Research map / Image generation / Text-to-image diffusion
Show, Don't Tell: Morphing Latent Reasoning into Image Generation
Harold Haodong Chen, Xinxiang Yin, Wen-Jie Shu, Hongfei Zhang, Zixin Zhang, Chenfei Liao, Litao Guo, Qifeng Chen, Ying-Cong Chen
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 LatentMorph latent reasoning | Janus-Pro · GenEval | overall: 0.80 -> 0.96 | Improved |
| add LatentMorph latent reasoning | Janus-Pro · T2I-CompBench | overall: 39.21 -> 64.53 | Improved |
| add LatentMorph latent reasoning instead of TwiG-RL explicit reasoning | TwiG-RL · T2I-CompBench Non-Spatial | score: 31.99 -> 39.27 (7.28%) | Improved |
| add LatentMorph latent reasoning instead of TwiG-RL explicit reasoning | TwiG-RL · T2I-CompBench++ 3D-Spatial | score: 38.87 -> 47.00 (8.13%) | Improved |
| add GRPO fine-tuning without reasoning | Janus-Pro · GenEval | overall: 0.80 -> 0.82 | Improved |
| add SFT fine-tuning without reasoning | Janus-Pro · GenEval | overall: 0.80 -> 0.79 | No clear effect |
| keep latent thought in continuous space instead of decode-re-encode into text (w/o latent removed) (ablation) | LatentMorph · IPV-Txt | Impossible Prompt Following | Improved |
| use learned adaptive invoker instead of random or fixed-step injection (ablation) | LatentMorph · Ablation on invoker (Appendix) | benchmark score | Improved |
| increase short-term memory length n_s from 2 to 16 (ablation) | LatentMorph · Condenser sensitivity | generation quality | Improved |
| use adaptive invocation instead of fixed-step reasoning | LatentMorph · T2I-CompBench efficiency | inference time reduction: 44.3% | Improved |
Closest papers by meaning
- Leveraging Latent Visual Reasoning in Silence (NeurIPS 2026 · similarity 0.92)
- Imagination Helps Visual Reasoning, But Not Yet in Latent Space (ICML 2026 · similarity 0.92)
- SLVR: Structured Latent Visual Reasoning via Human-like Reasoning Flows (NeurIPS 2026 · similarity 0.91)
- Machine Mental Imagery: Empower Multimodal Reasoning with Latent Visual Tokens (CVPR 2026 · similarity 0.91)
- Thinking-while-Generating: Interleaving Textual Reasoning throughout Visual Generation (CVPR 2026 · similarity 0.91)
- ThinkMorph: Emergent Properties in Multimodal Interleaved Chain-of-Thought Reasoning (ICLR 2026 · similarity 0.90)
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.
