As I understand you making following steps: 1. Fine tune dreambooth on all frames 2. Fine tune animatediff motion module 3. Extract controlnet from video 4. Combine them to infer Can you share hyperparameters of finetuning? 1. Do you finetune LoRA or Dreambooth? How many steps, what lr? 2. Do you finetune all module or selected layers?
As I understand you making following steps:
Can you share hyperparameters of finetuning?