Research map / Image generation / Text-to-image diffusion
LOTS of Fashion! Multi-Conditioning for Image Generation via Sketch-Text Pairing
Federico Girella, Davide Talon, Ziyue Liu, Zanxi Ruan, Yiming Wang, Marco Cristani
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 |
|---|---|---|---|
| use frozen DINOv2 sketch encoder instead of end-to-end trained encoder (ablation) | LOTS · Sketchy test | SSIM: .615 -> .678 | Improved |
| use DINOv2 instead of CLIP sketch encoder (ablation) | LOTS · Sketchy test | SSIM: .623 -> .678 | Improved |
| use DINOv2 instead of BLIP2 sketch encoder (ablation) | LOTS · Sketchy test | SSIM: .630 -> .678 | Improved |
| add Pair-Former pooling instead of no pooling (ablation) | LOTS · Sketchy test | SSIM: .623 -> .678 | Improved |
| add Pair-Former pooling instead of no pooling (ablation) | LOTS · Sketchy test | VQAScore: .724 -> .749 | Mixed |
| defer merging to diffusion instead of mean pooling pairs (ablation) | LOTS · Sketchy test | VQAScore: .676 -> .749 | Improved |
| defer merging to diffusion instead of mean pooling pairs (ablation) | LOTS · Sketchy test | SSIM: .659 -> .678 | Improved |
| use Pair-Former self-attention instead of cross-attention (ablation) | LOTS · Sketchy test | VQAScore: .681 -> .749 | Improved |
| use 32 pooling tokens instead of 64 (ablation) | LOTS · Sketchy test | VQAScore: .669 -> .749 | Improved |
| use localized sketch-text pairs (LOTS) instead of fine-tuned ControlNet | ControlNet · Sketchy test | GlobalCLIP: .645 -> .679 | Improved |
| use LOTS instead of fine-tuned IP-Adapter | IP-Adapter · Sketchy test | FID: 0.69 -> 0.79 | Hurt |
| use LOTS instead of Multi-T2I-Adapter | Multi-T2I-Adapter (zero-shot) · Sketchy test | SSIM: .723 -> .678 | Hurt |
Closest papers by meaning
- HieraFashDiff: Hierarchical Fashion Design with Multi-stage Diffusion Models (AAAI 2025 · similarity 0.89)
- TexFit: Text-Driven Fashion Image Editing with Diffusion Models (AAAI 2024 · similarity 0.88)
- Text to Sketch Generation with Multi-Styles (NeurIPS 2025 · similarity 0.88)
- DesignDiffusion: High-Quality Text-to-Design Image Generation with Diffusion Models (CVPR 2025 · similarity 0.87)
- Conditional Balance: Improving Multi-Conditioning Trade-Offs in Image Generation (CVPR 2025 · similarity 0.87)
- Style Injection in Diffusion: A Training-free Approach for Adapting Large-scale Diffusion Models for Style Transfer (CVPR 2024 · 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.
