Research map / Image generation / Editing and style transfer
Improved Texture Networks: Maximizing Quality and Diversity in Feed-Forward Stylization and Texture Synthesis
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 |
|---|---|---|---|
| replace batch normalization with instance normalization (ablation) | StyleNet · Image stylization | stylization quality | Helped |
| replace batch normalization with instance normalization (ablation) | StyleNet · Image stylization | training convergence | Helped |
| replace batch normalization with instance normalization (ablation) | texture synthesis networks · Texture synthesis | generation quality | No clear effect |
| add the entropy-based diversity term (ablation) | TextureNetV2 · Texture synthesis | sample diversity | Helped |
| add the entropy-based diversity term (ablation) | StyleNetV2 · Image stylization | stylization diversity | Helped |
| use a high-capacity architecture without the diversity term (ablation) | TextureNetV2 · Texture synthesis | sample diversity | Hurt |
| use a low-capacity architecture to induce diversity instead of adding the diversity term (ablation) | TextureNetV1 · Texture synthesis | visual fidelity | Hurt |
| increase the diversity-term weight too much (ablation) | texture and stylization networks · Texture synthesis and image stylization | visual quality | Hurt |
| use instance normalization rather than batch normalization (ablation) | AlexNet · ILSVRC | top-1 accuracy: 2-3% worse | Hurt |
Closest papers by meaning
- Diversified Texture Synthesis With Feed-Forward Networks (CVPR 2017 · similarity 0.92)
- Texture Networks: Feed-forward Synthesis of Textures and Stylized Images (ICML 2016 · similarity 0.90)
- Fast Texture Synthesis via Pseudo Optimizer (CVPR 2020 · similarity 0.88)
- Two-Stream Convolutional Networks for Dynamic Texture Synthesis (CVPR 2018 · similarity 0.87)
- Free-Lunch Color-Texture Disentanglement for Stylized Image Generation (NeurIPS 2025 · similarity 0.85)
- Conditional Balance: Improving Multi-Conditioning Trade-Offs in Image Generation (CVPR 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.