VIVAT: Virtuous Improving VAE Training through Artifact Mitigation
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866909935962423296 |
|---|---|
| author | Novitskiy, Lev Vasilev, Viacheslav Kovaleva, Maria Arkhipkin, Vladimir Dimitrov, Denis |
| author_facet | Novitskiy, Lev Vasilev, Viacheslav Kovaleva, Maria Arkhipkin, Vladimir Dimitrov, Denis |
| contents | Variational Autoencoders (VAEs) remain a cornerstone of generative computer vision, yet their training is often plagued by artifacts that degrade reconstruction and generation quality. This paper introduces VIVAT, a systematic approach to mitigating common artifacts in KL-VAE training without requiring radical architectural changes. We present a detailed taxonomy of five prevalent artifacts - color shift, grid patterns, blur, corner and droplet artifacts - and analyze their root causes. Through straightforward modifications, including adjustments to loss weights, padding strategies, and the integration of Spatially Conditional Normalization, we demonstrate significant improvements in VAE performance. Our method achieves state-of-the-art results in image reconstruction metrics (PSNR and SSIM) across multiple benchmarks and enhances text-to-image generation quality, as evidenced by superior CLIP scores. By preserving the simplicity of the KL-VAE framework while addressing its practical challenges, VIVAT offers actionable insights for researchers and practitioners aiming to optimize VAE training. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2506_07863 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | VIVAT: Virtuous Improving VAE Training through Artifact Mitigation Novitskiy, Lev Vasilev, Viacheslav Kovaleva, Maria Arkhipkin, Vladimir Dimitrov, Denis Computer Vision and Pattern Recognition Machine Learning Multimedia Variational Autoencoders (VAEs) remain a cornerstone of generative computer vision, yet their training is often plagued by artifacts that degrade reconstruction and generation quality. This paper introduces VIVAT, a systematic approach to mitigating common artifacts in KL-VAE training without requiring radical architectural changes. We present a detailed taxonomy of five prevalent artifacts - color shift, grid patterns, blur, corner and droplet artifacts - and analyze their root causes. Through straightforward modifications, including adjustments to loss weights, padding strategies, and the integration of Spatially Conditional Normalization, we demonstrate significant improvements in VAE performance. Our method achieves state-of-the-art results in image reconstruction metrics (PSNR and SSIM) across multiple benchmarks and enhances text-to-image generation quality, as evidenced by superior CLIP scores. By preserving the simplicity of the KL-VAE framework while addressing its practical challenges, VIVAT offers actionable insights for researchers and practitioners aiming to optimize VAE training. |
| title | VIVAT: Virtuous Improving VAE Training through Artifact Mitigation |
| topic | Computer Vision and Pattern Recognition Machine Learning Multimedia |
| url | https://arxiv.org/abs/2506.07863 |