” just got faster

You know that moment when you’re mid edit, you almost have the shot… but you need to isolate a subject, soften a background, or pull a clean mask on a moving object and suddenly it’s 45 minutes of tiny keyframe pain?

Adobe just aimed straight at that problem with new AI powered updates in Premiere and After Effects, announced around Sundance season.

The headline update (Premiere) 🧩

• AI Object Mask: hover over a subject/object, click, and Premiere generates a trackable mask in seconds.

• Redesigned Shape Masks: faster tracking and smoother controls for the usual ellipse/rectangle/pen masking work.

• Frame.io panel inside Premiere (beta): review comments + versions stay closer to your timeline (less tab ping pong).

The motion design side (After Effects) ✨

• Native SVG import as editable shape layers: cleaner “design → motion” handoff for logos, UI, icons.

• Parametric 3D meshes + Substance material support: more “build and texture inside AE” without instantly needing extra tools.

Why this matters for creative teams 👀

Key shift: masking is becoming a default capability, not a specialist chore. When isolation and tracking get faster, you naturally try more variations: selective grades, localized blur, product spotlighting, background tweaks, quick relights.

A practical insight

This is Adobe quietly reinforcing a new creative advantage: iteration speed becomes part of your taste.
When the technical friction drops, the winning editors aren’t the ones who can rotoscope the cleanest. It’s the ones who can:

• choose the right moment to isolate a subject

• know when not to add an effect

• move through options quickly without derailing the story

Also worth noting for client work: Adobe is emphasizing workflows that fit real production constraints (speed, collaboration, and “can we use this on brand work safely?”), which is where these updates start to pay off.

Read full details here.

The World API Is Live

A new way to generate fully explorable 3D worlds from text, images, and video.

Picture this: you have a moodboard, a few frames, maybe a quick phone video of a location… and instead of sending it to a 3D team (or opening Blender and spiraling), you call an API and get back a navigable 3D environment you can explore in a browser.

That’s the promise of World Labs’ new World API, built on their multimodal world model Marble. It generates explorable 3D worlds from text, images, panoramas, multi view inputs, and video, and is designed to plug into apps and workflows as “world creation on demand.” 🧩

This is “previs (pre-visualization) and spatial ideation” getting software-shaped

Instead of worldbuilding being a one-off asset project, it becomes something you can generate, iterate, version, and integrate like any other creative primitive.

If you’re a:

• filmmaker / studio 🎬: think virtual scouting, blocking, camera angle exploration, and grabbing high quality stills for pitch decks or shot planning.

• game / immersive creator 🎮: turn 2D references into spaces you can prototype inside faster.

• architect / spatial designer 🏛️: generate a walk-throughable concept early, before you commit to heavy modeling.

World Labs even points to early integrations in filmmaking workspaces and architecture tooling, which is a signal they’re aiming beyond “cool demo” into “workflow infrastructure.”

Practical bits (the producer brain stuff) ✅

• Credit based pricing 

• Async generation (you request a world, it generates, you fetch results)

• Rate limits apply

• You’ll need an account + API key 

Tiny “do / don’t”

Do: treat it like previs and alignment tooling first.
Don’t: assume it replaces production design or environment art, or use client assets without clear rights/approval.

Read full details here

🎬 Veo 3.1 just got way more “creator practical”

Ctto.

Veo 3.1’s “Ingredients to Video” workflow now puts mobile-first vertical video front and center, improves character and scene consistency using reference images, and adds upscaling to 1080p and 4K for cleaner outputs that hold up better in editing. Also, Google is emphasizing watermarking and verification as the content gets more lifelike.

What’s new (key details)

• Native vertical (9:16) output in the Ingredients to Video workflow, so it composes for phones from the start. 📲

• “Ingredients to Video” consistency upgrades: you can use up to 3 reference images to steer characters, backgrounds, textures, and reuse elements across clips. 🧩

• Upscaling options: base output can be enhanced to 1080p and 4K for cleaner edits and deliveries. ✨

• Where it’s showing up: rolling into places like the Gemini app, Flow, and integrations tied to YouTube Shorts / YouTube Create. 🎬

• Trust layer: Google continues to push watermarking / verification (SynthID) as outputs get more realistic. 🧾

Why this matters for creatives

This isn’t just “better gen.” It’s Google leaning into repeatable social production:

• Vertical-native means fewer compromises for mobile campaigns.

• Reference-image “ingredients” are a step toward shot families (multiple clips that feel like the same world), which is what brands actually need.

Insight 💡 

“Ingredients” is starting to behave like a creative kit system: instead of making one clip at a time, you can build a library (character + product + set + texture) and generate variations without reinventing the entire scene. That’s the bridge from experimental clips to campaign workflows.

Read full details here

🎬 A Bollywood prequel is going “AI-first”: Bal Tanhaji drops a first look

If you work in visuals, this one is worth a pause.

Ajay Devgn just shared the first look for Bal Tanhaji, described as an AI powered origin story that expands the world of Tanhaji: The Unsung Warrior into something built for multi platform audiences, not just theaters.

Why this matters for creatives

A genuinely “useful” assistant changes the shape of creative work, not by replacing taste, but by compressing the annoying parts:

• Pre production at speed: turning a messy brief into shot lists, prop lists, alt concepts, and production checklists faster.

• Cross app handoffs: moving from “idea” to “calendar, notes, tasks, and files” without you playing project manager for your own brain.

• Personal context as a feature: if Siri can reference your project history, brand notes, and prior decisions, your creative direction gets more consistent across iterations.

If this actually lands, the biggest win is less about novelty and more about fewer context resets.

The tradeoffs (quick version)

1. Style flattening is real
   When the same assistant becomes everyone’s default, outputs start to sound alike. Use it for structure, then add your taste so your work doesn’t blend into the feed.

2. “Privacy forward” ≠ “client safe”
   Even with Apple’s Private Cloud Compute, don’t paste sensitive briefs, unreleased product details, contracts, or performance data into any assistant. Think lower risk, not no risk.

3. This is a distribution deal, not just a feature
   Apple ships a stronger Siri faster. Google gets massive default placement. Creatives inherit new expectations around speed, volume, and “just ask the assistant.”

Read more details here.

‘Gym Rats’ — a reproducible AI film pipeline

What I found paints a picture of a filmmaker treating AI like a real production pipeline instead of a magic button — planning shots, building assets, and iterating between fast drafts and detailed final passes to make narrative work like shorts or ads feel intentional and controlled

Here’s what’s interesting from Gabe Michael’s public work and the surrounding material online:

🎬 A director’s mindset, not an “AI trick”

• Gabe Michael is known as an AI filmmaker and creative technologist focusing on storytelling with AI tools, not just one‑off generative outputs.

• His videos show real projects — like short films using Nano Banana Pro images animated through Veo (Veo 3 / Veo 3.1) — which aligns with the idea of layering tools for quality and control rather than relying on a single prompt.

🤖 Tool workflow insights

• Nano Banana Pro is highlighted in multiple tutorials as a go‑to for hero image generation and reference frames before animation.

• Veo (including Veo 3 / 3.1) appears frequently as the next step to animate those still frames into motion, reinforcing a two‑stage workflow (fast draft → refined move).

• Other creators also demo combining these tools (Nano Banana + Veo/VO3) to bridge generation and motion in video projects.

See his workshop and process breakdown here.

💡 Insight

This week’s updates feel less like “new toys” and more like the industry quietly agreeing on a new workflow:

AI is moving from “make me a clip” to “help me finish the cut.”

When World API turns explorable world creation into something you can generate on demand from text, images, or video, that’s not just another tool drop. It’s a signal that the next wave is edit and production workflows pulling generative capabilities into the pipeline, where iteration, versioning, and handoffs actually live. And when tools like Veo lean into creator practical improvements like vertical, reference images, and upscaling, they’re acknowledging what most of us need: not one impressive shot, but a set of shots that belong together.

Then you look at something like “Gym Rats” and the AI-first film headlines, and the message stays consistent: the creative edge isn’t “generate more.” It’s run a repeatable process without losing taste.

What’s changing in the workflow (from my seat)

• Editing becomes the center of gravity again. Gen tools are starting to orbit the timeline, not the other way around.

• Pre-pro matters more, not less. References, constraints, and shot planning are increasingly the difference between “cool” and “client-safe.”

