CosmicMan-SD

CosmicMan is a text-to-image foundation model specialized for generating high-fidelity human images. For more information, please refer to our research paper: CosmicMan: A Text-to-Image Foundation Model for Humans. Our model is based on runwayml/stable-diffusion-v1-5. This repository provide UNet checkpoints for CosmicMan-SD.

Requirements

conda create -n cosmicman python=3.10
source activate cosmicman
pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
pip install accelerate diffusers datasets transformers botocore invisible-watermark bitsandbytes gradio==3.48.0

Inference

import torch
from diffusers import StableDiffusionPipeline, UNet2DConditionModel, EulerDiscreteScheduler
from huggingface_hub import hf_hub_download
from safetensors.torch import load_file

base_path = "runwayml/stable-diffusion-v1-5"
unet_path = "cosmicman/CosmicMan-SD"

# Load model.
unet = UNet2DConditionModel.from_pretrained(unet_path, torch_dtype=torch.float16)
pipe = StableDiffusionPipeline.from_pretrained(base_path, unet=unet, torch_dtype=torch.float16, variant="fp16").to("cuda")
pipe.scheduler = EulerDiscreteScheduler.from_pretrained(base_path, subfolder="scheduler", torch_dtype=torch.float16)

 
# Generate image.
positive_prompt = "A closeup portrait shot against a white wall, a fit Caucasian adult female with wavy blonde hair falling above her chest wears a short sleeve silk floral dress and a floral silk normal short sleeve white blouse"
negative_prompt = ""
image = pipe(positive_prompt, num_inference_steps=30, 
        guidance_scale=7.5, height=1024, 
        width=1024, negative_prompt=negative_prompt, output_type="pil").images[0].save("output.png")

Citation Information

@article{li2024cosmicman,
  title={CosmicMan: A Text-to-Image Foundation Model for Humans},
  author={Li, Shikai and Fu, Jianglin and Liu, Kaiyuan and Wang, Wentao and Lin, Kwan-Yee and Wu, Wayne},
  journal={arXiv preprint arXiv:2404.01294},
  year={2024}
}