import cv2 import gradio as gr import numpy as np import torch from diffusers import ControlNetModel, StableDiffusionControlNetPipeline from PIL import Image from diffusion_webui.utils.model_list import ( controlnet_canny_model_list, stable_model_list, ) from diffusion_webui.utils.scheduler_list import ( SCHEDULER_LIST, get_scheduler_list, ) class StableDiffusionControlNetCannyGenerator: def __init__(self): self.pipe = None def load_model(self, stable_model_path, controlnet_model_path, scheduler): if self.pipe is None: controlnet = ControlNetModel.from_pretrained( controlnet_model_path, torch_dtype=torch.float16 ) self.pipe = StableDiffusionControlNetPipeline.from_pretrained( pretrained_model_name_or_path=stable_model_path, controlnet=controlnet, safety_checker=None, torch_dtype=torch.float16, ) self.pipe = get_scheduler_list(pipe=self.pipe, scheduler=scheduler) self.pipe.to("cuda") self.pipe.enable_xformers_memory_efficient_attention() return self.pipe def controlnet_canny( self, image_path: str, ): image = Image.open(image_path) image = np.array(image) image = cv2.Canny(image, 100, 200) image = image[:, :, None] image = np.concatenate([image, image, image], axis=2) image = Image.fromarray(image) return image def generate_image( self, image_path: str, stable_model_path: str, controlnet_model_path: str, prompt: str, negative_prompt: str, num_images_per_prompt: int, guidance_scale: int, num_inference_step: int, scheduler: str, seed_generator: int, ): pipe = self.load_model( stable_model_path=stable_model_path, controlnet_model_path=controlnet_model_path, scheduler=scheduler, ) image = self.controlnet_canny(image_path=image_path) if seed_generator == 0: random_seed = torch.randint(0, 1000000, (1,)) generator = torch.manual_seed(random_seed) else: generator = torch.manual_seed(seed_generator) output = pipe( prompt=prompt, image=image, negative_prompt=negative_prompt, num_images_per_prompt=num_images_per_prompt, num_inference_steps=num_inference_step, guidance_scale=guidance_scale, generator=generator, ).images return output def app(): with gr.Blocks(): with gr.Row(): with gr.Column(): controlnet_canny_image_file = gr.Image( type="filepath", label="Image" ) controlnet_canny_prompt = gr.Textbox( lines=1, placeholder="Prompt", show_label=False, ) controlnet_canny_negative_prompt = gr.Textbox( lines=1, placeholder="Negative Prompt", show_label=False, ) with gr.Row(): with gr.Column(): controlnet_canny_stable_model_id = gr.Dropdown( choices=stable_model_list, value=stable_model_list[0], label="Stable Model Id", ) controlnet_canny_guidance_scale = gr.Slider( minimum=0.1, maximum=15, step=0.1, value=7.5, label="Guidance Scale", ) controlnet_canny_num_inference_step = gr.Slider( minimum=1, maximum=100, step=1, value=50, label="Num Inference Step", ) controlnet_canny_num_images_per_prompt = gr.Slider( minimum=1, maximum=10, step=1, value=1, label="Number Of Images", ) with gr.Row(): with gr.Column(): controlnet_canny_model_id = gr.Dropdown( choices=controlnet_canny_model_list, value=controlnet_canny_model_list[0], label="ControlNet Model Id", ) controlnet_canny_scheduler = gr.Dropdown( choices=SCHEDULER_LIST, value=SCHEDULER_LIST[0], label="Scheduler", ) controlnet_canny_seed_generator = gr.Number( value=0, label="Seed Generator", ) controlnet_canny_predict = gr.Button(value="Generator") with gr.Column(): output_image = gr.Gallery( label="Generated images", show_label=False, elem_id="gallery", ).style(grid=(1, 2)) controlnet_canny_predict.click( fn=StableDiffusionControlNetCannyGenerator().generate_image, inputs=[ controlnet_canny_image_file, controlnet_canny_stable_model_id, controlnet_canny_model_id, controlnet_canny_prompt, controlnet_canny_negative_prompt, controlnet_canny_num_images_per_prompt, controlnet_canny_guidance_scale, controlnet_canny_num_inference_step, controlnet_canny_scheduler, controlnet_canny_seed_generator, ], outputs=[output_image], )