My workflow: | |
First, I rotoscope the footage using After Effects. | |
Then, I create a mask using the rotoscoped footage. | |
(Optional; depending on the computer), I convert the frames per second to 12 fps for both the footage and the mask. | |
After that, I export both the footage and masks as PNG sequences to their own folders. | |
Next up, I run automatic 1111 and head to the img2img tab. | |
I used the animelike25D model, and I set the CFG value to 7. Denose Strength to.25, and the sample steps to 25. | |
For the active control nets, I set one to open pose and set the other to depth. | |
Then, I run the batch tab within img2img and set the input and mask images, and an output folder. | |