Research map / Video / Video segmentation and tracking
DeepLandscape: Adversarial Modeling of Landscape Videos
Elizaveta Logacheva, Roman Suvorov, Oleg Khomenko, Anton Mashikhin, Victor Lempitsky
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 decaying discriminator frequency instead of freq=0.5 (ablation) | pairwise discriminator · Discriminator balancing (128px) | FID: 82.10 -> 56.13 | Improved |
| add decaying discriminator frequency instead of freq=0.3 (ablation) | pairwise discriminator · Discriminator balancing (128px) | FID: 63.48 -> 56.13 | Improved |
| use decay instead of freq=0.1 (ablation) | pairwise discriminator · Discriminator balancing (128px) | User study R: -0.03 | Improved |
| use decay instead of freq=0.1 (ablation) | pairwise discriminator · Discriminator balancing (128px) | FID: 54.15 -> 56.13 | No clear effect |
| add encoder initialization (E) to model-only optimization (ablation) | inference procedure · Inference ablation | SSIM: 0.95 -> 0.54 | Hurt |
| add optimization after encoder (EO vs E) (ablation) | inference procedure · Inference ablation | LPIPS: 0.43 -> 0.07 | Improved |
| add tying W to initial values (EOI vs EO) (ablation) | inference procedure · Inference ablation | Sky optical flow: 1.5 -> 2.5 | Improved |
| add generator fine-tuning (EOIF vs EOI) (ablation) | inference procedure · Inference ablation | LPIPS: 0.11 -> 0.04 | Improved |
| add segmentation guidance (EOIFS vs EOIF) (ablation) | inference procedure · Inference ablation | Sky optical flow: 2.3 -> 2.6 | Mixed |
| use EOIF instead of I2S inversion (ablation) | inference procedure · Inference ablation | LPIPS: 0.18 -> 0.04 | Improved |
| finetune Animating Landscape on our video dataset (ablation) | Animating Landscape · Animated Landscape on our video data | User preference: 0.68 -> 0.32 | Hurt |
| finetune Animating Landscape on our video dataset (ablation) | Animating Landscape · Animated Landscape on our video data | FVD: 162 -> 159 | No clear effect |
Closest papers by meaning
- StyleCineGAN: Landscape Cinemagraph Generation using a Pre-trained StyleGAN (CVPR 2024 · similarity 0.90)
- Landscape Learning for Neural Network Inversion (ICCV 2023 · similarity 0.87)
- StyleInV: A Temporal Style Modulated Inversion Network for Unconditional Video Generation (ICCV 2023 · similarity 0.86)
- On the "steerability" of generative adversarial networks (ICLR 2020 · similarity 0.85)
- Learning Modulated Transformation in GANs (NeurIPS 2023 · similarity 0.85)
- Optimizing 4D Gaussians for Dynamic Scene Video from Single Landscape Images (ICLR 2025 · similarity 0.84)
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.
