Research map / Image generation / Text-to-image diffusion
Multimodal Latent Diffusion Model for Complex Sewing Pattern Generation
Shengqi Liu, Yuhao Cheng, Zhuo Chen, Xingyu Ren, Wenhan Zhu, Lincheng Li, Mengxiao Bi, Xiaokang Yang, Yichao Yan
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 |
|---|---|---|---|
| compress sewing pattern latent to 256*6 with n=2 instead of 256*8 n=2 (ablation) | SewingLDM · Latent compression ablation | clothes-to-body distance (cm): 2.87 cm -> 2.20 cm | Improved |
| compress sewing pattern latent to 256*6 with n=2 instead of 256*8 n=2 (ablation) | SewingLDM · Latent compression ablation | codebook usage (%): 91% -> 100% | Improved |
| add compact quantized latent (256*8, n=2) instead of no compression (ablation) | SewingLDM · Latent compression ablation | generation ability | Improved |
| use more compressed 256*4 latent with n=2 (ablation) | SewingLDM · Latent compression ablation | reconstruction | Hurt |
| use larger quantization n=32 or n=8 with 256*32 or 256*12 latents (ablation) | SewingLDM · Latent compression ablation | codebook usage (%): 0% | Hurt |
| fine-tune output layers of both self- and cross-attention instead of only one (ablation) | SewingLDM · multi-modal conditioning (qualitative) | sketch alignment | Improved |
| inject body and sketch condition at shallower layers (after block 0) rather than block 5, 10, 15, 20 (ablation) | SewingLDM · multi-modal conditioning (qualitative) | garment completeness | Improved |
| use two-step training instead of one-step training (ablation) | SewingLDM · sketch-only generation (qualitative) | garment fidelity | Improved |
| use extended sewing pattern representation instead of original (ablation) | SewingLDM · representation ablation (qualitative) | garment diversity | Improved |
| use SewingLDM instead of DressCode | sewing pattern generation · Sewing pattern generation comparison | clothes-to-body distance (cm): 3.69 cm -> 2.20 cm | Improved |
| use SewingLDM instead of Sewformer | sewing pattern generation · Sewing pattern generation comparison | user value (1-5): 2.10 -> 4.60 | Improved |
| use SewingLDM instead of fine-tuned SewFormer | sewing pattern generation · GarmentCode test set reconstruction | Stitch Acc: 2.8 -> 84.2 | Improved |
Closest papers by meaning
- Design2GarmentCode: Turning Design Concepts to Tangible Garments Through Program Synthesis (CVPR 2025 · similarity 0.90)
- AIpparel: A Multimodal Foundation Model for Digital Garments (CVPR 2025 · similarity 0.90)
- AnyDressing: Customizable Multi-Garment Virtual Dressing via Latent Diffusion Models (CVPR 2025 · similarity 0.89)
- DreamFit: Garment-Centric Human Generation via a Lightweight Anything-Dressing Encoder (AAAI 2025 · similarity 0.88)
- HieraFashDiff: Hierarchical Fashion Design with Multi-stage Diffusion Models (AAAI 2025 · similarity 0.88)
- Cascaded Diffusion Models for Virtual Try-On: Improving Control and Resolution (AAAI 2025 · similarity 0.87)
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.
