import os import sys from pathlib import Path from collections import OrderedDict import gradio as gr import shutil import uuid import torch from PIL import Image # Force CPU usage and disable CUDA completely torch.backends.cudnn.enabled = False os.environ['CUDA_VISIBLE_DEVICES'] = '-1' torch.cuda.is_available = lambda: False torch.cuda.device_count = lambda: 0 torch.cuda.get_device_name = lambda x: 'cpu' torch.cuda.current_device = lambda: 0 torch.cuda.set_device = lambda x: None torch.Tensor.cuda = lambda self, device=None: self torch.nn.Module.cuda = lambda self, device=None: self demo_path = Path(__file__).resolve().parent root_path = demo_path sys.path.append(str(root_path)) from src import models from src.methods import rasg, sd, sr from src.utils import IImage, poisson_blend, image_from_url_text TMP_DIR = root_path / 'gradio_tmp' if TMP_DIR.exists(): shutil.rmtree(str(TMP_DIR)) TMP_DIR.mkdir(exist_ok=True, parents=True) os.environ['GRADIO_TEMP_DIR'] = str(TMP_DIR) on_huggingspace = os.environ.get("SPACE_AUTHOR_NAME") == "PAIR" negative_prompt_str = "text, bad anatomy, bad proportions, blurry, cropped, deformed, disfigured, duplicate, error, extra limbs, gross proportions, jpeg artifacts, long neck, low quality, lowres, malformed, morbid, mutated, mutilated, out of frame, ugly, worst quality" positive_prompt_str = "Full HD, 4K, high quality, high resolution" examples_path = root_path / '__assets__/demo/examples' example_inputs = [ [f'{examples_path}/images_1024/a40.jpg', f'{examples_path}/images_2048/a40.jpg', 'medieval castle'], [f'{examples_path}/images_1024/a4.jpg', f'{examples_path}/images_2048/a4.jpg', 'parrot'], [f'{examples_path}/images_1024/a65.jpg', f'{examples_path}/images_2048/a65.jpg', 'hoodie'], [f'{examples_path}/images_1024/a54.jpg', f'{examples_path}/images_2048/a54.jpg', 'salad'], [f'{examples_path}/images_1024/a51.jpg', f'{examples_path}/images_2048/a51.jpg', 'space helmet'], [f'{examples_path}/images_1024/a46.jpg', f'{examples_path}/images_2048/a46.jpg', 'stack of books'], [f'{examples_path}/images_1024/a19.jpg', f'{examples_path}/images_2048/a19.jpg', 'antique greek vase'], [f'{examples_path}/images_1024/a2.jpg', f'{examples_path}/images_2048/a2.jpg', 'sunglasses'], ] thumbnails = [ 'https://lh3.googleusercontent.com/pw/ABLVV87bkFc_SRKrbXuk5BTp18dETNm18MLbjoJo6JvwbIkYtjZXrjU_H1dCJIP799OJjHTZmo19mYVyMCC1RLmwqzoZrgwQzfB-SCtxLa83IbXBQ23xzmKoZgsRlPztxNJD6gmXzFyatdLRzDxHIusBQLUz=w3580-h1150-s-no-gm', 'https://lh3.googleusercontent.com/pw/ABLVV85RWtrpTf1tMp2p3q37eg5DlFp5znifALK_JTjvxJua8UYMjytVoEy2GUW2cLXgBvQyYKg7GvrWXQ5hkdAsyih5Rf4rFnDq-JoiQYhVZHStCZLKxmeAlQna5ZwMPVTKG1TK63DH_OdK58gvSjWtF2ww=w3580-h1152-s-no-gm', 'https://lh3.googleusercontent.com/pw/ABLVV84dkaU6SQs9fyDjajpk1X9JkYp_zQBEnPVL67oi11_05U6-Ys5ydQpuny8GBQCMyVbFKxJ5unn9w__gmP9K0cKQ4_IVoT7Hvfmya71klDqSI7vu9Iy_5P2Il5-0giJFpumtffBA3kryn1xtJdR4vSA0=w2924-h1858-s-no-gm', 'https://lh3.googleusercontent.com/pw/ABLVV853ZyjvS4LvcPpVMY9BWz-232omt3-hgRiGcky_3ojE6WLKgtsrftsg1jSrUm2ccT_UOa279CulZy6fdnH_Xg1SunyRBxaRjOK0uxAkUFwb60rR1S4hI2MmhLV7KCi3tw1A-oiGi0f9JINyade-322A=w2622-h1858-s-no-gm', 'https://lh3.googleusercontent.com/pw/ABLVV86AJGUVGjb0i6CPg8zlJlWObNY0xdOzM1x5Bq9gKhP-ZWre5aaexRJDxQUO2gmJtRIyohD88FJDG_aVX2G5M0QOyGRWlZmx7tOVXLh-Kbesobxo9MfD-wqk9Ts9O8NUGtIwkWzo9SEs2opKdu83gB9F=w2528-h1858-s-no-gm', 