t̶e̶x̶t̶ ̶t̶o̶ ̶v̶i̶d̶e̶o̶ ̶
̶i̶m̶a̶g̶e̶ ̶t̶o̶ ̶v̶i̶d̶e̶o̶ ̶
Previs to video ✔
https://t.co/BtrUfccmMV took 49 secs to create the pre-vis video.
passed it to seedance 2.0 mini - cost: $0.721
both single shotted
#seedance#3d#threejs#animation#AIVideo
The shark was chasing him before we gave it skin. 🦈
Clay previs below, final shot above. We blocked the swim, the shark’s approach, and the camera cuts in Dimension, right in the browser.
Build your own shot at https://t.co/BtrUfccmMV
This entire 3D scene was built and directed in a browser.
The run, the falling debris, the camera cuts. You can see the original previs in the top right and how the video model brought it to life.
Build your own scene at https://t.co/KfujfrdgPK.
#seedance#3d#threejs #animation #AIVideo
Stop burning credits re-prompting your video model.
Block out the movement, camera, and timing in 3D, then feed the model your previs so it can follow the shot you actually want.
Blender-like control, right in your browser. Try it at https://t.co/BtrUfcbOXn.
Your video model shouldn't be guessing the shot.
We brought Blender into the browser to solve that.
Set the camera, place your objects, then generate exactly the scene you planned.
try it: https://t.co/BtrUfccmMV
demo:
Today, we're shipping real-time graphics.
Using Sol 6.1 Ultrafast we can generate custom overlays, diagrams, and even interactive 3D scenes right on a user's screen, no chat.
It's a new visual interface for AI.
Here's me learning about LLMs + hardware.
Demo:
@ZeroBorne @emmanuel_2m lmao ye fully a scam.
And also I can legitimately do this rigged animation, directly on the browser with https://t.co/BtrUfccmMV
Top is agent made, bottom is seedance 2.0 mini pass:
I think Blender is dead for this workflow.
Agents in Dimension built the entire scene directly in the browser, then handed it off to @fal's new 3D-to-video model for the photoreal pass.
That browser-native agentic workflow is where this gets really interesting.
Agents using Blender will have a massive impact on media production.
I asked Astra to make a scene where people stand in a field watching a UFO - and iterated on exact timing + camera angles.
Then I sent it to @fal's new 3D to video model and got this photoreal clip 🤯
@LuisHiluy Yep, exactly the problem I was trying to solve. Get the camera, body motion, and scene layout right in a lightweight browser tool first, then let diffusion do what its actually good at.
Is Blender useless now for previs?
I was able to block the whole scene, camera, and character motion right in the browser with Dimension, then hand it straight to a diffusion model.
Same control. No install, no learning curve.
#blender#Astra#threejs#3danimation