• Taste becomes visible. When iteration is cheap, the value is in what you choose to keep, what you cut, and how you maintain continuity.

The upside for creatives

• Less time lost redoing sequences just to fix one transition

• More continuity across shots without brute-force prompting

• A clearer path to production-ready workflows, especially for small teams

The risk

• You can ship faster while saying less. Bridge shots + upscaling can make everything “smooth”… and also forgettable.

• Workflow templates can quietly turn into style templates. If your process doesn’t include “what makes this ours,” the outputs won’t either.

A small rule to try next week
Before you generate anything, write one sentence:
“This piece should feel like ___, not like ___.”
Then use the tools to serve that sentence, not replace it.

Hi Creatives! 👋 

This week’s theme is simple: the “default” is getting smarter, and that’s going to change what clients, teams, and audiences expect from your work.

When Apple and Google start reshaping the assistant stack (yes, Siri gets Gemini), it’s not just tech gossip, it’s a signal that creative workflows are moving closer to the OS level. At the same time, tools like LTX-2 and Freepik’s Change Camera are pushing generative video and image edits toward something that looks less like “a demo clip” and more like an editable pipeline. Add Niji 7 for creators who live in illustration and stylization, plus “Variations + Storyboard” features that nudge people toward planning instead of random spawning, and you’ve got a week that quietly screams:

Speed is cheap now. Direction is the premium.

This Week’s Highlights:

• Apple + Google redraw the assistant map, Siri gets Gemini

• Real time captions that actually keep up

• 🎬 LTX-2, the “ship ready” AI video update

• Niji 7, cleaner style control for illustration workflows

• Freepik Change Camera

• Variations + Storyboard, planning becomes a feature

Everything Everywhere AI at Once (Upscale Day LA)

I’ll be at Everything Everywhere AI at Once (Upscale Day LA), a director’s cut style meetup for the people building and using AI in film. Expect real talk on what’s exciting, what’s risky, and what “craft” should still mean when workflows get faster.

Why it’s worth your time

• Keynote from Joaquín Cuenca (Freepik CEO)

• AI in film workflow session, plus an AI video showcase

• Panel on how AI is reshaping production, talent, ownership, and ethics

• Networking, DJ, and a long hang for meeting collaborators

Seats are limited + approval required. If you’re in LA and working in film, VFX, directing, production, or creative tech, come through.
👉 RSVP here: Upscale Day LA event page

Apple + Google redraw the assistant map, Siri gets Gemini

Ctto

Apple and Google announced a multi year collaboration where the next generation of Apple Foundation Models will be based on Google’s Gemini models and cloud tech, and those models will power future Apple Intelligence features, including a more personalized Siri expected this year. Apple also says Apple Intelligence will continue to run on device and through its Private Cloud Compute, keeping Apple’s privacy posture.

Why this matters for creatives

A genuinely “useful” assistant changes the shape of creative work, not by replacing taste, but by compressing the annoying parts:

• Pre production at speed: turning a messy brief into shot lists, prop lists, alt concepts, and production checklists faster.

• Cross app handoffs: moving from “idea” to “calendar, notes, tasks, and files” without you playing project manager for your own brain.

• Personal context as a feature: if Siri can reference your project history, brand notes, and prior decisions, your creative direction gets more consistent across iterations.

If this actually lands, the biggest win is less about novelty and more about fewer context resets.

The tradeoffs (quick version)

1. Style flattening is real
   When the same assistant becomes everyone’s default, outputs start to sound alike. Use it for structure, then add your taste so your work doesn’t blend into the feed.

2. “Privacy forward” ≠ “client safe”
   Even with Apple’s Private Cloud Compute, don’t paste sensitive briefs, unreleased product details, contracts, or performance data into any assistant. Think lower risk, not no risk.

3. This is a distribution deal, not just a feature
   Apple ships a stronger Siri faster. Google gets massive default placement. Creatives inherit new expectations around speed, volume, and “just ask the assistant.”

Read more details here.

Real time captions that actually keep up

Credit to ElevenLabs

You know that moment when you’re mid interview, mid livestream, or mid client review and someone says the one line you need to clip later… and you’re stuck scrubbing the timeline to find it?

ElevenLabs just released Scribe v2 Realtime to make that pain smaller: live transcription in under ~150ms, designed for real time captioning, meeting assistants, and voice agents. 📝⚡

What’s new (creator relevant highlights)

• Real time transcription under 150ms so captions feel in sync

  90+ languages plus auto language detection so bilingual sessions are easier.

• “Negative latency” prediction that guesses next words and punctuation to keep text feeling instant.

• Manual commit + Voice Activity Detection so you can control when text “locks” for cleaner captions and notes.

• Enterprise privacy options like Zero retention mode and EU and India data residency for sensitive client work.

Why this matters for creatives

If you edit, produce, or run sessions live, this can tighten your whole workflow:

• Faster clips: searchable transcript while you’re still recording.

• Cleaner captions: fewer rewrites mid sentence if you use commit control.

• Better collab: producer, editor, and social can pull selects from the same text doc.

Read full details here.

🎬 LTX-2, the “ship ready” AI video update

Credit to LTX

If you’ve ever loved a generated clip… then immediately hated it in the edit because motion got weird or identity drifted, LTX-2 is trying to solve that part of the job.

Key specs creatives will care about

• Up to 20 seconds per clip 

• Native 4K output

• Up to 50 FPS for smoother motion and better slowdowns

• Audio + video generated together so timing can feel more cohesive

• Two model options: ltx-2-fast (speed) and ltx-2-pro (quality)

Why this matters for your workflow

• More usable iterations: Longer clips plus stronger control means fewer “restart from scratch” cycles.

• Better for story and brand: Consistent motion and camera behavior is what makes content feel intentional, not random.

• Audio included: Even if you replace it later, synced audio can speed up early cuts and pacing decisions.

Read full details here.

Niji 7, cleaner style control for illustration workflows

Niji 7 (niji Midjourney’s newest model) is aimed at one thing creatives actually care about: cleaner, more consistent anime outputs with fewer rerolls.

What’s new (key details)

• Better anime coherence: faces, eyes, and small details hold together more reliably 👁️

• Stronger prompt understanding: more literal + more controllable (great for art direction) 🧭

• Improved text rendering: nicer for posters, panels, signage, UI-ish concepts 📝

• Better sref performance: style reference is a bigger win in this release 🎨

• No cref in v7: character reference isn’t supported (they hint a different solution later) 👀

Why this matters for creatives

• Production-friendly iteration: fewer “almost” images means faster client rounds and series work (key art, webtoon panels, campaign visuals).

• More direction, less chaos: when a model is more literal, your art direction actually sticks, but you’ll want to be more specific in prompts.

Quick prompting tip ✅

If you want consistency: add counts + placement (left/right, foreground/background) and keep the first version simple before you go cinematic.

Read more details here.

Change Camera (simple body) 📸🔁

Ever get one perfect image… then need a different angle for an ad, deck, or storyboard?

Freepik’s Change Camera lets you take one image and generate new camera viewpoints (think: same scene, different angle) so you don’t have to start from scratch.

Why it matters for creatives

• Faster coverage: hero shot, 3/4 view, close-up, alternate framing

• More consistency: keeps the vibe while you explore angles

• Better storyboarding: build a sequence from one strong frame

Read full details here: the official announcement + tutorial.

Variations + Storyboard (simple body) 🎬

Introducing Variations One visual is all you need to create complete narratives.

One image → multiple directions → a full sequence.

Start with a single frame on Freepik, open Variations to generate different takes, then use Storyboard to turn them into a multi-shot flow fast.

Why it matters for creatives

• Turn key art into a sequence for campaigns, pitches, and social series

• Move like a director: pick the path, not random outputs

• Speed: from one image to a storyboard-ready set of shots

Read full details here: the official Variations post + storyboard tutorial.

💡 Insight

Here’s the uncomfortable truth that’s also kind of freeing: as tools get better, “good enough” gets crowded. When anyone can generate a decent image, a decent clip, a decent caption, the work that stands out is the work with taste, constraints, and intent.

This week’s updates point to a shift in creative advantage:

• Assistants are becoming infrastructure, not apps. That means creative teams will soon be judged on how well they design their workflows, not how fast they can click around.