'https://lh3.googleusercontent.com/pw/ABLVV87MplTciS7z-4i-eY3B3L0YhaK8UEQ3pTQD6W6uYVGR4hPD9u1WGEGyfg5ddqU-Bx2BrKskDhwxzF746cRhgFU5aPtbYA_-O7KfqXe9IsMxYCgUKxEHBm2ncqy64V-w-N8XOFgUMkAQqcuuNZ8Xapqp=w3580-h1186-s-no-gm', 'https://lh3.googleusercontent.com/pw/ABLVV877Esi6l2Kuw3akH5QBlmDAbWydZDZEEJqlZ_N-X7g33NQZU8nv_UKdAVETS7q23byTuldIAhW-q99zCycFB8Yfc-5e_WPNIM9icU0p3gd6DUVZR233ZNUtLca384MYGIhMGud9Y_Xed1I3PpiMhrpG=w2846-h1858-s-no-gm', 'https://lh3.googleusercontent.com/pw/ABLVV85hMQbSB6fCokdyut4ke7xTUqjERhuYygnj7T8IIA1k48e9GkaowDywPZzi5QJzZfj7wU3bgBHzjxop19qK1zOi5XDrjfXkn5bwj4MxicHa3TG-Rc-V-c1uyZVUyviyUlkGZ62FxuVROw2x0aGJIcr0=w3580-h1382-s-no-gm' ] example_previews = [ [thumbnails[0], 'Prompt: medieval castle'], [thumbnails[1], 'Prompt: parrot'], [thumbnails[2], 'Prompt: hoodie'], [thumbnails[3], 'Prompt: salad'], [thumbnails[4], 'Prompt: space helmet'], [thumbnails[5], 'Prompt: stack of books'], [thumbnails[6], 'Prompt: antique greek vase'], [thumbnails[7], 'Prompt: sunglasses'], ] # Monkey patch any remaining CUDA calls in the models original_to = torch.Tensor.to def patched_to(self, *args, **kwargs): if len(args) > 0 and isinstance(args[0], str) and args[0] == 'cuda': return original_to(self, 'cpu') if 'device' in kwargs and kwargs['device'] == 'cuda': kwargs['device'] = 'cpu' return original_to(self, *args, **kwargs) torch.Tensor.to = patched_to # Load models with CPU only models.pre_download_inpainting_models() inpainting_models = OrderedDict([ ("Dreamshaper Inpainting V8", 'ds8_inp'), ("Stable-Inpainting 2.0", 'sd2_inp'), ("Stable-Inpainting 1.5", 'sd15_inp') ]) # Patch model loading to ensure CPU usage original_load_model = models.load_inpainting_model def patched_load_model(*args, **kwargs): kwargs['device'] = 'cpu' model = original_load_model(*args, **kwargs) model.to('cpu') return model models.load_inpainting_model = patched_load_model original_sr_load_model = models.sd2_sr.load_model def patched_sr_load_model(*args, **kwargs): kwargs['device'] = 'cpu' model = original_sr_load_model(*args, **kwargs) # Handle DDIM object which doesn't have to() method if hasattr(model, 'model'): # If there's a main model component model.model.to('cpu') if hasattr(model, 'diffusion'): # Some diffusion models have this model.diffusion.to('cpu') return model models.sd2_sr.load_model = patched_sr_load_model original_sam_load_model = models.sam.load_model def patched_sam_load_model(*args, **kwargs): kwargs['device'] = 'cpu' model = original_sam_load_model(*args, **kwargs) # SAM predictor doesn't have to() method but its model does if hasattr(model, 'model'): model.model.to('cpu') return model models.sam.load_model = patched_sam_load_model # Load models with CPU sr_model = models.sd2_sr.load_model(device='cpu') sam_predictor = models.sam.load_model(device='cpu') inp_model_name = list(inpainting_models.keys())[0] inp_model = models.load_inpainting_model( inpainting_models[inp_model_name], device='cpu', cache=True) def set_model_from_name(new_inp_model_name): global inp_model global inp_model_name if new_inp_model_name != inp_model_name: print(f"Activating Inpaintng Model: {new_inp_model_name}") inp_model = models.load_inpainting_model( inpainting_models[new_inp_model_name], device='cpu', cache=True) inp_model_name = new_inp_model_name def