A little different post but here is my workflow. More models coming again tomorrow :)!!
1. First we use ComfyUI with an image generation library (Stable Diffusion 3). I have a python script that posts tons of prompts over and the images start getting generated.
2. Next we import the images into our “AI Image Tinder” app where we swipe left to reject and swipe right to approve – not all images are going to work so this is critical. I think we have maybe 20% of the images that will work.
3. With these images I manually edit them in Seashore to clean up stuff that won’t render well (extra cape bits, broken weapons etc…).
4. Then I plug it into another “blender processor” program which posts the API calls to Rodin and has a helper script to convert those obj + shades.png files to processed.stl files
5. Finally I do my best to cleanup the model in Blender (and then post here!)
