import os import random import uuid import gradio as gr import numpy as np from PIL import Image import spaces import torch from diffusers import StableDiffusionXLPipeline, DPMSolverSinglestepScheduler, AutoencoderKL DESCRIPTION = """# Stable Diffusion 3""" if not torch.cuda.is_available(): DESCRIPTION += "\n

Running on CPU 🥶 This demo may not work on CPU.

" MAX_SEED = np.iinfo(np.int32).max CACHE_EXAMPLES = False MAX_IMAGE_SIZE = int(os.getenv("MAX_IMAGE_SIZE", "1536")) USE_TORCH_COMPILE = False ENABLE_CPU_OFFLOAD = os.getenv("ENABLE_CPU_OFFLOAD", "0") == "1" device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu") vae = AutoencoderKL.from_pretrained("madebyollin/sdxl-vae-fp16-fix", torch_dtype=torch.float16) if torch.cuda.is_available(): pipe = StableDiffusionXLPipeline.from_single_file( "https://huggingface.co/kadirnar/Black-Hole/blob/main/tachyon.safetensors", torch_dtype=torch.float16, use_safetensors=True, add_watermarker=False, variant="fp16", vae=vae, ) if ENABLE_CPU_OFFLOAD: pipe.enable_model_cpu_offload() else: pipe.to(device) print("Loaded on Device!") if USE_TORCH_COMPILE: pipe.unet = torch.compile(pipe.unet, mode="reduce-overhead", fullgraph=True) print("Model Compiled!") def save_image(img): unique_name = str(uuid.uuid4()) + ".png" img.save(unique_name) return unique_name def randomize_seed_fn(seed: int, randomize_seed: bool) -> int: if randomize_seed: seed = random.randint(0, MAX_SEED) return seed @spaces.GPU(enable_queue=True) def generate( prompt: str, negative_prompt: str = "", use_negative_prompt: bool = False, seed: int = 0, width: int = 1024, height: int = 1024, guidance_scale: float = 3, randomize_seed: bool = False, num_inference_steps=5, NUM_IMAGES_PER_PROMPT=1, use_resolution_binning: bool = True, progress=gr.Progress(track_tqdm=True), ): pipe.to(device) seed = int(randomize_seed_fn(seed, randomize_seed)) generator = torch.Generator().manual_seed(seed) sampling_schedule = [999, 845, 730, 587, 443, 310, 193, 116, 53, 13, 0] #pipe.scheduler = DPMSolverSinglestepScheduler(use_karras_sigmas=True).from_config(pipe.scheduler.config) #pipe.scheduler = DPMSolverMultistepScheduler(algorithm_type="sde-dpmsolver++").from_config(pipe.scheduler.config) if not use_negative_prompt: negative_prompt = None # type: ignore output = pipe( prompt=prompt, negative_prompt=negative_prompt, width=width, height=height, guidance_scale=guidance_scale, num_inference_steps=num_inference_steps, generator=generator, num_images_per_prompt=NUM_IMAGES_PER_PROMPT, use_resolution_binning=use_resolution_binning, output_type="pil", ).images return output examples = [ "neon holography crystal cat", "a cat eating a piece of cheese", "an astronaut riding a horse in space", "a cartoon of a boy playing with a tiger", "a cute robot artist painting on an easel, concept art", "a close up of a woman wearing a transparent, prismatic, elaborate nemeses headdress, over the should pose, brown skin-tone" ] css = ''' .gradio-container{max-width: 1000px !important} h1{text-align:center} ''' with gr.Blocks(css=css) as demo: with gr.Row(): with gr.Column(): gr.HTML( """

Fast Tachyon SDXL

""" ) gr.HTML( """

Follow me for more! Twitter | Github | Linkedin

""" ) with gr.Group(): with gr.Row(): prompt = gr.Text( label="Prompt", show_label=False, max_lines=1, placeholder="Enter your prompt", container=False, ) run_button = gr.Button("Run", scale=0) result = gr.Gallery(label="Result", elem_id="gallery", show_label=False) with gr.Accordion("Advanced options", open=False): with gr.Row(): use_negative_prompt = gr.Checkbox(label="Use negative prompt", value=True) negative_prompt = gr.Text( label="Negative prompt", max_lines=1, value = "deformed, distorted, disfigured, poorly drawn, bad anatomy, wrong anatomy, extra limb, missing limb, floating limbs, mutated hands and fingers, disconnected limbs, mutation, mutated, ugly, disgusting, blurry, amputation, NSFW", visible=True, ) seed = gr.Slider( label="Seed", minimum=0, maximum=MAX_SEED, step=1, value=0, ) steps = gr.Slider( label="Steps", minimum=0, maximum=15, step=1, value=4, ) number_image = gr.Slider( label="Number of Image", minimum=1, maximum=4, step=1, value=1, ) randomize_seed = gr.Checkbox(label="Randomize seed", value=True) with gr.Row(visible=True): width = gr.Slider( label="Width", minimum=256, maximum=MAX_IMAGE_SIZE, step=32, value=1024, ) height = gr.Slider( label="Height", minimum=256, maximum=MAX_IMAGE_SIZE, step=32, value=1024, ) with gr.Row(): guidance_scale = gr.Slider( label="Guidance Scale", minimum=0.1, maximum=10, step=0.1, value=2.0, ) gr.Examples( examples=examples, inputs=prompt, outputs=[result], fn=generate, cache_examples=CACHE_EXAMPLES, ) use_negative_prompt.change( fn=lambda x: gr.update(visible=x), inputs=use_negative_prompt, outputs=negative_prompt, api_name=False, ) gr.on( triggers=[ prompt.submit, negative_prompt.submit, run_button.click, ], fn=generate, inputs=[ prompt, negative_prompt, use_negative_prompt, seed, width, height, guidance_scale, randomize_seed, steps, number_image, ], outputs=[result], api_name="run", ) if __name__ == "__main__": demo.queue().launch()