Research map / Image generation / Image restoration
TryOn-Refiner: Conditional Rectified-flow-based TryOn Refiner for More Accurate Detail Reconstruction
Wen Qian
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 |
|---|---|---|---|
| add TryOn-Refiner | OOTD · VITON-HD | LPIPS: 0.071 -> 0.069 | Improved |
| add TryOn-Refiner | OOTD · VITON-HD | FID: 8.81 -> 8.60 | Improved |
| add TryOn-Refiner | StableVITON · VITON-HD | LPIPS: 0.084 -> 0.070 | Improved |
| add TryOn-Refiner | SDXL-TryOn · VITON-HD | FID: 9.07 -> 8.75 | Improved |
| add TryOn-Refiner | OOTD · Dresscode | FID: 4.20 -> 4.11 | Improved |
| use noise-start diffusion refiner (Diff-Refiner, 30 steps) instead of coarse-image-start rectified flow | StableVITON · VITON-HD | LPIPS: 0.084 -> 0.092 | Hurt |
| increase inference steps from 1 to 10 (ablation) | TryOn-Refiner · VITON-HD | SSIM: 0.882 -> 0.886 | Improved |
| increase inference steps from 10 to 30 (ablation) | TryOn-Refiner · VITON-HD | LPIPS: 0.069 -> 0.069 | No clear effect |
| increase resolution from 1024x768 to 1600x1200 (ablation) | TryOn-Refiner · VITON-HD | LPIPS: 0.069 -> 0.066 | Improved |
| increase resolution from 512x384 to 1024x768 (ablation) | TryOn-Refiner · VITON-HD | LPIPS: 0.073 -> 0.069 | Improved |
| add Gaussian low-pass filtering of targets (ablation) | TryOn-Refiner · VITON-HD | FID: 9.21 -> 8.75 | Improved |
| add VAE decoder finetuning on top of Gaussian filtering (ablation) | TryOn-Refiner · VITON-HD | FID: 8.75 -> 8.60 | Improved |
Closest papers by meaning
- Cascaded Diffusion Models for Virtual Try-On: Improving Control and Resolution (AAAI 2025 · similarity 0.86)
- Balanced Conic Rectified Flow (NeurIPS 2025 · similarity 0.85)
- OpenVTON-Bench: A Large-Scale High-Resolution Benchmark for Controllable Virtual Try-On Evaluation (NeurIPS 2026 · similarity 0.85)
- Latent Diffusion-Enhanced Virtual Try-On via Optimized Pseudo-Label Generation (AAAI 2025 · similarity 0.85)
- SCoT: Unifying Consistency Models and Rectified Flows via Straight-Consistent Trajectories (NeurIPS 2025 · similarity 0.85)
- HiFlow: Training-free High-Resolution Image Generation with Flow-Aligned Guidance (NeurIPS 2025 · similarity 0.85)
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.
