Research map / Image generation / Text-to-image diffusion
Text-based Tactile Graphics Generation for the Visually Impaired
Ruihan Gao, Joonghyuk Shin, Ava Pun, Jaesik Park, Wenzhen Yuan, Jun-Yan Zhu
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 plate template conditioning (ablation) | Qwen-Image-Edit 4-step · Base geometry (50 prompts) | CLIP score: 34.82 -> 36.64 | Improved |
| add plate template conditioning (ablation) | Qwen-Image-Edit 40-step · Base geometry (50 prompts) | plate height SD: 0.205 -> 0.123 | Improved |
| add plate template conditioning (ablation) | Nano Banana Pro · Base geometry (50 prompts) | CLIP score: 36.83 -> 37.66 | Improved |
| add plate template conditioning (ablation) | Nano Banana Pro · Base geometry (50 prompts) | plate detection rate: 6/50 failures -> 100% | Improved |
| fine-tune Qwen-Image on texture dataset (4-step) | Qwen-Image · Text-to-texture (50 prompts) | patch self-similarity (LPIPS): 0.496 -> 0.412 | Improved |
| fine-tune Qwen-Image on texture dataset (4-step) | Qwen-Image · Text-to-texture (50 prompts) | HF ratio: 0.305 -> 0.396 | Improved |
| fine-tune Qwen-Image on texture dataset (40-step) | Qwen-Image · Text-to-texture (50 prompts) | CLIP score: 27.68 -> 28.43 | Improved |
| add center crop before tiling (ablation) | Resize baseline · Tiling (50 normal maps, 512x512) | border continuity: 29.80 -> 26.10 | Improved |
| add high-pass filtering after crop (ablation) | Crop · Tiling (50 normal maps, 512x512) | border continuity: 26.10 -> 22.84 | Improved |
| add intra-tile diffusion tiling after high-pass filtering (ablation) | Crop + HPF · Tiling (50 normal maps, 512x512) | border continuity: 22.84 -> 1.61 | Improved |
| use intra-tile instead of inter-tile tiling (ablation) | Ours tiling · Tiling (50 normal maps, 512x512) | gradient continuity: 14.42 -> 3.19 | Improved |
| use intra-tile tiling instead of Tiled Diffusion (SDXL) | Tiling · Tiling (50 normal maps, 512x512) | run time (s): 10.9 -> 3.18 | Improved |
Closest papers by meaning
- Tactile DreamFusion: Exploiting Tactile Sensing for 3D Generation (NeurIPS 2024 · similarity 0.89)
- TextToucher: Fine-Grained Text-to-Touch Generation (AAAI 2025 · similarity 0.88)
- PhysTacGen: Physics-Aware Visual-Tactile Sensor Image Generation (NeurIPS 2026 · similarity 0.87)
- Tactile-Augmented Radiance Fields (CVPR 2024 · similarity 0.87)
- A Touch, Vision, and Language Dataset for Multimodal Alignment (ICML 2024 · similarity 0.85)
- AnyTouch: Learning Unified Static-Dynamic Representation across Multiple Visuo-tactile Sensors (ICLR 2025 · similarity 0.83)
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.
