Research map / Image generation / Autoregressive and latent generation
Diversified Texture Synthesis With Feed-Forward Networks
Check 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 |
|---|---|---|---|
| subtract the feature mean before computing Gram statistics (ablation) | multi-texture synthesis network · 3-texture synthesis | synthesis quality | Helped |
| use the original, non-centered Gram matrix (ablation) | multi-texture synthesis network · 3-texture synthesis | synthesis quality | Hurt |
| add a feature-space diversity loss between samples sharing a texture (ablation) | multi-texture synthesis network · single-texture output generation | output diversity | Helped |
| remove the diversity loss (ablation) | multi-texture synthesis network · single-texture output generation | output diversity | Hurt |
| introduce textures incrementally and retain earlier textures in subsequent sampling (ablation) | 60-texture synthesis network · 60-texture synthesis | texture loss | Helped |
| sample target textures randomly throughout training (ablation) | 60-texture synthesis network · 60-texture synthesis | synthesis quality | Hurt |
| add selector guidance at each upsampling scale (ablation) | 60-texture synthesis network · 60-texture synthesis | training convergence | Helped |
| remove the selector network (ablation) | 60-texture synthesis network · 60-texture synthesis | training convergence | Hurt |
| use one feed-forward network for many textures | Diversified Texture Synthesis · DTD, 300-texture synthesis | texture synthesis quality | Helped |
| use the proposed multi-style transfer network | 16-style transfer · test content images | style transfer quality | Mixed |
Closest papers by meaning
- Texture Networks: Feed-forward Synthesis of Textures and Stylized Images (ICML 2016 · similarity 0.94)
- Fast Texture Synthesis via Pseudo Optimizer (CVPR 2020 · similarity 0.93)
- Improved Texture Networks: Maximizing Quality and Diversity in Feed-Forward Stylization and Texture Synthesis (CVPR 2017 · similarity 0.92)
- TextureDreamer: Image-Guided Texture Synthesis Through Geometry-Aware Diffusion (CVPR 2024 · similarity 0.86)
- Paint-it: Text-to-Texture Synthesis via Deep Convolutional Texture Map Optimization and Physically-Based Rendering (CVPR 2024 · similarity 0.86)
- Single Mesh Diffusion Models with Field Latents for Texture Generation (CVPR 2024 · similarity 0.86)
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.