ComfyUI workflow: animate characters/objects using LoRAs this is not for my demo (game)I hope it helped some of you
I’ve been building a workflow in ComfyUI that lets me generate consistent character and object animations using LoRAs driven by video motion.
The setup basically takes a video, extracts frames, optionally upscales or removes the background, then runs a dual-stage sampling process with LoRA conditioning to keep the character consistent across frames, and finally reconstructs everything back into an animation.
One of the biggest advantages is that it helps solve the usual inconsistency issues you get with diffusion across sequences, and it works pretty well for both characters and objects or even sprite-style outputs. I’m currently using this for a real project a historical narrative game set during the Hussite Wars mainly to prototype animations quickly and test gameplay systems before committing to final assets as a solo dev.
I’ve shared the full workflow through screenshots, so you can recreate the node setup directly and plug in your own LoRAs or models. If anyone wants help setting it up, improving results, or adapting it for their own use case, feel free to ask questions or DM me, I’m happy to help 👍.
