Closed MoonMoon82 closed 1 year ago
Ok, just saw your "multicn_depth+seg.blend" file. Awesome! But it's just passing the depthmap file. In my case it doesn't seem to calculate the segmentation map. Any suggestions?
You could bring the depth and segmentation maps to the Web UI directly and see if it works. In some of my previous cases, I have found that the segmentation map may not be as useful in certain situations.
Currently, Blender renders the WHOLE image, passes it to ControlNet with depth-preprocessor to get the (ai-calculated) depthmap of a fully rendered image.
Why not directly passing the different data- and light-passes to controlnet without any preprocessor?