VTON 360: High-Fidelity Virtual Try-On from Any Viewing Direction
Paper • 2503.12165 • Published
Image-to-Image
Diffusers
Safetensors
virtual-try-on
image-editing
multi-view
human
vton360
cvpr-2025
import torch
from diffusers import DiffusionPipeline
from diffusers.utils import load_image
# switch to "mps" for apple devices
pipe = DiffusionPipeline.from_pretrained("nielsr/VTON360-THuman2.0", dtype=torch.bfloat16, device_map="cuda")
prompt = "Turn this cat into a dog"
input_image = load_image("https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/cat.png")
image = pipe(image=input_image, prompt=prompt).images[0]VTON360 checkpoint for THuman2.0
This repository contains the checkpoint-30000 model bundle for the
THuman2.0 experiment released with VTON 360: High-Fidelity Virtual Try-On
from Any Viewing Direction (CVPR 2025).
Use it with the custom implementation in the
official VTON360 repository. The
repository README instructs users to place this checkpoint under
src/multiview_consist_edit/checkpoints and run
src/multiview_consist_edit/infer_tryon_multi.py.
from huggingface_hub import snapshot_download
snapshot_download(
repo_id="nielsr/VTON360-THuman2.0",
local_dir="src/multiview_consist_edit/checkpoints/thuman_tryon_mvattn_multi",
)
The released configuration also expects:
Provenance and attribution
- Official code: https://github.com/scnuhealthy/VTON360
- Project page: https://scnuhealthy.github.io/VTON360/
- Paper: https://huggingface.co/papers/2503.12165
- arXiv: https://arxiv.org/abs/2503.12165
- Original Baidu release: https://pan.baidu.com/s/1SJH3QI30UKihOaU9owta5Q (code
32h3) - Community Hub source used for this mirror: https://huggingface.co/NTUST-DDRC/thuman_tryon_mvattn_multi
The checkpoint is a derivative of Stable Diffusion v1.5 and retains the CreativeML Open RAIL-M license. Users must also follow the licenses and terms of VTON360, its code dependencies, and the underlying THuman data.
- Downloads last month
- -
Model tree for nielsr/VTON360-THuman2.0
Base model
stable-diffusion-v1-5/stable-diffusion-v1-5