layerdiffusion
Doesn't work with any Illustrious (SDXL) based models
I tested it on base SDXL models and Pony models, those work for both the attention injection and conv. Switching to Illustirous v1.0, NoobXL, or some merged like WAIIllustrious only produces a slight pink hue over the image for either attention injection or conv.
Yea doesn't work for me either, tried several illustrious models.
EDIT: When using Preview Image node in ComfyUI or in Forge fetching the pre-process image before layerdiffuse kicks in, it is most noticeable that usually the backgrounds generated for these models are not optimal for LayerDiffuse, producing non-gray backgrounds, either with blue-ish/yellowish tint of white. It works if you manage to prompt it to make a simple plaing slightly dark variant of neutral gray background. (from for example #757575-#898989)