• Camera control and storyboarding features are a tell, the winners won’t be the people who generate the most, they’ll be the people who can art direct, revise, and keep continuity.

• Real time speech and captions reduce friction, which is great for accessibility and speed, but also raises expectations. If everything is easier to produce, audiences get pickier faster.

So if you’re wondering “what’s my edge now,” try this for the next week:

1. Write the brief like a director, not a prompter.

2. Choose constraints on purpose, style bible, camera rules, references, what you refuse to do.

3. Use the tools for iteration, not identity. Your taste is the brand.

If any of these highlights hit your workflow in a real way, reply with the one you’re most curious about, and I’ll do a creator focused breakdown next edition.

That is all for this week, folks.

Hi Creatives! 👋 

This week is about creative control with fewer surprises, in your pipeline and in your rights. Tools are blending together faster, “motion control” is becoming a default expectation, and 2026 is shaping up to be a very real year for AI copyright decisions that will land on creators first.

This Week’s Highlights:

• Adobe + Runway are moving closer (what it could mean for end-to-end workflows) 🎬

• 2026 is a big year for AI copyright, and creators will feel it in contracts and distribution

• Kling 2.6 Motion Control is now in Freepik (more direction, fewer rerolls)

• 🎬 Creative Feature

  • Tests LTX-V2 (Open-Weight AI Video)

  • AI Productions x Stereocolor on Hybrid Advertising

2025 was the year generative media became operational.

In 2025, generative media stopped being a “cool demo” and started behaving like a real production tool. Not because the outputs suddenly became perfect, but because editing and consistency finally got usable enough for teams to iterate without restarting from scratch. Once identity, style, and continuity hold up, you can run the boring (and necessary) parts of production: shot lists, continuity notes, review cycles, controlled revisions, and clean handoffs.

As we head into 2026, the momentum is shifting from prompting in chats to building systems. Node based workflows are turning creative experiments into repeatable pipelines, and enterprises are moving from pilots to actual playbooks with approvals, brand controls, and clearer norms around rights and disclosure. The next pressure point is already forming: less “generate more video,” more “edit fewer clips with more control,” especially for motion behavior and continuity. The creators who win here are not just producing outputs, they are building direction, taste, and repeatable processes that scale.

Read the full LinkedIn post

A few weeks ago, Adobe + Runway quietly did a “this changes the workflow” kind of collab 🎬

Photo Credit to Adobe

In mid December, Adobe announced a multi year partnership with Runway, and the very first outcome is concrete, Runway’s Gen 4.5 shows up inside Adobe Firefly. The pitch is basically: generate motion where you already start ideas (Firefly), then finish where you already polish (Premiere Pro and After Effects).

Below are the upsides, plus the stuff that might bite later.

What’s great for creators ✅

• Less export gymnastics
  If you live in Creative Cloud, having Runway video generation inside Firefly reduces the “download, re upload, convert, repeat” loop. That’s time back for actual directing and editing.

• Faster concept to cut
  For pitch reels, social variations, animatics, and mood tests, this is a clean path: idea → rough sequence → export → refine.

• Model choice becomes more practical
  Adobe is leaning into “partner models” in Firefly, meaning you can pick a model based on the job instead of forcing every brief through one engine.

• A good excuse to stress test
  Adobe also marketed a short “unlimited generations” window for some plans, which is useful for creatives who want to do a proper bake off without obsessing over credits.

Read the full article here.

Here is a workflow by Giovanni Nakpil

Adobe Firefly + Runway Gen-4.5 Video Generation workflow by | Giovanni Nakpil.

Kling 2.6 Motion Control is now in Freepik

Kling 2.6 Motion Control: Drive your character's movements

Freepik added Kling 2.6 Motion Control to its AI Video Generator, letting you control motion using your own reference video (performance), while swapping the character look with a reference image.

How it works (inputs)

• Motion reference video = the movement, pacing, expression you want

• Appearance reference image = the character/look

• Text prompt = style, background, mood refinements

Key specs

• Length: 3–30 seconds

• Resolution: 720p (standard), 1080p (Pro)

Why creators should care

• Better for sports, martial arts, dance, fast action where prompting motion is usually messy

• Freepik claims improved hand gestures, synced facial expressions, and one-take consistency up to 30s

Quick workflow

Open AI Video Generator → choose Kling 2.6 Motion Control → upload motion video → add character image → prompt → generate.

Read full details here.

AI-generated image

2026 is a big year for AI copyright, and creators will feel it

US courts are weighing training as fair use, while the EU is pushing clearer labels for AI-generated content.

Two rule-making forces are converging in 2026:

US courts are moving toward decisions on whether training genAI on copyrighted work is protected by fair use, or whether companies need to pay and license.

This 2026, the US courts are set up for more rulings after fresh lawsuits and a major 2025 settlement, with billions potentially on the line depending on how fair use is applied to AI training.

So far, courts are split.

• Judge William Alsup described AI training as “transformative” under fair use, but still flagged liability around storing pirated books not tied to training.

• Judge Vince Chhabria ruled for Meta in one case yet warned AI training often may not be fair use, raising concerns about content flooding the market and harming creator incentives.

• Licensing and settlement are already happening alongside the lawsuits (including big entertainment and music examples), which suggests the market may move toward paid access even before courts fully settle the doctrine.

• More hearings are expected in 2026 across disputes involving music, visual art, and model developers, and outcomes could either clarify fair use or extend the uncertainty.

Insight (US)

This is heading toward a world where “Where did it train?” becomes a standard question in creative procurement, similar to “Do we have usage rights for this stock photo?”

Read full details here.

AI-generated image

The EU is formalizing a Code of Practice to help companies comply with the AI Act’s transparency obligations for marking and labelling AI-generated or manipulated content, including deepfakes.

If you ship creative work for clients, publish for an audience, or build a brand, this can affect tool choice, budgets, approvals, and disclosure expectations.

The EU’s Code of Practice is meant to help providers and professional deployers comply with Article 50 transparency obligations around AI-generated or manipulated content.

• AI tool makers should add markings to AI outputs that computers can detect, and that can work across different platforms when possible.

• Deployers (professional use) are expected to label deepfakes and AI generated or manipulated text on matters of public interest, with an exception when there’s human review and editorial responsibility.

• The Commission’s update outlines a timeline: draft now, iteration through early 2026, final target June 2026, with transparency rules becoming applicable August 2, 2026.

Insight (EU)

“Labeling” is becoming less of a personal preference and more of a workflow requirement, especially if you work with EU based brands, platforms, or audiences.

Read full details here.

🎬 Creative Feature

Tests LTX-V2 (Open-Weight AI Video)

László Gaál puts Lightricks’ LTX-V2 through a fast local test and explains why open-weight matters: you can train your own style, use any aspect ratio, and get tighter control with seed locking, FPS, and up to 4K renders. Their setup: local rendering on an RTX Blackwell, ControlNets (depth + canny), and Veo 2–style prompts (LTX-V2 uses a Google Gemma text encoder).

👉 Check it here.

AI Productions x Stereocolor on Hybrid Advertising

Billy Boman AI Productions shares a behind-the-scenes look at their collaboration with Stereocolor AB on Allente Nordic’s new concept, “Allente World of Entertainment.” As the AI Creative Technologist partner, Billy helped build open-source video-to-video workflows to blend real soccer movement with Allente’s brand colors, alongside traditional VFX/CGI (including those dragon scenes). It’s a clean example of hybrid production done right: live action realism, AI-assisted motion integration, and classic post all working together.

👉 Check it here.

💡 Insight

That is all for this week, folks.

Hi Creatives! 👋 

This week is about control and continuity—research that flows straight into drafts with Atlas, brand-safe outputs at scale with Adobe’s Foundry, and node-based canvases that let you branch ideas without losing the thread in Runway Workflows and Figma Weavy. We break down likeness and consistency with Nano Banana, stress-test models in the Image Editing Showdown

This Week’s Highlights:

• ChatGPT Atlas — a browser with ChatGPT built in

• GenAI Image Editing Showdown

• Adobe’s new “AI Foundry” is here — brand-trained models for teams

• 🎧 Nano Banana, in plain speak

• Runway’s new Workflows

• Figma acquires Weavy, launches “Figma Weave” for AI-native creation

• 🎬 Creative Feature

  • Lovis Odin — Next Scene V2

  • AI-assisted circular workflow

Life Update 🎨✨

I’ve joined Krea.ai to help grow partnerships and marketing initiatives around creative AI. If you’re building or experimenting in this space, I’d love to connect and learn what you’re working on. Let’s make something beautiful — and a little bit smarter — together.

