zef

[zef @dddjef]

@tintwotin @ltx Seems great, but it's unclear to me if it's designed to infere locally or from web services ?

[Dr. Dreams @DrDreams]

New pod up! Hollywood is NOT cooked by AI 🔥

Top productions are already using hybrid AI + live action workflows — and it’s only getting better.

New chat with the great and powerful @cfryant breaking down:
• His Amazon commercial behind the scenes
• The viral Jurassic Park rubber duck experiment
• Why moving cameras still break AI
• The insane “Brain Halo Device” future

...and a lot more

Watch here 👇
https://t.co/UjLwzNExZh

#AI #Filmmaking #Hollywood

[zef @dddjef]

@F_Rimasson Very very usefull. I kind if remember a lightwave plugin using vector to initialize the fur direction

[Claude @claudeai]

Claude now connects to the tools creative professionals already use.

With the new Blender connector, you can debug a scene, build new tools, or batch-apply changes across every object, directly from Claude.

[LTX Studio @LTXStudio]

Seedance 2.0 is now live in LTX Studio.

1080p generation. No restrictions.

Follow + Retweet to get 800 credits in your DMs + 33% off Pro plan

24 HOURS ONLY ⏱️

[zef @dddjef]

@PixVerse_ Hey

[PixVerse @PixVerse_]

720p was fine. 1080P changes everything.
Seedance 2.0 is live on PixVerse — sharper details, smoother motion, cinema-grade lighting.
72H early access starts now.
RT+Follow+Reply=300Creds(72H ONLY)

[zef @dddjef]

@_pepeland_ It's well deserved

[zef @dddjef]

@Anima_Labs @MartiniArt_ I can't see the seedance access in the web page

[Riley Brown @rileybrown]

Openclaw can control blender and create 3D models that are animated....

That you can insert into websites with a single prompt.

[Henry Daubrez 🌸💀 @henrydaubrez]

Seedance 2.0 won't save you from learning the basics of telling a good story.

Sure you can create never ending mashed up IP battles but...what does it say about you, and is there a lasting public for this?

[Francois Rimasson @F_Rimasson]

Mikros Animation is looking for a blender dev
https://t.co/FgEVXZjdgA

[zef @dddjef]

@AimewillSo @Paroles_auteurs Tu peux donner la source de cette stat?

[TomLikesRobots🤖 @TomLikesRobots]

Exciting - a new model to test.
Q2 from @ViduAI_official has just been announced and is very, very good.
Animated fight scenes have always been difficult for AI video models but Q2 got this one right straight away.
Input image from Midjourney.

[zef @dddjef]

@F_Rimasson There's was something named "particular" I think a long time ago

[Logan Kilpatrick @OfficialLoganK]

Introducing Genie 3, the most advanced world simulator ever created, enabled by numerous research breakthroughs. 🤯

Featuring high fidelity visuals, 20-24 fps, prompting on the go, world memory, and more.

[Runway @runwayml]

Introducing Runway Aleph, a new way to edit, transform and generate video.

Aleph is a state-of-the-art in-context video model, setting a new frontier for multi-task visual generation, with the ability to perform a wide range of edits on an input video such as adding, removing and transforming objects, getting new angles of a scene and modifying style and lighting, among many other tasks.

[Hyper3D by Deemos @DeemosTech]

@elonmusk @grok @xai This is incredible! @grok can understand the internal relationships between objects, and when combined with Rodin Gen-2, it can directly generate .urdf ready for production.

This will revolutionize 3D Gen, video games, embodied AI, robotics, and even 3D printing.
#Grok4

[Kimi.ai @Kimi_Moonshot]

🚀 Kimi K2 is here!
Built for code & agentic tasks.

Try it now at https://t.co/7YeIld97BT or via API.

[Henry Daubrez 🌸💀 @henrydaubrez]

MULTI-MODEL (Google VEO2's ingredients + VEO3) ANIMATION WORKFLOW 👇

(🔊 ON)

I built a short sequence using a mix of tools to push both aesthetic complexity and shot control...turns out it worked better than expected.

🔹 Backgrounds: I started with a customized version of Midjourney, tuned toward the stop-motion universe I wanted to explore and generated different backgrounds reflecting a more artistic and painterly universe while sticking to the stop-motion feel.

🔹 Character: Generated a full sheet in Imagen 3, then stylized it in Flux Kontext to match the tactile feel I wanted to have. Imagen understood what I wanted to create but Flux enabled me to fine-tune, fix, and relight..which is the next point

🔹 Lighting: Used Flux again to match light direction and color between my background and character..small tweak, big payoff as VEO2's ingredients don't feel like the relighting is really strong in most of my attempts.

🔹 Action shots: Here’s where VEO2 came in: thanks to ingredients, I could feed both the background and character, then define precise actions.
Pro tip: start the prompt with “transition to” and ask for a close-up or angle change...this lets you chain scenes with intentional camera moves. Props to Martin Nebelong for that one, and it seems like the Google team made it even easier now...it used to be a lot more of a dice rolling thingy.

🔹 Dialogue: Once I had a strong close-up, I isolated the first frame and fed it into VEO3, now with image-to-video and audio. That gave me synchronized motion and voice. I could easily get more of those shots by training a voice on Eleven Labs pro and then passing each new generation through a voice modifier with that cloned voice...

🔹 Final pass: A very small bit of compositing, stock FX where needed, and Udio for the score.

What I like about this workflow is the separation of ingredients...almost like having a visual toolbox.

VEO2 remains the one model letting you control scene elements this precisely, and VEO3 now picks up the baton with expressive audio-based moments.
Still early, still messy, but there’s definitely something here.
Building scenes like this used to take weeks. Now it’s a few hours of prototyping, revision, and style tuning.

More tests soon.