Research map / Image generation / Autoregressive and latent generation
L-CAD: Language-based Colorization with Any-level Descriptions using Diffusion Priors
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 luminance-guided image compression (luminance encoder) (ablation) | L-CAD · Extended COCO-Stuff | PSNR: 23.87 -> 25.97 | Helped |
| add luminance-guided image compression (ablation) | L-CAD · Extended COCO-Stuff | LPIPS: .190 -> .142 | Helped |
| add semantic-aligned latent representation (channel-extended convolution) (ablation) | L-CAD · Extended COCO-Stuff | PSNR: 17.26 -> 25.97 | Helped |
| add semantic-aligned latent representation (ablation) | L-CAD · Multi-instance | PSNR: 15.61 -> 25.51 | Helped |
| add instance-aware sampling strategy instead of standard DDIM (ablation) | L-CAD · Extended COCO-Stuff | PSNR: 25.32 -> 25.97 | Helped |
| add instance-aware sampling strategy (ablation) | L-CAD · Multi-instance | LPIPS: .149 -> .127 | Helped |
| add instance-aware sampling strategy (ablation) | L-CAD (complete/partial) · Extended COCO-Stuff | SSIM: .9174 -> .9191 | Helped |
| add luminance-guided compression (ablation) | L-CAD · Multi-instance | PSNR: 23.96 -> 25.51 | Helped |
| add instance-aware sampling strategy (ablation) | L-CAD* · Extended COCO-Stuff (scarce) | PSNR: 24.24 -> 24.33 | Helped |
| swap in L-CAD with scarce descriptions for CT^2 baseline | L-CAD* · ImageNet 5K val | colorfulness | Mixed |
Closest papers by meaning
- Automatic Controllable Colorization via Imagination (CVPR 2024 · similarity 0.86)
- ColorConceptBench: A Benchmark for Probabilistic Color-Concept Understanding in Text-to-Image Models (NeurIPS 2026 · similarity 0.85)
- Color Alignment in Diffusion (CVPR 2025 · similarity 0.85)
- MangaNinja: Line Art Colorization with Precise Reference Following (CVPR 2025 · similarity 0.85)
- Color Conditional Generation with Sliced Wasserstein Guidance (NeurIPS 2025 · similarity 0.84)
- MegaCoin: Enhancing Medium-Grained Color Perception for Vision-Language Models (AAAI 2026 · 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.