Perks: Get 20% off your first 3 months on Krea when you sign up with my link.

ChatGPT Atlas — a browser with ChatGPT built in

Atlas is a new macOS browser from OpenAI with a built-in ChatGPT sidebar, optional “Agent Mode,” and tight privacy controls. It helps you research, summarize, compare products, draft copy, and even complete tasks without leaving the page. Windows, iOS, and Android are coming soon.

What dropped

• Atlas (macOS): Available to Free, Plus, Pro, Go. Business beta live; Enterprise/Edu via admin. Windows + mobile coming soon.

• Sidebar: Summarize pages, grab quotes, outline, draft emails, parse tables.

• Agent Mode: For select paid tiers, handles multi-step tasks like research and shopping.

• Privacy: Browsing data isn’t used for training by default; optional “browser memories.”

• Release notes: Early fixes and improvements are already rolling out.

Why this matters

• Faster research for briefs, treatments, and pitch decks.

• Easy comparisons for gear and tools without tab overload.

• Draft as you browse: captions, outlines, shot lists.

• Offload repetitive work with Agent Mode.

• More control over what gets remembered.

👉 Read full details here: OpenAI

GenAI Image Editing Showdown

Picking an image-editing model for real work? This quick feature lines up multiple tools on the same prompts so you can spot strengths and weak spots fast. Check this site to see the side-by-side comparison of different models.

GenAI Image Showdown

A comparison of various SOTA generative image models on specific prompts and challenges with a strong emphasis placed on adherence.

genai-showdown.specr.net/image-editing

Adobe’s new “AI Foundry” is here — brand-trained models for teams

Image Credits:Adobe

Adobe just introduced AI Foundry, a service that helps companies build custom generative models trained on their own brand assets and IP. These models can output text, images, video, and even 3D scenes, and the service uses usage-based pricing so teams pay for what they actually run.

Why it matters for creatives 💡

• On-brand at scale. Train once on style guides, product shots, palettes, and tone, then spin seasonal, multilingual, and format variants without starting from zero.

• Cleaner rights posture. Firefly is trained on licensed data, and Foundry fine-tunes with your IP to help reduce copyright headaches.

• Faster cycles. Great for always-on social, promo cut-downs, and product page refreshes where consistency beats one-off hero shots.

Try this next 🧰

1. Assemble a “training kit.” Logo packs, typography, color tokens, product angles, VO examples, and proof of rights in one place.

2. Write guardrails. Prompt do’s and don’ts plus a short house style for copy and visuals.

3. Pilot one campaign. Narrow goal, clear metrics, weekly review of outputs.

4. Watch consumption. Set caps and track generations since pricing is usage-based.

Who’s already using it: Early adopters include Home Depot and Walt Disney Imagineering; reports also point to Red Sea Global evaluating Adobe’s enterprise AI stack

Note: AI Foundry is enterprise-only. You access it via Adobe’s business page and a “contact sales/request info” flow; it’s positioned for organizations with Admin Console–managed licenses, not individual creators.

Read full details here.

🎧 Nano Banana, in plain speak

The team spills that Nano Banana (Gemini 2.5 Flash Image) is built to cut the boring stuff so creators can spend way more time actually creating. Think conversational edits with the image quality you want, plus likeness and consistency that hold up for real work.

• Less tedious, more creative: Removes the grind so ideas stay front and center.

• Origin: Imagine’s visual fidelity meets Gemini’s interactive editing.

• Zero-shot likeness: One reference photo can look like you—no fine-tuning.

• Character consistency: Same subject across frames for storyboards and brand work.

• Speed as a feature: Low latency encourages more tries and better results.

• Raise the floor: Focus on improving the worst outputs, not cherry-picks.

• Right UI for the job: Chat/voice for quick tweaks, workflow graphs when you need control.

• Visual reasoning: Strong at diagrams/explainers for teaching and pitching.

• Enterprise path: Long brand guides in context for tighter compliance over time.

• Sleeper feature: Multi-image “interleaved” stories for consistent sequences.

Watch the podcast here.

Runway’s new Workflows

Runway’s new Workflows is a big, node-based canvas where you chain text→image, image→video, and video→video steps. You’re not limited to Runway models — you can mix in third-party models — then tweak shots, swap backgrounds, and keep motion and dialogue intact across steps.

Why creatives should care

• Build repeatable pipelines for brand-safe looks

• Branch variations fast for A/B shots and social cuts

• Save templates so your team stays consistent across scenes

• Mix LLM nodes to auto-polish prompts between steps

Quick start

1. Go to Workflows then New Workflow

2. Add nodes: Text → Gen-4 Image → Gen-4 Image→Video

3. Connect outputs to inputs and click Run to test a node or Run All for the chain

4. Save as template and branch prompts for versions

Heads-up: You’ll need a paid Runway account to use Workflows.

Read more here.

Figma acquires Weavy, launches “Figma Weave” for AI-native creation

Figma

Figma just bought Weavy and is rolling it into Figma Weave, a model-agnostic, node-based workspace for generating and editing images, video, animation, motion design, and VFX right on a browser canvas. Think pro controls plus AI outputs you can branch, remix, and refine. 🎬✨

What’s new

• Model-agnostic: Pick the right model per task. Examples mentioned include Seedance, Sora, Veo for cinematic video, Flux and Ideogram for realism, and Nano-Banana or Seedream for precision.

• Node-based workflow: Chain steps, branch ideas, and keep versions you can tweak without breaking the flow.

• Built-in editing: Masking, lighting tweaks, color grading, and more on the same canvas.

• Community and scale: Already used by indie creators, startups, and Fortune 100 teams. Figma is growing the Weave team in Tel Aviv to accelerate the roadmap.

• All inside Figma: Expands the platform from design files to a broader media pipeline spanning concept to final polish.

Early use cases to try

• Pitch videos and mood films: Block scenes with AI shots, color grade, and export faster for stakeholder reviews.

• Brand kits to content: Feed brand palettes and references, generate social banners, then fine-tune typography and grading.

• Product mocks and motion tests: Swap models for realism or stylization, keep both branches for side-by-side review.

• VFX previz: Rough in shots with Seedance or Sora, then refine elements and passes without leaving the canvas.

Read more and the official announcement here.

🎬 Creative Feature

Lovis Odin — Next Scene V2

A purpose-built LoRA for Qwen Image Edit 2509 that generates seamless “next shots” while keeping characters, lighting, and mood consistent. Trained on thousands of paired cinematic shots for smoother, more emotional transitions.

• Stronger shot-to-shot consistency

• Better lighting and character preservation

• Cleaner framing with no black bar artifacts

Use it in ComfyUI or any diffusers pipeline. Type “Next Scene:” and describe what happens next.
Open-source and built for filmmakers, animators, and storytellers.

Try it: Hugging Face workflow + fal thread

👉 Check it here.

AI-assisted circular workflow

How AI-driven workflow changed the Toyota ad

László has shared a longer Director’s Cut of his Toyota spot, expanding on an AI-assisted circular workflow where ideas are generated, tested, approved, and refined in quick loops. The film traces “mobility” from life’s first movement to horse-powered travel to modern cars, showing how freedom of movement connects people and places. Created with Toyota and collaborators at Google, including Randy Han and Mitsuhito Hirose.

👉 Check it here.

Chroma Awards

Heads up: Chroma Awards deadline is now Nov 17, 11:45 PM PST. I’m on the finals jury and would love to see your AI film, music video, or game in the mix. $175,000+ in prizes and strong exposure to studios and brands.

Enter at chromaawards.com and share with a friend who needs the nudge.

