Research map / Image generation / Editing and style transfer
Time-Efficient and Identity-Consistent Virtual Try-On Using A Variant of Altered Diffusion Models
Phuong Hoang Dam, Jihoon Jeong, Anh T Tran, Daeyoung Kim
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| remove DensePose condition (ablation) | FIP-VITON try-on module · VITON-HD 512x384 | FID: 24.49 -> 8.43 | Improved |
| add generated noise condition image instead of Gaussian noise (ablation) | FIP-VITON · VITON-HD 512x384 | FID: 11.00 -> 8.43 | Improved |
| add CLIP global condition (ablation) | FIP-VITON · VITON-HD 512x384 | FID: 9.11 -> 8.43 | Improved |
| add unconditional post-processing block (ablation) | FIP-VITON · VITON-HD 512x384 | FID: 8.52 -> 8.43 | No clear effect |
| add cross-attention to the warping module (ablation) | FIP-VITON · VITON-HD 512x384 | FID: 8.52 -> 8.43 (parameters 128.4M -> 168.5M) | Improved |
| add mask-aware conditional post-processing (ablation) | FIP-VITON · VITON-HD 512x384 | LPIPS: 0.0706 -> 0.0675 | Improved |
| plug in mask-aware post-processing to HR-VTON | HR-VTON · VITON-HD | LPIPS: 0.1106 -> 0.1068 | Improved |
| use noise level alpha_n=5 instead of 2 or 7 (ablation) | FIP-VITON · VITON-HD 512x384 | FID: 9.61 (2), 8.74 (7) -> 8.43 (5) | Improved |
| use single-step try-on module instead of 1000-step DDIM diffusion (ablation) | FIP-VITON · VITON-HD 512x384 | FID and inference time: FID 13.93 -> 8.43; time 460 -> 1.01 s | Improved |
| lower overlap threshold for post-processing from 0.80 to 0.75 (ablation) | mask-aware post-processing · VITON-HD 512x384 | FID: 8.428 -> 8.423 | No clear effect |
| FIP-VITON versus DCI-VTON on paired LPIPS | FIP-VITON · VITON-HD 512x384 | LPIPS: 0.043 (DCI-VTON) vs 0.067 | Hurt |
Closest papers by meaning
- M&M VTO: Multi-Garment Virtual Try-On and Editing (CVPR 2024 · similarity 0.93)
- Texture-Preserving Diffusion Models for High-Fidelity Virtual Try-On (CVPR 2024 · similarity 0.92)
- Cascaded Diffusion Models for Virtual Try-On: Improving Control and Resolution (AAAI 2025 · similarity 0.91)
- Shining Yourself: High-Fidelity Ornaments Virtual Try-on with Diffusion Model (CVPR 2025 · similarity 0.89)
- Virtual Try-On: Real-Time Interactive Hybrid Network with High-Fidelity (AAAI 2024 · similarity 0.89)
- Incorporating Visual Correspondence into Diffusion Model for Virtual Try-On (ICLR 2025 · similarity 0.89)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
