Zero-shot cloning has never been enough for studio work.
So we made our own model: Robin
A fully opensource 48khz English and Japanese voice model. (links below)
An Autoregressive architecture from scratch and designed with the OS community in mind from the start.
Been playing around with some pixel art thinking grids for agents. almost like loading loops
Try it here or remix it into something different
https://t.co/Cc2tAQBnnL
3/4
• ComfyUI nodes for inference and training already available
• An open license allowing commercial use
• Direct the tone of speech with emotion tags like [laugh] [cough] [shout] [angry]
@TheRenderRig@daodao166888 a lot of that look is code-driven, not keyframed. we built lattice so you can prompt infographic-style motion and adjust it live, then export. glad to walk you through it
@jacksteiner01@oiharshit most of these are code, not video gen. we built lattice for exactly this: describe the motion, claude builds it live, you tweak and export mp4. a few examples are on our profile
@aathikesavan457 the render-feedback loop matters most imo. we built lattice so claude sees the canvas live as it builds the motion graphics, which cuts the blind guessing a lot. can share our setup if useful
@ziwenxu_ lighting and post are most of the aaa feel. we built lattice, where claude builds real-time 3d on a gpu canvas and you push the look by prompting. happy to show what we'd try on this
@kelly_zhuoer@kaushikimmadi@X we built Lattice for basically this. it's a creative IDE where Claude builds the motion graphics live on a GPU canvas, you iterate by prompting, then export mp4. happy to share more if you want to try it
Lattice allowed me to remix @Machinedelusion realtime shadow box project and so I created these little dudes. Not sure why they're scared of me.
What would you name them?
Get a closer look here: https://t.co/X1F87bjzrN @fixedseed