Our current pipeline uses multi controlnet with canny and inpaint and use the controlnetinpaint pipeline
Is the inpaint control net checkpoint available for SD XL?
Reference Code:
controlnet_inpaint_model = ControlNetModel.from_pretrained(
CONTROLNET_INPAINT_MODEL_ID, torch_dtype=torch.float16, cache_dir=DIFFUSION_CHECKPOINTS_PATH).to(
MODEL_DEVICE)
controlnet_hed_model = ControlNetModel.from_pretrained(
CONTROLNET_HED_MODEL_ID, torch_dtype=torch.float16, cache_dir=DIFFUSION_CHECKPOINTS_PATH).to(
MODEL_DEVICE)
controlnet_canny_model = ControlNetModel.from_pretrained(
CONTROLNET_CANNY_MODEL_ID, torch_dtype=torch.float16, cache_dir=DIFFUSION_CHECKPOINTS_PATH).to(
MODEL_DEVICE)
controlnet_depth_model = ControlNetModel.from_pretrained(
CONTROLNET_DEPTH_MODEL_ID, torch_dtype=torch.float16, cache_dir=DIFFUSION_CHECKPOINTS_PATH).to(
MODEL_DEVICE)
self.inpaint_pipeline = StableDiffusionControlNetInpaintPipeline.from_pretrained(
CUSTOM_SD_MODEL,
controlnet=[controlnet_inpaint_model, controlnet_hed_model],
torch_dtype=torch.float16).to(MODEL_DEVICE)