Research map / Image generation / Text-to-image diffusion
OrthoTailor: Geometric Orthogonalization for Conflict-Free Unified Fashion Generation
Zhaotong Yang, Ying Tai, Jiahui Zhan, Yu Zheng, Jianjun Qian, Jian Yang
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 Orthogonal Subspace Projection (task-specific orthogonal rotations) to shared LoRA (ablation) | joint LoRA on LongCat-Image-Edit · DeepFashion pose transfer | FID: 18.703 -> 6.632 | Improved |
| add OSP to shared LoRA (ablation) | joint LoRA on LongCat-Image-Edit · VITON-HD garment reconstruction | FID: 16.157 -> 9.787 | Improved |
| add OSP to shared LoRA (ablation) | joint LoRA on LongCat-Image-Edit · VITON-HD virtual try-on | FID: 10.587 -> 8.386 | Improved |
| naive sharing of a single LoRA across tasks versus task-specific LoRAs (ablation) | LongCat-Image-Edit · DeepFashion pose transfer | FID: 7.445 -> 18.703 | Hurt |
| use orthogonal rotations instead of non-orthogonal random matrices in the bottleneck (ablation) | OSP · VITON-HD garment reconstruction | FID: 20.412 -> 9.787 | Improved |
| use orthogonal rotations instead of non-orthogonal random matrices in the bottleneck (ablation) | OSP · DeepFashion pose transfer | FID: 15.423 -> 6.632 | Improved |
| add Fisher-guided Negative Guidance (highest-similarity task) on top of OSP (ablation) | OSP · DeepFashion pose transfer | FID: 6.632 -> 6.364 | Improved |
| add FNG on top of OSP (ablation) | OSP · VITON-HD virtual try-on | LPIPS: 0.084 -> 0.064 | Improved |
| choose highest-Fisher-similarity task as negative instead of lowest (ablation) | FNG · DeepFashion pose transfer | FID: 7.318 -> 6.364 | Improved |
| use lowest-Fisher-similarity task as negative (FNG*) on top of OSP (ablation) | OSP · DeepFashion pose transfer | FID: 6.632 -> 7.318 | Hurt |
| add OSP to Any2AnyTryon | Any2AnyTryon · VITON-HD virtual try-on | LPIPS: 0.077 -> 0.058 | Improved |
| add OSP to Any2AnyTryon | Any2AnyTryon · DeepFashion pose transfer | FID: 12.250 -> 11.609 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- Any2AnyTryon — Any2AnyTryon: Leveraging Adaptive Position Embeddings for Versatile Virtual Clothing Tasks (ICCV 2025) · better on 4 of 4 shared results
- GP-VTON — GP-VTON: Towards General Purpose Virtual Try-On via Collaborative Local-Flow Global-Parsing Learning (CVPR 2023) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- GarmentGPT: Compositional Garment Pattern Generation via Discrete Latent Tokenization (ICLR 2026 · similarity 0.85)
- ChatGarment: Garment Estimation, Generation and Editing via Large Language Models (CVPR 2025 · similarity 0.84)
- Easy Regional Contrastive Learning of Expressive Fashion Representations (NeurIPS 2024 · similarity 0.83)
- IMAGDressing-v1: Customizable Virtual Dressing (AAAI 2025 · similarity 0.83)
- Enhancing Virtual Try-On with Synthetic Pairs and Error-Aware Noise Scheduling (CVPR 2025 · similarity 0.83)
- AnyFit: Controllable Virtual Try-on for Any Combination of Attire Across Any Scenario (NeurIPS 2024 · similarity 0.83)
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.