💡 Insight

This week’s thread is “control without chaos.” Use ChatGPT Atlas when you need research to flow straight into first drafts and moodboards with less context-switching, but keep a human eye on nuance and citations; treat the GenAI Image Editing Showdown as your fast picker for who’s best at hair edges, product cleanup, text, and multi-pass retouch so the team isn’t guessing; reach for Adobe’s AI Foundry when you need on-brand, governed outputs at SCALE lean on Nano Banana to keep faces and product forms consistent across variations while watching for subtle drift; build repeatable spots in Runway Workflows with node-based templates you can rerun for each campaign; and explore Figma Weave to keep AI image/video iterations, prompts, and feedback in the same file so handoffs shrink and creative history stays reviewable.

That is all for this week, folks.

Hi Creatives! 👋 

This week is all about control, sound, and speed. Veo 3.1 lands inside Flow by Google with richer native audio, stronger prompt adherence, and inline edits that feel like real post. OpenAI outlines a chips plan and long-range build that points to shorter queues and steadier fidelity. Microsoft debuts MAI-Image-1 for photoreal images that slot neatly into Copilot workflows. And Google rolls Nano Banana into Search via Lens and NotebookLM, with Photos following soon. Net result for your pipeline: cleaner first passes, fewer handoffs, and more time to shape the story instead of wrestling the tools

This Week’s Highlights:

• Veo 3.1 lands in Flow: smarter controls, richer audio, more cinematic results

• Big Build for Bigger Creativity: OpenAI’s new chips plan and a five-year roadmap

• Microsoft’s first in-house image model just dropped: MAI-Image-1

• Google’s “Nano Banana” lands in Search

• 🎬 Creative Feature

  • Veo 3.1 inside Flow by Google.

  • Playful mini-story

  • AI integrates into a rebranded product.

  • Kling delivers consistently.

Veo 3.1 lands in Flow: smarter controls, richer audio, more cinematic results

Google just rolled out Veo 3.1 and a big refresh to Flow, its AI-powered filmmaking workspace. If you’ve been experimenting with AI video for social, ads, or pitch films, this update levels up control, sound, and editability so your clips feel less “AI” and more “production-ready.”

What’s new

• Rich, generated audio across Flow features, so clips don’t feel silent or temp. Works with Ingredients to Video, Frames to Video, and Extend.

• Stronger narrative control and realism in Veo 3.1, with better prompt adherence and image-to-video quality.

• Inline editing in Flow: Insert new elements with proper shadows and lighting; Remove is coming soon for clean object removal.

• Where to use it: Flow, Gemini app, Gemini API, and Vertex AI.

👉 Read full details here.

🚀 Big Build for Bigger Creativity: OpenAI’s new chips plan and a five-year roadmap

📷️OpenAI

OpenAI x Broadcom: 10 GW of custom AI accelerators

OpenAI announced a multi-year collaboration with Broadcom to design and deploy OpenAI-designed AI accelerators and Ethernet networking for next-generation clusters. The rollout targets first racks in the second half of 2026 and completion by the end of 2029. OpenAI says this helps move model know-how directly into silicon, and notes it now serves over 800 million weekly active users.

Why this matters to creatives

• Faster queues and renders: more compute means shorter waits for image, video, and audio generations at higher fidelity.

• Bigger canvases: expect longer video durations, richer motion, and more consistent subjects as training and inference capacity scales.

• Better control: purpose-built accelerators can align hardware with model behavior, which often shows up as steadier styles, tighter motion control, and more reliable edits.

Read full details here.

💼 The bigger picture: a five-year plan for massive spend

Two days later, OpenAI is crafting a five-year plan to support more than one trillion dollars in pledged spending, exploring new revenue lines, debt partnerships, and additional fundraising to keep scaling infrastructure and products.

Why this matters to creatives

• More tools in one place: sustained investment typically brings new features into familiar apps, so you can ideate, edit, and publish without bouncing across tools.

• Reliability as you scale: stable funding plus long-range planning reduces outages and throttling during peak launches and client deadlines.

• Wider access: diversified revenue and financing can support geographies beyond the usual hubs, which means better performance wherever your team is.

Read full details here.

Microsoft’s first in-house image model just dropped: MAI-Image-1

🖼️MAI-Image1

🖼️MAI-Image1

Microsoft AI unveiled MAI-Image-1, its first text-to-image model built entirely in-house. It focuses on photorealistic lighting, landscapes, and speed, and it’s already sitting in the Top 10 on LMArena. This signals Microsoft’s steady move to make more of its own creative models that can plug into Copilot and Windows workflows.

What happened

• Microsoft announced MAI-Image-1 on October 13, 2025. It’s positioned as an image generator trained with feedback from creative pros to avoid “repetitive or generically stylized outputs.” Microsoft also calls out faster turnaround than many larger models.

• Early signal of quality: MAI-Image-1 entered LMArena’s text-to-image Top 10 shortly after launch.

• Context: This follows Microsoft’s August reveal of its first two in-house models, MAI-Voice-1 and MAI-1-preview, and its broader strategy of mixing models from multiple providers inside Microsoft 365.

👉 Read full details here

🍌 Google’s Nano Banana is now in Search — with NotebookLM today and Photos soon

Google is rolling Nano Banana into Search via Lens with a new Create mode, and it’s already working under the hood in NotebookLM to power Video Overviews with new styles like watercolor and anime. Google also says Photos integration is “coming in the weeks ahead.” They note 5B+ images generated since launch in August. Initial rollout is US and India in English.

How to try it: Open the Google app, tap Lens, then Create to transform or generate images without leaving Search.

🎬 Creative Feature

Veo 3.1 inside Flow by Google.

Made with Veo 3.1

Filmmaker Dave Clark just shared his first tests with Veo 3.1 inside Flow by Google. His team at Promise has early API access and is already integrating it into their MUSE platform, led by Mariana Acuña Acosta.

• Workflow: mix of image and text-to-video

• Noticed: more dynamic motion; the new Expand Prompt feature is especially useful

👉 Watch the tests.

Playful mini-story

Just more Sora2 tests.

Meet Mauricio Tonon. His latest Sora 2 test is a playful mini-story that shows you can keep tight control and still have fun. He wrote the characters’ lines, sketched the sound design, and shaped a loose, coherent narrative — a great reminder that learning-by-making counts.

Watch the three trailers here.

AI integrates into a rebranded product.

Martin Haerlin uses AI to integrate a rebranded product directly into existing shots, so teams can repurpose assets without a reshoot. It’s especially smooth when the original product is already in frame, since tracking and lighting cues are there.

Check the full post here.

Kling delivers consistently.

In just 10 minutes, creator Tianyu Xu turned 2 prompts into 13 crisp video clips using Kling 2.1 with start and end frames in a Weavy workflow. His takeaway for designers and architects is simple: you don’t always need the newest model. You need the one that performs for your use case. Sora 2 and Veo 3 are exciting, but for fast, reliable image-to-video, Kling delivers consistently.

What you’ll learn next

• LoRA for Architects workshop: Ar. June Chow on practical LoRA training for your preferred styles

• Tianyu’s video best practices to speed up your pipeline

👉 Watch the original post and details here.

💡 Insight

Treat AI output like production plates you can trust. Start your first cut in Flow with Veo 3.1 so mood and pacing are baked in from the audio. Use inline edits to add or remove elements instead of regenerating whole shots. When you need quick stills with believable lighting, spin up MAI-Image-1 and comp as you would any plate. Keep an eye on OpenAI’s buildout because more capacity typically means longer durations, steadier motion, and fewer throttles on deadline days. For fast mockups and concept frames on the go, Nano Banana inside Lens and NotebookLM is a handy pocket tool. Small proofs, quick passes, cleaner finals — that’s the move this week.

Hi Creatives! 👋 

This week is about native audio, tighter control, and quicker handoffs. Sora 2 levels up motion realism and sound, plus a social-style app for fast remixing. ChatGPT Apps let you build inside the chat instead of tab-hopping. Runway’s case study shows AI nesting cleanly in a traditional post pipeline. Meta’s AI signals are shifting how work gets discovered. In short, better tools, shorter loops, clearer results.

This Week’s Highlights:

• 🎬 Sora 2 is live — and there’s a new social-style Sora app

• OpenAI DevDay: ChatGPT Apps + Apps SDK (built on MCP) 🚀

• How Eggplant Picture & Sound finished Life After People with Runway 🎬

• Meta will use your AI chats to shape your feed and ads

• 🎬 Creative Feature

  • GOLF le FLEUR Maritime ⛴️

  • Generating three HD trailers.

