Research map / Video / Video segmentation and tracking
ViSt3D: Video Stylization with 3D CNN
Ayush Pande, Gaurav Sharma
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 appearance subnetworks that disentangle appearance from motion before AdaIN3D (ablation) | ViSt3D (C3D encoder + AdaIN3D) · video stylization, qualitative | motion artifacts | Improved |
| add temporal loss (optical-flow based) (ablation) | ViSt3D · Optical flow error, 20 test pairs (mean) | mean abs optical flow difference: 1.63 -> 1.00 | Improved |
| add temporal loss (optical-flow based) (ablation) | ViSt3D · video stylization, qualitative | color consistency and motion preservation | Improved |
| add intra-clip loss on top of temporal loss (ablation) | ViSt3D+temporal · video stylization, qualitative | flashing color inconsistencies | Improved |
| add intra-clip loss on top of temporal loss (ablation) | ViSt3D+temporal · Optical flow error, 20 test pairs (mean) | mean abs optical flow difference: 1.00 -> 1.11 | Hurt |
| use static style clip with 3D encoder instead of 2D CNN feature statistics transfer | ViSt3D encoder · initial experiments | stylization quality | Improved |
| compare ViSt3D against AdaAttN on content preservation | ViSt3D · User study | content preservation preference (%): 41.0 -> 43.3 | Improved |
| compare ViSt3D against MCCNet on style degree | ViSt3D · User study | style degree preference (%): 31.3 vs 16.7 | Hurt |
| compare ViSt3D against AdaAttN on overall preference | ViSt3D · User study | overall preference (%): 28.3 -> 28.7 | Improved |
| 3D CNN instead of 2D AdaAttN on compute | ViSt3D · 144 frames 640x360 | processing time (s): 14 seconds -> 60 seconds | Hurt |
| compare ViSt3D (full) with AdaAttN on optical flow error | ViSt3D+temporal+intra · Optical flow error, 20 test pairs (mean) | mean abs optical flow difference: 1.26 vs 1.11 | Improved |
Closest papers by meaning
- FreeViS: Training-free Video Stylization with Inconsistent References (ICLR 2026 · similarity 0.89)
- Styl3R: Instant 3D Stylized Reconstruction for Arbitrary Scenes and Styles (NeurIPS 2025 · similarity 0.88)
- Generative Human Motion Stylization in Latent Space (ICLR 2024 · similarity 0.87)
- Morpheus: Text-Driven 3D Gaussian Splat Shape and Color Stylization (CVPR 2025 · similarity 0.87)
- PNeSM: Arbitrary 3D Scene Stylization via Prompt-Based Neural Style Mapping (AAAI 2024 · similarity 0.86)
- Multi-StyleGS: Stylized Gaussian Splatting with Multiple Styles (AAAI 2025 · 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.
