Research map / Image generation / Editing and style transfer
LayerD: Decomposing Raster Graphic Designs into Layers
Tomoyuki Suzuki, Kang-Jun Liu, Naoto Inoue, Kota Yamaguchi
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 inverse-blending foreground color estimation instead of naive alpha replacement (ablation) | LayerD · Crello test | RGB L1 | Improved |
| add palette-based background refinement (ablation) | LayerD · Crello test | RGB L1 and Alpha IoU | Improved |
| add palette-based foreground refinement (ablation) | LayerD · Crello test | Alpha IoU | Improved |
| fine-tune the matting model on top-layer matting instead of using BiRefNet without additional training (ablation) | LayerD · Crello test | RGB L1 and Alpha IoU | Improved |
| train with text layers included, evaluated on designs without text (ablation) | LayerD · Crello test (without text layers) | RGB L1 and Alpha IoU | Improved |
| use LayerD alone instead of replacing the first iteration with Hi-SAM text segmentation (ablation) | LayerD · Crello test (all layers) | RGB L1 and Alpha IoU | Improved |
| use larger matting backbone (Swin-L over Swin-T, PVT) (ablation) | LayerD · Crello test | RGB L1 and Alpha IoU | Improved |
| replace LaMa inpainting with FLUX.1 Fill [dev] generative inpainting (ablation) | LayerD · Crello test | RGB L1 and Alpha IoU | Hurt |
| use LayerD instead of the YOLO and VLM baselines | layer decomposition · User study | average score: 3.52 -> 3.74 | Improved |
Later papers that compared with this method
3 later accepted papers report LayerD as a baseline in their results tables; 2 of them report a better number on most of the results they share with it.
- ReDesign: Recovering Editable Design Structures from Raster Images via Agentic Decomposition (ECCV 2026 · better on 5 of 5 shared results)
- PixelART: Image-to-Layer Decomposition without Latents or Text-to-Image Pretraining (NeurIPS 2026 · better on 2 of 4 shared results)
- Qwen-Image-Layered: Towards Inherent Editability via Layer Decomposition (CVPR 2026 · better on 4 of 4 shared results)
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
- ReDesign: Recovering Editable Design Structures from Raster Images via Agentic Decomposition (ECCV 2026 · similarity 0.91)
- Decomposition of Graphic Design with Unified Multimodal Model (ICML 2025 · similarity 0.89)
- Illustrator's Depth: Monocular Layer Index Prediction for Image Decomposition (CVPR 2026 · similarity 0.88)
- Generative Image Layer Decomposition with Visual Effects (CVPR 2025 · similarity 0.88)
- PixelART: Image-to-Layer Decomposition without Latents or Text-to-Image Pretraining (NeurIPS 2026 · similarity 0.86)
- LayerCraft: Enhancing Text-to-Image Generation with CoT Reasoning and Layered Object Integration (NeurIPS 2025 · similarity 0.84)
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.