  • Gucci’s THE TIGER

Chroma Awards — the global stage for AI-powered creativity

I’m proud to announce I’m judging the Chroma Awards finals for Short Form, Social, and Film — the global stage for AI-powered creativity across film, music videos, and games.

• 💰 $175,000 in cash prizes

• 🤝 Backed by ElevenLabs, Freepik, fal, Google Cloud, Dreamina, and CapCut

• 🚀 Open to AI creators pushing what’s possible

🎬 Sora 2 is live — and there’s a new social-style Sora app

OpenAI rolled out Sora 2, a big step up in text-to-video, and paired it with a TikTok-like iOS app where you can make 10-second clips, remix others, and even add your own face and voice with a one-time “cameo” setup. Think short, swipeable, AI-generated videos built for play, prototyping, and quick storytelling.

✨ What’s new in Sora 2

• More realistic motion and physics. Fewer weird warps on action shots.

• Native audio. Synchronized dialogue, sound effects, and ambient tracks generated with the video.

• Better control. Clearer adherence to prompts and continuity across shots.

What’s inside the Sora app

• Vertical feed with like, comment, and remix. All videos are AI-generated. Clips are up to 10 seconds.

• Cameos. Verify your likeness once, then drop yourself or friends into scenes. You’ll be notified when your cameo is used.

• Early rollout on iOS in the U.S. and Canada. Invite-based access for now. Free to start.

Who’s already tapping into it

All Pro and Max users in Krea will have unlimited access to Sora 2.

• Krea  added “Sora 2” as a selectable model in its Video tool; notes a promo for Pro/Max users.

• Picsart  blog post walks through using “Sora 2” and “Sora 2 Pro” inside the Picsart AI Video Generator on web and mobile.

• Artlist  “Sora 2 lands on Artlist” with a model dropdown for Sora 2 / Sora 2 Pro in its Image & Video Generator; included on certain plans.

• Artificial Studio says Sora 2 and Sora 2 Pro are available in-app, plus an API for embedding.

• Mattel is piloting Sora 2 so designers can turn toy sketches into shareable video concepts faster. This is a real signal that big brands see Sora as a rapid pre-viz tool.

• Freepik: Sora 2 is listed on pricing with “1,600 credits per 4s” for paid tiers.

Creative caution checklist

• Copyright. If your concept touches known characters, franchises, or artist styles, do a rights pass first and document approvals. The opt-out system means owners may need to take action, and policies could evolve.

• Brand safety. Early feeds on new platforms can surface messy content before moderation catches up. Keep a human review layer on anything public-facing.

• Territory and access. The iOS app rollout is staged. Plan for mixed team access and keep alt workflows ready.

Read full details here.

OpenAI DevDay: ChatGPT Apps + Apps SDK (built on MCP) 🚀

🖼️OpenAI

ChatGPT is now an app platform. You can call apps like Canva right in the chat, and developers can build their own using the new Apps SDK, which sits on the Model Context Protocol (MCP).

• Apps run inside ChatGPT with interactive UIs. Early partners available today include Canva, Spotify, Zillow, Booking.com, Coursera, Expedia, Figma. Available to logged-in users outside the EU on Free, Go, Plus, and Pro.

• Apps SDK is in preview and is built on MCP, the open spec for connecting LLMs to tools and data. OpenAI says the SDK is open source so apps can run anywhere that adopts the standard. App submissions and monetization come later this year via an Agentic Commerce Protocol.

• Live demo used Canva to generate designs directly in chat.

Why it matters for creatives 🎨

• Fewer context switches. Brainstorm in chat, then have Canva turn an outline into slides or posters without leaving the thread.

• New distribution. Apps surface contextually inside conversations, giving builders access to a very large ChatGPT audience.

• Commerce is coming. Devs will be able to charge for in-chat experiences once monetization opens.

MCP in one minute 🔧

• What it is: An open spec that lets a model call your backend “tools” and optionally render your UI inline.

• How Apps SDK uses it: Your MCP server lists tools, handles call_tool requests, and can return structured data plus a component reference that ChatGPT renders in an iframe. Transport is SSE or streaming HTTP.

• Official SDKs: Python and TypeScript are provided, with guides to design components, auth, and state.

Build notes for your devs 🛠️

• Spin up an MCP server in Python or TypeScript and define tools with JSON Schemas.

• Return a UI component with your tool response so ChatGPT renders it inline.

• Test locally, expose with ngrok, enable Developer Mode, then add your app in Settings → Connectors. Sample repos are provided.

Test locally

Read full details here: OpenAI’s announcement and docs. OpenAI Developers

How Eggplant Picture & Sound finished Life After People with Runway 🎬

Eggplant Picture & Sound is a Toronto-based full-service post-production studio offering picture, sound, motion graphics, animation, and VFX. They helped bring an 8-episode History Channel series back from the shelf using Runway’s Gen-4, Aleph, and reference tools. The team cut weeks of VFX time, kept quality high, and slotted AI into a traditional pipeline without blowing budget or schedule.

Concrete wins

• 1 month of post-production time saved overall.

• 4 hours of compositing saved on a single “roof damage” shot by asking Aleph for “more damage on the roof, please.” It went straight into the final episode.

• Project unshelved: AI made a previously unaffordable vision achievable

The workflow, simply

• Runway + classic VFX: They combined Runway generations with matte painting, Nuke, and compositing for polish. Example: adding practical trees into a hurricane scene.

• Prompt craft matters: wording and tense changed results. “The mall is closed” removed people because the model inferred that closed malls are empty.

• Why Runway: quality plus enterprise/legal support that fits their pipeline.

🔗 Read full details here.

Meta will use your AI chats to shape your feed and ads

Illustration by Nick Barclay / The Verge

Meta just confirmed it will start using what you talk about with its AI assistant to personalize what you see across Facebook and Instagram, starting December 16.

Key points 🔎

• Affects suggested posts, Reels, groups, and ads.

• Cross-app effects if your accounts are linked in Accounts Center.

• No specific opt-out for this new signal, though you can still tweak general ad settings.

• Not rolling out in the UK, EU, or South Korea.

• Sensitive topics like religion, politics, health, sexual orientation, ethnicity, philosophical beliefs, and union membership are excluded.

Why it matters for creatives 🎨

• Treat chat-style questions as keywords. Write hooks and captions like real queries your audience would ask.

• Keep brand voice and visuals consistent across Facebook and Instagram to ride cross-app momentum.

• Run quick A/Bs that answer a question vs inspire a mood, then watch saves and CTR.

Read full details here.
Official update from Meta.

🎬 Creative Feature

GOLF le FLEUR Maritime ⛴️

Arantxa Barcia and the MITO Studio team’s newest spec: a playful maritime sequel to their fan-favorite “Golf le Fleur Express.” This version uses real GOLF le FLEUR* products and branding, plus a tiny baby capsule — and even tests timing and continuity with a Tyler, The Creator lipsync.

Why it’s cool

• Shows how much generative video quality jumped this year

• Smart brand integration with authentic assets

• Crisp lipsync and pacing that feel production-ready

Tools ⚙️ Freepik, Higgsfield AI, Kling AI, Runway, Google DeepMind, WAN 2.1 for lipsync

Watch the clip and meet the team in the links.

Generating three HD trailers.

OpenAI Sora 2 API: Creating HD Trailers with a Single Prompt | Chad Nelson

Filmmaker Chad Nelson put Sora 2 Pro mode through its paces by generating three HD trailer concepts from single prompts. Sora handled the script, shot design, edit, music, and dialogue in one pass. He produced four variations for each, then did light swaps on title cards or a shot to lock a final cut. No color grade or mix.

• Alice’s Wonderland — a modern, darker London-set take on Alice, blending wonder with horror and ending on a towering spider Queen.

• Fathoms Below — a 1960s slapstick musical vibe with kitschy submarine miniatures, octopus mayhem, and a catchy theme.

• The Cosmonaut — a tense, muted historical drama around Yuri Gagarin’s Vostok 1 training, launch, and white-knuckle re-entry.

Watch the three trailers here.