save_user_session(hr_image, hr_mask, lr_results, prompt, session_id=None): if session_id == '': session_id = str(uuid.uuid4()) session_dir = TMP_DIR / session_id session_dir.mkdir(exist_ok=True, parents=True) hr_image.save(session_dir / 'hr_image.png') hr_mask.save(session_dir / 'hr_mask.png') lr_results_dir = session_dir / 'lr_results' if lr_results_dir.exists(): shutil.rmtree(lr_results_dir) lr_results_dir.mkdir(parents=True) for i, lr_result in enumerate(lr_results): lr_result.save(lr_results_dir / f'{i}.png') with open(session_dir / 'prompt.txt', 'w') as f: f.write(prompt) return session_id def recover_user_session(session_id): if session_id == '': return None, None, [], '' session_dir = TMP_DIR / session_id lr_results_dir = session_dir / 'lr_results' hr_image = Image.open(session_dir / 'hr_image.png') hr_mask = Image.open(session_dir / 'hr_mask.png') lr_result_paths = list(lr_results_dir.glob('*.png')) gallery = [] for lr_result_path in sorted(lr_result_paths): gallery.append(Image.open(lr_result_path)) with open(session_dir / 'prompt.txt', "r") as f: prompt = f.read() return hr_image, hr_mask, gallery, prompt def inpainting_run(model_name, use_rasg, use_painta, prompt, imageMask, hr_image, seed, eta, negative_prompt, positive_prompt, ddim_steps, guidance_scale=7.5, batch_size=1, session_id='' ): set_model_from_name(model_name) method = ['default'] if use_painta: method.append('painta') if use_rasg: method.append('rasg') method = '-'.join(method) if use_rasg: inpainting_f = rasg.run else: inpainting_f = sd.run seed = int(seed) batch_size = max(1, min(int(batch_size), 4)) image = IImage(hr_image).resize(512) mask = IImage(imageMask['mask']).rgb().resize(512) method = ['default'] if use_painta: method.append('painta') method = '-'.join(method) inpainted_images = [] blended_images = [] for i in range(batch_size): seed = seed + i * 1000 inpainted_image = inpainting_f( ddim=inp_model, method=method, prompt=prompt, image=image, mask=mask, seed=seed, eta=eta, negative_prompt=negative_prompt, positive_prompt=positive_prompt, num_steps=ddim_steps, guidance_scale=guidance_scale ).crop(image.size) blended_image = poisson_blend( orig_img=image.data[0], fake_img=inpainted_image.data[0], mask=mask.data[0], dilation=12 ) blended_images.append(blended_image) inpainted_images.append(inpainted_image.pil()) session_id = save_user_session( hr_image, imageMask['mask'], inpainted_images, prompt, session_id=session_id) return blended_images, session_id def upscale_run( ddim_steps, seed, use_sam_mask, session_id, img_index, negative_prompt='', positive_prompt='high resolution professional photo' ): hr_image, hr_mask, gallery, prompt = recover_user_session(session_id) if len(gallery) == 0: return Image.open(root_path / '__assets__/demo/sr_info.png') seed = int(seed) img_index = int(img_index) img_index = 0 if img_index < 0 else img_index img_index = len(gallery) - 1 if img_index >= len(gallery) else img_index inpainted_image = gallery[img_index if img_index >= 0 else 0] output_image = sr.run( sr_model, sam_predictor, inpainted_image, hr_image, hr_mask, prompt=f'{prompt}, {positive_prompt}', noise_level=20, blend_trick=True, blend_output=True, negative_prompt=negative_prompt, seed=seed, use_sam_mask=use_sam_mask ) return output_image with gr.Blocks(css=demo_path / 'style.css') as demo: gr.HTML( """

