ComfyUI-See-through

Why we included this project

If you produce Live2D or 2.5D animation from anime artwork, this plugin removes one of the most tedious manual steps in the pipeline. Feed it a single character illustration and it splits the image into up to 24 semantic transparent layers (hair, face, eyes, clothing, accessories) with depth ordering computed automatically, so the pieces come out already stacked in the correct draw order. It also splits symmetric parts like eyes and ears into left/right, and separates hair into front and back using depth clustering. That's the kind of prep work that normally eats hours in an editor. The result is a layered PSD you can download straight from the browser, and a separate depth PSD is available for parallax or 3D work. Because it ships as ComfyUI custom nodes, it slots into an existing node-based workflow rather than forcing a separate tool, and the VRAM options make it usable on mid-range GPUs.