Research map / Image generation / Image restoration
Latent Convolutional Models
ShahRukh Athar, Evgeny Burnaev, 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 |
|---|---|---|---|
| remove convolutional manifold constraint at test time (optimize z directly) (ablation) | LCM · CelebA-HQ inpainting | MSE (inpainted pixels): 0.00171 -> 0.00854 | Hurt |
| remove convolutional manifold constraint at test time (ablation) | LCM · CelebA-HQ inpainting | perceptual metric: 0.02381 -> 0.02736 | Hurt |
| remove convolutional manifold constraint at test time (ablation) | LCM · CelebA-HQ inpainting | MSE (known pixels): 0.00307 -> 0.00141 | Hurt |
| remove manifold constraint at test time (ablation) | LCM · CelebA and Bedrooms | restoration quality | No clear effect |
| use convolutional manifold latent instead of free latent vector (LCM vs GLO) | LCM · CelebA inpainting | MSE: 0.0038 -> 0.0034 | Improved |
| use convolutional manifold latent instead of free latent vector (LCM vs GLO) | LCM · LSUN-Bedrooms inpainting | MSE: 0.0085 -> 0.0065 | Improved |
| LCM vs GLO on colorization | LCM · LSUN-Bedrooms colorization | MSE: 0.0075 -> 0.0066 | Mixed |
| LCM vs per-image DIP prior | LCM · CelebA inpainting | MSE: 0.0091 -> 0.0034 | Improved |
| LCM vs DIP on super-resolution | LCM · CelebA super-resolution | MSE: 0.0052 -> 0.0061 | Hurt |
| LCM vs DIP on Bedrooms inpainting (user study) | LCM · LSUN-Bedrooms inpainting | user preference: 63.7% vs 29.7% | Hurt |
| LCM vs GAN latent models | LCM · CelebA inpainting | MSE: 0.0344 -> 0.0034 | Improved |
| add Gaussian prior on GAN latent z (ablation) | WGAN/PGAN restoration · CelebA/Bedrooms | realism | No clear effect |
Closest papers by meaning
- Stabilize the Latent Space for Image Autoregressive Modeling: A Unified Perspective (NeurIPS 2024 · similarity 0.89)
- Masked Autoencoders Are Effective Tokenizers for Diffusion Models (ICML 2025 · similarity 0.87)
- Latent Zoning Network: A Unified Principle for Generative Modeling, Representation Learning, and Classification (NeurIPS 2025 · similarity 0.87)
- Isometric Representation Learning for Disentangled Latent Space of Diffusion Models (ICML 2024 · similarity 0.86)
- Simpler Diffusion: 1.5 FID on ImageNet512 with Pixel-space Diffusion (CVPR 2025 · similarity 0.86)
- Sparse Autoencoders, Again? (ICML 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.
