Research map / Image generation / Editing and style transfer
Two-Stream Convolutional Networks for Dynamic Texture Synthesis
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 concat-layer activations rather than the predicted-flow output for dynamics statistics (ablation) | Two-stream dynamic texture synthesis · Dynamic texture user study | real-versus-generated identification accuracy | Helped |
| remove dynamics constraints and synthesize from appearance statistics only (ablation) | Two-stream dynamic texture synthesis · Fish sequence | temporal coherence | Hurt |
| remove the appearance stream and synthesize with dynamics loss only (ablation) | Two-stream dynamic texture synthesis · Dynamic texture synthesis | appearance fidelity | Hurt |
| Add the appearance stream to dynamics-only synthesis (ablation) | Two-stream dynamic texture synthesis · Dynamic texture synthesis | appearance fidelity | Helped |
| Use stochastic or near-stochastic appearance rather than large-scale regularity | Two-stream dynamic texture synthesis · Textures grouped by appearance | real-versus-generated identification accuracy: 53.8% ± 7.4%; 55.6% ± 7.4%; 98.9% ± 1.6%; 90.8% ± 4.3% | Mixed |
| Use spatially consistent rather than spatially inconsistent dynamics | Two-stream dynamic texture synthesis · Textures grouped by dynamics | real-versus-generated identification accuracy: 48.6% ± 7.4%; 63.2% ± 7.2%; 92.4% ± 4%; 92.1% ± 4% | Helped |
| Use concat-layer orientation representations instead of flow estimates (ablation) | Two-stream dynamic texture synthesis · Textures with transparency or chaotic motion | synthesis quality | Helped |
Closest papers by meaning
- Diversified Texture Synthesis With Feed-Forward Networks (CVPR 2017 · similarity 0.89)
- Texture Networks: Feed-forward Synthesis of Textures and Stylized Images (ICML 2016 · similarity 0.89)
- Texture Mixer: A Network for Controllable Synthesis and Interpolation of Texture (CVPR 2019 · similarity 0.87)
- Improved Texture Networks: Maximizing Quality and Diversity in Feed-Forward Stylization and Texture Synthesis (CVPR 2017 · similarity 0.87)
- Deep TEN: Texture Encoding Network (CVPR 2017 · similarity 0.87)
- TextureDreamer: Image-Guided Texture Synthesis Through Geometry-Aware Diffusion (CVPR 2024 · 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.