Gucci’s THE TIGER

Shared by Cartel & Co. Spotlighting AI artist Sam Finn for his work on THE TIGER, a short by Spike Jonze and Halina Reijn presented by Gucci. Starring Demi Moore, Edward Norton, Keke Palmer, Alia Shawkat, Elliot Page, and Ed Harris, the film blends cinematic craft with seamless AI-driven visuals led by Sam and collaborators.

Key credits: DoP Jasper Wolf. VFX Supervisor Janelle Croshaw Ralla. Color by Harbor Picture Company with Senior Colorist Damien Vandercuyssen. Produced by MJZ. Edited at Final Cut.
AI: Sam Finn @cartelandco with conartistproductions, electric_dreama, feremony, and partners Perlin, SSVFX, TEAM VFX ARTIST PVT LTD, Suplex FX.

Watch the post and full credits.

💡 Insight

Production is sliding from experiment to execution. Use Sora 2 for rapid pre-viz with audio baked in, then finish where it counts. Keep your team close to the work by testing Apps inside ChatGPT and handing off assets smoothly to post. Treat AI outputs like plates you’ll comp and grade. Watch distribution shifts as Meta learns from AI chats and tune captions like keyworded questions. Small proofs, fast iterations, cleaner finals.

That’s a wrap for this week!

Hi Creatives! 👋 

This week’s AI news feels like a mash-up of Hollywood, fashion, and cutting-edge tools. Adobe Firefly just leveled up with Ray3 for AI video, the world’s biggest studios are clashing with MiniMax, and OpenAI is teaming with US and UK institutes to stress-test agent security. Add in Nick Knight’s take on AI as the next camera, Comfy Cloud making workflows setup-free, and some sleek creative experiments in character consistency and stylized video — and you’ve got a snapshot of just how wide AI is spreading into our worlds.

This Week’s Highlights:

• Ray3 Lands in Adobe Firefly — The Next Big Step for AI Video Creation

• 🎥 Disney, Warner Bros & Universal vs. MiniMax

• OpenAI + US CAISI and UK AISI 🔐

• 📷 Nick Knight on AI and the Future of Photography

• ☁️ Comfy Cloud: Generative AI Made Simple

• 🎬 Creative Feature

  • Workflow for AI character consistency

  • Stylized AI video worlds

Ray3 Lands in Adobe Firefly — The Next Big Step for AI Video Creation

An image from a Ray3 prompt. Luma AI is releasing the new model hoping it becomes a creator staple. Luma AI

Luma AI’s new Ray3 model is now integrated into Adobe Firefly. It generates short video clips (up to 10s) with smoother motion, better frame consistency, and native HDR support — meaning higher-quality visuals straight out of the tool.

What you should know

• Available inside Firefly for Adobe users, with unlimited generations until Oct 1

• Exports in pro formats like ACES2065-1 EXR at 10, 12, and 16-bit depth

• Draft mode lets you quickly test ideas and iterate faster

• Fits directly into the Adobe ecosystem (Boards, Creative Cloud)

What it means
For creatives, this makes Firefly more than an image tool — it’s evolving into a hub where you can storyboard, generate test footage, and even produce polished assets with AI. Ray3 brings professional-level output into the short-form space, raising the bar for reels, promos, and concept work.

👉 Read full details here

🎥 Disney, Warner Bros & Universal vs. MiniMax

The minions in 'Despicable Me 4.' Illumination & Universal Pictures

Hollywood’s biggest players—Disney, Warner Bros, and Universal—are suing Chinese AI company MiniMax. Why? Because MiniMax’s tool, Hailuo AI, used famous characters like Darth Vader, Minions, and Wonder Woman in its promotions without permission .

Why this matters for you 👩‍🎨

• Protecting creativity: These lawsuits are about making sure studios (and by extension, all creators) have control over how their work is used.

• Setting the rules: Cases like this will help decide what AI platforms can or can’t do with copyrighted material.

• Global impact: MiniMax is huge, with 157M users worldwide. What happens here could shape how AI and creative industries work everywhere.

👉 Read more here on Reuters

OpenAI + US CAISI and UK AISI 🔐

Quick explainer

• US CAISI is the Center for AI Standards and Innovation at NIST. It works with industry to test AI systems and shape standards.

• UK AISI is the UK AI Security Institute inside the Department for Science, Innovation and Technology. It evaluates risks from advanced AI and shares guidance.

The gist
OpenAI posted a fresh update on September 12 about working with the US Center for AI Standards and Innovation and the UK AI Security Institute to pressure-test agent security and bio-misuse safeguards.

Highlights

• CAISI red-teamed ChatGPT Agent, found two new vulnerabilities that could have allowed remote control and site impersonation, then OpenAI patched within one business day. The exploit chain worked about half the time before fixes.

• UK AISI stress-tested biosecurity safeguards across ChatGPT Agent and GPT-5, filing more than a dozen detailed reports that led to monitoring and config improvements.

Do this next ✅

• Run agents in a sandbox with least-privilege access when they touch client files or cloud drives.

• Threat-model prompt injection in your workflow and test multiple attempts before rollout.

Insights 💡

Read full details here: OpenAI’s update

📷 Nick Knight on AI and the Future of Photography

Instagram post

For over four decades, Nick Knight has challenged how we see images—moving fluidly from photography to film, virtual realities, and now, AI. In a recent talk, he compared today’s debate around generative AI to the skepticism photography once faced when it was new. Painters dismissed photography as “mechanical,” yet over time it proved its power to capture soul and emotion. Knight suggests AI is following a similar path.

🔑 His Perspectives at a Glance

• History repeating: Photography was once dismissed, now AI gets the same critique.

• It’s still art: What matters is the artist’s choice, curation, and vision—not the tool itself.

• Borrowing isn’t new: Artists have always drawn inspiration from the past; AI simply accelerates this.

• Optimism for the future: Knight believes this is one of the most exciting times for creativity.

✨ Insight for Creatives: Like the camera once did, AI offers a new lens on how we can create, remix, and reimagine art. The challenge—and opportunity—lies in how we choose to use it.

🎥 Watch his full reflection here

☁️ Comfy Cloud: Generative AI Made Simple

Comfy Cloud

Join the private beta

blog.comfy.org/p/comfy-cloud?utm_campaign=post&utm_medium=web×tamp=4.4

If you’ve ever wrestled with installing Python dependencies, hunting for the right GPU, or crossing your fingers during an update… you’ll love this. Comfy has just launched Comfy Cloud — a streamlined way to use ComfyUI without the tech.

✨ What it means for you:

• Skip the setup: No more complicated installs, everything just works.

• Powerful GPUs: Run workflows faster and smoother.

• Creative tools on tap: Access popular models plus an ever-growing library of custom nodes.

• Community-first: They’re even exploring revenue sharing for developers who create custom nodes.

Right now, Comfy Cloud is in private beta, offering free access for those who sign up and share feedback. Long-term, they’ll move to a subscription model based on GPU usage, but the local open-source ComfyUI remains free.

For creatives, this means more time making and less time troubleshooting. Imagine testing ideas quickly, collaborating easily, and not worrying about whether your setup will crash. 🚀

👉 Read full details here

🎬 Creative Feature

Workflow for AI character consistency

Sebastien is breaking down a hidden workflow for AI character consistency that blends Seedream 4, Kling 2.1, and Freepik into a powerful pipeline. From swapping backgrounds and outfits to turning still edits into moving video, he shows how a full photoshoot concept can now be prototyped in under an hour.

👉 See his full breakdown here

Stylized AI video worlds

How AI can recreate products in stylized worlds

James’ latest experiment explores how products can live naturally inside stylized AI video worlds. In this Tide spec ad, the bottle isn’t just placed in — it’s fully reimagined in felt to match the scene, while staying instantly recognizable. The result shows how AI video models like Veo 3 can bring both characters and products into new creative spaces without losing brand identity.

👉 See the full post here

💡 Industry Insight

What jumps out this week is how AI is blending industry, creativity, and accountability. On one side, we see heavyweights like Adobe pushing video forward with Ray3, and Nick Knight reframing AI as just another tool in the artist’s kit. On the other, Hollywood is pushing back hard against MiniMax, showing how copyright battles are shaping the rules in real time. And in the middle, OpenAI’s work with CAISI and AISI reminds us that governments and labs are stress-testing the very systems we’ll soon use every day.

