Spaces:
Running
on
A10G
Running
on
A10G
| from diffusers import DiffusionPipeline, AutoencoderTiny | |
| from compel import Compel | |
| import torch | |
| try: | |
| import intel_extension_for_pytorch as ipex # type: ignore | |
| except: | |
| pass | |
| import psutil | |
| from config import Args | |
| from pydantic import BaseModel, Field | |
| from PIL import Image | |
| base_model = "SimianLuo/LCM_Dreamshaper_v7" | |
| taesd_model = "madebyollin/taesd" | |
| default_prompt = "Portrait of The Terminator with , glare pose, detailed, intricate, full of colour, cinematic lighting, trending on artstation, 8k, hyperrealistic, focused, extreme details, unreal engine 5 cinematic, masterpiece" | |
| page_content = """<h1 class="text-3xl font-bold">Real-Time Latent Consistency Model</h1> | |
| <h3 class="text-xl font-bold">Text-to-Image</h3> | |
| <p class="text-sm"> | |
| This demo showcases | |
| <a | |
| href="https://huggingface.co/SimianLuo/LCM_Dreamshaper_v7" | |
| target="_blank" | |
| class="text-blue-500 underline hover:no-underline">LCM</a> | |
| Image to Image pipeline using | |
| <a | |
| href="https://huggingface.co/docs/diffusers/main/en/using-diffusers/lcm#performing-inference-with-lcm" | |
| target="_blank" | |
| class="text-blue-500 underline hover:no-underline">Diffusers</a> with a MJPEG stream server | |
| </p> | |
| <p class="text-sm text-gray-500"> | |
| Change the prompt to generate different images, accepts <a | |
| href="https://github.com/damian0815/compel/blob/main/doc/syntax.md" | |
| target="_blank" | |
| class="text-blue-500 underline hover:no-underline">Compel</a | |
| > syntax. | |
| </p>""" | |
| class Pipeline: | |
| class Info(BaseModel): | |
| name: str = "txt2img" | |
| title: str = "Text-to-Image LCM" | |
| description: str = "Generates an image from a text prompt" | |
| input_mode: str = "text" | |
| page_content: str = page_content | |
| class InputParams(BaseModel): | |
| prompt: str = Field( | |
| default_prompt, | |
| title="Prompt", | |
| field="textarea", | |
| id="prompt", | |
| ) | |
| seed: int = Field( | |
| 2159232, min=0, title="Seed", field="seed", hide=True, id="seed" | |
| ) | |
| steps: int = Field( | |
| 4, min=2, max=15, title="Steps", field="range", hide=True, id="steps" | |
| ) | |
| width: int = Field( | |
| 768, min=2, max=15, title="Width", disabled=True, hide=True, id="width" | |
| ) | |
| height: int = Field( | |
| 768, min=2, max=15, title="Height", disabled=True, hide=True, id="height" | |
| ) | |
| guidance_scale: float = Field( | |
| 8.0, | |
| min=1, | |
| max=30, | |
| step=0.001, | |
| title="Guidance Scale", | |
| field="range", | |
| hide=True, | |
| id="guidance_scale", | |
| ) | |
| def __init__(self, args: Args, device: torch.device, torch_dtype: torch.dtype): | |
| if args.safety_checker: | |
| self.pipe = DiffusionPipeline.from_pretrained(base_model) | |
| else: | |
| self.pipe = DiffusionPipeline.from_pretrained( | |
| base_model, safety_checker=None | |
| ) | |
| if args.taesd: | |
| self.pipe.vae = AutoencoderTiny.from_pretrained( | |
| taesd_model, torch_dtype=torch_dtype, use_safetensors=True | |
| ).to(device) | |
| if args.sfast: | |
| from sfast.compilers.stable_diffusion_pipeline_compiler import ( | |
| compile, | |
| CompilationConfig, | |
| ) | |
| config = CompilationConfig.Default() | |
| config.enable_xformers = True | |
| config.enable_triton = True | |
| config.enable_cuda_graph = True | |
| self.pipe = compile(self.pipe, config=config) | |
| self.pipe.set_progress_bar_config(disable=True) | |
| self.pipe.to(device=device, dtype=torch_dtype) | |
| if device.type != "mps": | |
| self.pipe.unet.to(memory_format=torch.channels_last) | |
| if args.torch_compile: | |
| self.pipe.unet = torch.compile( | |
| self.pipe.unet, mode="reduce-overhead", fullgraph=True | |
| ) | |
| self.pipe.vae = torch.compile( | |
| self.pipe.vae, mode="reduce-overhead", fullgraph=True | |
| ) | |
| self.pipe(prompt="warmup", num_inference_steps=1, guidance_scale=8.0) | |
| if args.compel: | |
| self.compel_proc = Compel( | |
| tokenizer=self.pipe.tokenizer, | |
| text_encoder=self.pipe.text_encoder, | |
| truncate_long_prompts=False, | |
| ) | |
| def predict(self, params: "Pipeline.InputParams") -> Image.Image: | |
| generator = torch.manual_seed(params.seed) | |
| prompt_embeds = None | |
| prompt = params.prompt | |
| if hasattr(self, "compel_proc"): | |
| prompt_embeds = self.compel_proc(params.prompt) | |
| prompt = None | |
| results = self.pipe( | |
| prompt_embeds=prompt_embeds, | |
| prompt=prompt, | |
| generator=generator, | |
| num_inference_steps=params.steps, | |
| guidance_scale=params.guidance_scale, | |
| width=params.width, | |
| height=params.height, | |
| output_type="pil", | |
| ) | |
| nsfw_content_detected = ( | |
| results.nsfw_content_detected[0] | |
| if "nsfw_content_detected" in results | |
| else False | |
| ) | |
| if nsfw_content_detected: | |
| return None | |
| return results.images[0] | |