Research map / Image generation / Editing and style transfer
Neural Texture Synthesis With Guided Correspondence
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 |
|---|---|---|---|
| Use the Guided Correspondence loss instead of L2 distance (ablation) | CNNMRF texture optimization · Textures in Fig. 5 | Correspondence switches: 14.09±5.33 -> 6.83±4.12; 8.01±5.91 -> 3.93±3.93; 2.43±4.17 -> 1.32±3.21; 1.39±3.97 -> 0.62±2.14 | Helped |
| Add an occurrence penalty to the correspondence search (ablation) | CNNMRF texture optimization · Uncontrolled texture synthesis | Repetition and natural-looking quality | Helped |
| Increase the occurrence-penalty weight (ablation) | CNNMRF texture optimization · Uncontrolled texture synthesis | Visual naturalness | Helped |
| Use L2 distance instead of contextual similarity in the loss (ablation) | CNNMRF texture optimization · Textures in Fig. 5 | Correspondence stability: 6.83±4.12 -> 14.09±5.33; 3.93±3.93 -> 8.01±5.91; 1.32±3.21 -> 2.43±4.17; 0.62±2.14 -> 1.39±3.97 | Hurt |
| Add Guided Correspondence loss to the conditional GAN loss (ablation) | SPADE · Real-time progression control | Texture and control quality | Helped |
| Remove the discriminator and use Guided Correspondence loss only (ablation) | SPADE · Real-time progression control | Fine-detail quality | Hurt |
| Remove Guided Correspondence loss from the conditional GAN objective (ablation) | SPADE · Real-time progression control | Correct texture-pattern generation | Hurt |
| Replace Guided Correspondence loss with Sliced Wasserstein loss (ablation) | SPADE · Real-time controlled synthesis | Synthesis quality | Hurt |
| Use Self-tuning instead of Ours | Uncontrolled texture synthesis · 50 textures | Average Color Distance: 9.40 -> 2.65 | Helped |
| Use Ours instead of CNNMRF | Uncontrolled texture synthesis · 50 textures | Average Color Distance: 23.16 -> 9.40 | Helped |
| Use Ours instead of CNNMRF | Uncontrolled texture synthesis · 50 textures | User preference: 33.1% preference for CNNMRF | Helped |
| Use Ours instead of SWD | Uncontrolled texture synthesis · 50 textures | Average Color Distance: 24.21 -> 9.40 | Helped |
Closest papers by meaning
- Combining Markov Random Fields and Convolutional Neural Networks for Image Synthesis (CVPR 2016 · similarity 0.90)
- Diversified Texture Synthesis With Feed-Forward Networks (CVPR 2017 · similarity 0.89)
- TextureGAN: Controlling Deep Image Synthesis With Texture Patches (CVPR 2018 · similarity 0.88)
- TextureDreamer: Image-Guided Texture Synthesis Through Geometry-Aware Diffusion (CVPR 2024 · similarity 0.86)
- Single Mesh Diffusion Models with Field Latents for Texture Generation (CVPR 2024 · similarity 0.85)
- One-Shot Reference-based Structure-Aware Image to Sketch Synthesis (AAAI 2025 · similarity 0.85)
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.