Five tools, trained by the community, open to everyone.
Congratulations to the winners of the LTX LoRA Jam. We shipped the LTX Trainer, the community trained on it, and what came back were LoRAs with weights, ComfyUI nodes, and documentation behind them.
Check out the winners ⬇️
@ZorviaLux@ltx_io It a visual style I always liked. Mary Poppins has an incredible sequence blending live action and traditional 2d animation. Although there they have the environment 2d and the character 3d. Maybe it should try that too :)
Built this for the @ltx_io LoRA Jam 🎬
An IC-LoRA that turns ONE character in live-action footage into 2D cel animation while the background stays photographic.
LTX-2.5 22B. 20 training pairs I made myself.
Model + dataset public:
https://t.co/Yv4D6HlX4G
Slice of Life" is my entry for the @capcutapp × TIFF Creative Challenge 🎬
A one-take short film made entirely in CapCut Video Studio with Seedance 2.5: 9 reference images for characters and location.
One street. One minute. Many lives.
#CapCutAI#CapCutCPP
Hardest part of training a video LoRA isn't the training.
It's that the data doesn't exist.
I needed the same shot twice — once real, once with one character drawn.
Generate a clip → take frame 1 → swap the character to 2D → re-animate.
All pairs:
https://t.co/sK5aYZ6LzD
Slice of Life" is my entry for the @capcutapp × TIFF Creative Challenge 🎬
A one-take short film made entirely in CapCut Video Studio with Seedance 2.5: 9 reference images for characters and location.
One street. One minute. Many lives.
#CapCutAI#CapCutCPP