Train an IC-LoRA that regenerates only the masked region of a video, guided by the kept pixels and a separate reference/control video.
No supported capabilities recorded.
/v1/models/fal/fal-ai/ltx23-trainer-v2/ic-lora/v2v-masked