xiaozaa commited on
Commit
4fd56c0
1 Parent(s): 1beac4e

replace image

Browse files
README.md CHANGED
@@ -1,17 +1,19 @@
1
  # catvton-flux
2
 
3
  An advanced virtual try-on solution that combines the power of [CATVTON](https://arxiv.org/abs/2407.15886) (Contrastive Appearance and Topology Virtual Try-On) with Flux fill inpainting model for realistic and accurate clothing transfer.
 
4
 
5
  ## Showcase
6
- | Original | Result |
7
- |----------|--------|
8
- | ![Original](example/person/1.jpg) | ![Result](example/result/1.png) |
9
- | ![Original](example/person/00008_00.jpg) | ![Result](example/result/2.png) |
10
- | ![Original](example/person/00008_00.jpg) | ![Result](example/result/3.png) |
11
 
12
  ## Model Weights
 
 
13
  The model weights are trained on the [VITON-HD](https://github.com/shadow2496/VITON-HD) dataset.
14
- 🤗 [catvton-flux-alpha](https://huggingface.co/xiaozaa/catvton-flux-alpha)
15
 
16
  ## Prerequisites
17
  ```bash
@@ -39,9 +41,20 @@ python tryon_inference.py \
39
  ## Citation
40
 
41
  ```bibtex
42
- @misc{jiang2024catvton,
43
- title={CATVTON: A Contrastive Approach for Virtual Try-On Network},
44
- author={Chao Jiang and Xujie Zhang}
 
 
 
 
 
 
 
 
 
 
 
45
  }
46
  ```
47
 
 
1
  # catvton-flux
2
 
3
  An advanced virtual try-on solution that combines the power of [CATVTON](https://arxiv.org/abs/2407.15886) (Contrastive Appearance and Topology Virtual Try-On) with Flux fill inpainting model for realistic and accurate clothing transfer.
4
+ Also inspired by [In-Context LoRA](https://arxiv.org/abs/2410.23775) for prompt engineering.
5
 
6
  ## Showcase
7
+ | Original | Garment | Result |
8
+ |----------|---------|---------|
9
+ | ![Original](example/person/1.jpg) | ![Garment](example/garment/1.jpg) | ![Result](example/result/1.png) |
10
+ | ![Original](example/person/1.jpg) | ![Garment](example/garment/04564_00.jpg) | ![Result](example/result/2.png) |
11
+ | ![Original](example/person/00008_00.jpg) | ![Garment](example/garment/00034_00.jpg) | ![Result](example/result/3.png) |
12
 
13
  ## Model Weights
14
+ Hugging Face: 🤗 [catvton-flux-alpha](https://huggingface.co/xiaozaa/catvton-flux-alpha)
15
+
16
  The model weights are trained on the [VITON-HD](https://github.com/shadow2496/VITON-HD) dataset.
 
17
 
18
  ## Prerequisites
19
  ```bash
 
41
  ## Citation
42
 
43
  ```bibtex
44
+ @misc{chong2024catvtonconcatenationneedvirtual,
45
+ title={CatVTON: Concatenation Is All You Need for Virtual Try-On with Diffusion Models},
46
+ author={Zheng Chong and Xiao Dong and Haoxiang Li and Shiyue Zhang and Wenqing Zhang and Xujie Zhang and Hanqing Zhao and Xiaodan Liang},
47
+ year={2024},
48
+ eprint={2407.15886},
49
+ archivePrefix={arXiv},
50
+ primaryClass={cs.CV},
51
+ url={https://arxiv.org/abs/2407.15886},
52
+ }
53
+ @article{lhhuang2024iclora,
54
+ title={In-Context LoRA for Diffusion Transformers},
55
+ author={Huang, Lianghua and Wang, Wei and Wu, Zhi-Fan and Shi, Yupeng and Dou, Huanzhang and Liang, Chen and Feng, Yutong and Liu, Yu and Zhou, Jingren},
56
+ journal={arXiv preprint arxiv:2410.23775},
57
+ year={2024}
58
  }
59
  ```
60
 
example/result/2.png CHANGED
example/result/3.png CHANGED
flux_inpaint_garment.png ADDED
flux_inpaint_tryon.png ADDED
tryon_inference.py CHANGED
@@ -97,6 +97,8 @@ def main():
97
  parser.add_argument('--steps', type=int, default=50, help='Number of inference steps')
98
  parser.add_argument('--guidance-scale', type=float, default=30, help='Guidance scale')
99
  parser.add_argument('--seed', type=int, default=0, help='Random seed')
 
 
100
 
101
  args = parser.parse_args()
102
 
@@ -110,7 +112,8 @@ def main():
110
  output_tryon_path=args.output_tryon,
111
  num_steps=args.steps,
112
  guidance_scale=args.guidance_scale,
113
- seed=args.seed
 
114
  )
115
  print("Successfully saved garment and try-on images")
116
 
 
97
  parser.add_argument('--steps', type=int, default=50, help='Number of inference steps')
98
  parser.add_argument('--guidance-scale', type=float, default=30, help='Guidance scale')
99
  parser.add_argument('--seed', type=int, default=0, help='Random seed')
100
+ parser.add_argument('--width', type=int, default=768, help='Width')
101
+ parser.add_argument('--height', type=int, default=576, help='Height')
102
 
103
  args = parser.parse_args()
104
 
 
112
  output_tryon_path=args.output_tryon,
113
  num_steps=args.steps,
114
  guidance_scale=args.guidance_scale,
115
+ seed=args.seed,
116
+ size=(args.width, args.height)
117
  )
118
  print("Successfully saved garment and try-on images")
119