For creatives, the signal is clear: the projects we’re experimenting with now — from AI-driven video edits to stylized brand worlds — aren’t just tests. They’re the same skills that will matter when the industry lines between film, fashion, and AI finally blur.

That’s a wrap for this week!

Hi Creatives! 👋 

This week’s AI updates show just how fast the field is branching out. Runway, the tool we know for videos, is now training robots. OpenAI is backing a full-length AI feature film, China’s giants are flexing trillion-parameter models, and Warner Bros. has officially sued Midjourney over AI-generated images of its iconic characters. ⚡

From robotics labs to Hollywood courtrooms, the reach of creative AI keeps expanding, and for us, that means the skills we’re building today may shape far more than just content.

This Week’s Highlights:

• Runway’s not just generating videos anymore—it’s eyeing robotics.” 👀🤖

• 🎥 OpenAI Backs an AI-Made Animated Film

• 🚀 China’s Giants Drop Trillion-Parameter Models

• ⚡️Warner Bros. sues Midjourney for AI images

• 🎬 Creative Feature

  • The Chroma Awards

  • Runway’s 48-Hour AI Film

  • AI Music Video Experiment

Runway’s not just generating videos anymore—it’s eyeing robotics.” 🤖

AI-generated Image

Runway, best known for its creative AI video tools, is now stepping into the robotics and self-driving space. Why? Their powerful world models, originally built for content creation, turn out to be perfect for training robots in simulated environments.

What this means for us creatives:

• Tools we love are scaling up: the same tech behind stunning videos is teaching machines how to move.

• Crossovers are real: artistry and engineering are blending—your creative workflows may fuel real-world impact.

• Big growth ahead: Runway has raised over $500M and is building a team dedicated to robotics.

👉 Read the full story on TechCrunch

❓ Will Runway stop making image & video tools?

NOPE! Runway isn’t ditching its creative roots. Image and video generation remains their core focus for creators.

What’s new is that they’ve discovered their world models are also perfect for training robots and self-driving systems. Instead of replacing creative tools, they’re expanding—serving both filmmakers and robotics companies with the same underlying tech.

Think of it as Runway doubling its stage: 🎬 for creators, 🤖 for robotics.

🎥 OpenAI Backs an AI-Made Animated Film

Photo: Critterz Film Ltd.

The Wall Street Journal reports that OpenAI is backing Critterz, an animated feature film created with heavy use of AI. Directed by Chad Nelson (OpenAI creative specialist) and co-developed with Nik Kleverov of Native Foreign, the project is being produced by Vertigo Films (UK) and Native Foreign (LA), with a script from James Lamont & Jon Foster (Paddington in Peru). Executive producers Allan Niblo, James Richardson, and Jane Moore are leading the production, supported by Federation Studios (Paris). The film is being created in just nine months on a budget of under $30 million

•  Speed & affordability → A full-length film produced in months, not years, with a fraction of the budget.

•  Collaboration over replacement → Human sketches and voices combined with AI enhancements keep creativity and copyright intactIndustry ripple effect → Hollywood is watching closely, expect more debates on ethics, jobs, and the future of storytellingOpportunities for independents → Lower costs and faster timelines could empower smaller studios and individual creators.

👉 Read the full article here

🚀 China’s Giants Drop Trillion-Parameter Models

Two of China’s biggest players just made bold moves in AI.

Twitter tweet

Alibaba unveiled Qwen-3-Max-Preview, a trillion-parameter model designed to rival OpenAI and DeepMind. It’s built for multilingual nuance, long-form storytelling, and complex prompts—and available through Alibaba Cloud and OpenRouter. Backed by a $52B investment in AI infrastructure, this isn’t just a flex, it’s a commitment to building serious creative tools.

👉 You can try Qwen yourself here.

Meanwhile, Moonshot AI released Kimi-K2-Instruct-0905, another trillion-parameter powerhouse. It uses a Mixture-of-Experts architecture with a massive 256K token context window, making it perfect for sprawling projects. It’s also strong at agentic coding, front-end generation, and tool-calling—helpful if you’re blending code with creative work. Access is live through Groq and OpenRouter.

Try it here.

Why Creatives Should Care 🎨

•  Deeper comprehension → great for layered stories or multilingual projects

•  Long-form ready → holds context across big scripts or novels

•  Coding + creative blend → helps design, prototype, or integrate tools

•  On-demand access → pay-as-you-go, easier to experiment

👉 Read more on Alibaba here
👉 Explore Kimi-K2-Instruct on Hugging Face

⚡️Warner Bros. sues Midjourney for AI images

Image Credits:Victor J. Blue/Bloomberg / Getty Images

Warner Bros. Discovery has just filed a lawsuit against Midjourney, claiming the AI image platform is generating unauthorized versions of Superman, Batman, Bugs Bunny, and more. This case could be a turning point for how AI platforms operate—and how creatives like us use them.

🔑 Key Highlights

• The Lawsuit: Filed in LA federal court on Sept. 4, 2025, alleging Midjourney trained on Warner Bros. content without permission.

• Character Use: Users can prompt “classic superhero battle” and still get DC heroes in high detail.

• High Stakes: Warner Bros. is seeking up to $150,000 per infringing work plus an injunction to stop further use.

• Bigger Picture: This follows similar lawsuits from Disney and Universal earlier this year—meaning the pressure is mounting across Hollywood.

For creatives, this isn’t just another legal headline—it’s a signal of where the industry is headed. If the studios succeed, we can expect AI tools to tighten their filters, restrict character generation, and introduce more guardrails. That means fewer shortcuts, but also a push for more original, distinctive work from artists who embrace AI as a creative partner rather than a cloning machine.

Read the full article here.

When AI Product Placement Gets Tricky 🎥

NanoBanana nails single-product shots—but struggles when more items are added, especially products with complex text labels.

Instagram post

Check it here.

🎬 Creative Feature

The Chroma Awards

The Chroma Awards is the first global competition celebrating AI-powered creativity across Film, Music Videos, and Games. Season 1 is now live, bringing together creators worldwide with over $150,000 in cash prizes and $1M in AI tool credits.

What’s Happening Now

Submissions are officially open until November 2, 2025. Winners will be announced in a livestream ceremony on December 7, 2025.

How to Join

• Create an original project (after Feb 1, 2025).

• Post it publicly on YouTube, TikTok, or Vimeo.

• Submit your entry through the Chroma Awards portal.

💡 If you’ve been experimenting with AI in your films, music videos, or game concepts—this is your moment to step forward, share your work, and get recognized.

👉 Don’t just watch from the sidelines. Submit your project and join the movement at the Chroma Awards.

Runway’s 48-Hour AI Film

Ron Baranov recently won ( CONGRATULATIONS! ) Runway’s 48-hour film challenge, creating an entire short film from scratch in just two days. Unlike traditional filmmaking teams of dozens, Ron leaned on AI tools Runway, Midjourney, LTX Studio, ElevenLabs, and Epidemic Sound to bring his story, characters, and multiple locations to life.

We especially liked what he said about the importance of storytelling:
“The fundamentals of storytelling matter most. Tools will keep shifting, techniques will keep evolving. But script, logic, and visual storytelling are timeless. If you know how to tell a story, everything else can be learned.”

👉 Watch Ron’s winning film here

AI Music Video Experiment

🍌 NanoBanana strikes again, this time, a full AI music video experiment!

Sara Nemati recently shared her latest experiment — a full AI-generated music video. Built with first-frame/last-frame sequences in NanoBanana, enhanced with PixVerse and Higgsfield AI, and scored with original music composed in Suno, the project shows how seamlessly multiple tools can come together. Final image cleanup in Photoshop and editing in DaVinci Resolve tied it all into one flowing piece.

👉 Watch Sara’s AI music video experiment here

💡 Industry Insight

What stood out to me this week is how creative AI is pushing into places we didn’t see coming. Runway’s world models are teaching robots, OpenAI is investing in films, China’s trillion-parameter releases are rewriting what’s possible, and now Warner Bros. is challenging Midjourney in court.

For creatives, that’s a clear signal: the projects we’re testing now, mock-ups, visuals, short films, aren’t just experiments. They’re the same skills that could shape new industries, new opportunities, and even new rules.

So keep creating, keep experimenting, and most importantly, keep your perspective at the center. That’s what makes these updates more than features—it’s what makes them matter.

That’s a wrap for this week!