🧑‍🎨 HD-Painter Demo (CPU Mode)

Hayk Manukyan1*, Andranik Sargsyan1*, Barsegh Atanyan1, Zhangyang Wang1,2, Shant Navasardyan1 and Humphrey Shi1,3

1Picsart AI Resarch (PAIR), 2UT Austin, 3Georgia Tech

[arXiv] [GitHub]

HD-Painter enables prompt-faithfull and high resolution (up to 2k) image inpainting upon any diffusion-based image inpainting method.
Note: Running on CPU may be slower than GPU.

""") if on_huggingspace: gr.HTML("""

For faster inference without waiting in queue, you may duplicate the space and upgrade to the suggested GPU in settings.
Duplicate Space

""") with open(demo_path / 'script.js', 'r') as f: js_str = f.read() demo.load(_js=js_str) with gr.Row(): with gr.Column(): model_picker = gr.Dropdown( list(inpainting_models.keys()), value=list(inpainting_models.keys())[0], label="Please select a model!", ) with gr.Column(): use_painta = gr.Checkbox(value=True, label="Use PAIntA") use_rasg = gr.Checkbox(value=True, label="Use RASG") prompt = gr.Textbox(label="Inpainting Prompt") with gr.Row(): with gr.Column(): imageMask = gr.ImageMask(label="Input Image", brush_color='#ff0000', elem_id="inputmask", type="pil") hr_image = gr.Image(visible=False, type="pil") hr_image.change(fn=None, _js="function() {setTimeout(imageMaskResize, 200);}", inputs=[], outputs=[]) imageMask.upload( fn=None, _js="async function (a) {hr_img = await resize_b64_img(a['image'], 2048); dp_img = await resize_b64_img(hr_img, 1024); return [hr_img, {image: dp_img, mask: null}]}", inputs=[imageMask], outputs=[hr_image, imageMask], ) with gr.Row(): inpaint_btn = gr.Button("Inpaint", scale=0) with gr.Accordion('Advanced options', open=False): guidance_scale = gr.Slider(minimum=0, maximum=30, value=7.5, label="Guidance Scale") eta = gr.Slider(minimum=0, maximum=1, value=0.1, label="eta") ddim_steps = gr.Slider(minimum=10, maximum=100, value=50, step=1, label='Number of diffusion steps') with gr.Row(): seed = gr.Number(value=49123, label="Seed") batch_size = gr.Number(value=1, label="Batch size", minimum=1, maximum=4) negative_prompt = gr.Textbox(value=negative_prompt_str, label="Negative prompt", lines=3) positive_prompt = gr.Textbox(value=positive_prompt_str, label="Positive prompt", lines=1) with gr.Column(): with gr.Row(): output_gallery = gr.Gallery( [], columns=4, preview=True, allow_preview=True, object_fit='scale-down', elem_id='outputgallery' ) with gr.Row(): upscale_btn = gr.Button("Send to Inpainting-Specialized Super-Resolution (x4)", scale=1) with gr.Row(): use_sam_mask = gr.Checkbox(value=False, label="Use SAM mask for background preservation (for SR only, experimental feature)") with gr.Row(): hires_image = gr.Image(label="Hi-res Image") label = gr.Markdown("## High-Resolution Generation Samples (2048px large side)") with gr.Column(): example_container = gr.Gallery( example_previews, columns=4, preview=True, allow_preview=True, object_fit='scale-down' ) gr.Examples( [example_inputs[i] + [[example_previews[i]]] for i in range(len(example_previews))], [imageMask, hr_image, prompt, example_container], elem_id='examples' ) session_id = gr.Textbox(value='', visible=False) html_info = gr.HTML(elem_id=f'html_info', elem_classes="infotext") inpaint_btn.click( fn=inpainting_run, inputs=[ model_picker, use_rasg, use_painta, prompt, imageMask, hr_image, seed, eta, negative_prompt, positive_prompt, ddim_steps, guidance_scale, batch_size, session_id ], outputs=[output_gallery, session_id], api_name="inpaint" ) upscale_btn.click( fn=upscale_run, inputs=[ ddim_steps, seed, use_sam_mask, session_id, html_info ], outputs=[hires_image], api_name="upscale", _js="function(a, b, c, d, e){ return [a, b, c, d, selected_gallery_index()] }", ) demo.queue(max_size=20) demo.launch(share=True, allowed_paths=[str(TMP_DIR)])