Making Reconstruction FID Predictive of Diffusion Generation FID
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866918487563173888 |
|---|---|
| author | Xu, Tongda He, Mingwei Abu-Hussein, Shady Hernandez-Lobato, Jose Miguel Zheng, Chunhang Zhao, Kai Zhou, Chao Zhang, Ya-Qin Wang, Yan |
| author_facet | Xu, Tongda He, Mingwei Abu-Hussein, Shady Hernandez-Lobato, Jose Miguel Zheng, Chunhang Zhao, Kai Zhou, Chao Zhang, Ya-Qin Wang, Yan |
| contents | It is well known that the reconstruction FID (rFID) of a VAE is poorly correlated with the generation FID (gFID) of a latent diffusion model. We propose interpolated FID (iFID), a simple variant of rFID that exhibits a strong correlation with gFID. Specifically, for each dataset element, we retrieve its nearest neighbor in latent space, interpolate between their latent representations, decode the interpolated latent, and compute the FID between the decoded samples and the original dataset. We provide an intuitive explanation for why iFID correlates well with gFID, and why reconstruction metrics can be negatively correlated with gFID, by connecting iFID to recent results on diffusion generalization and hallucination. Theoretically, we show that iFID evaluates decoded interpolations aligned with the ridge set around which diffusion samples concentrate, thereby measuring a quantity closely related to diffusion sample quality. Empirically, iFID is the first metric shown to strongly correlate with diffusion gFID across diverse VAEs, achieving Pearson and Spearman correlations of approximately $0.85$. The project page is available at https://tongdaxu.github.io/pages/ifid.html. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_05630 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Making Reconstruction FID Predictive of Diffusion Generation FID Xu, Tongda He, Mingwei Abu-Hussein, Shady Hernandez-Lobato, Jose Miguel Zheng, Chunhang Zhao, Kai Zhou, Chao Zhang, Ya-Qin Wang, Yan Computer Vision and Pattern Recognition Machine Learning It is well known that the reconstruction FID (rFID) of a VAE is poorly correlated with the generation FID (gFID) of a latent diffusion model. We propose interpolated FID (iFID), a simple variant of rFID that exhibits a strong correlation with gFID. Specifically, for each dataset element, we retrieve its nearest neighbor in latent space, interpolate between their latent representations, decode the interpolated latent, and compute the FID between the decoded samples and the original dataset. We provide an intuitive explanation for why iFID correlates well with gFID, and why reconstruction metrics can be negatively correlated with gFID, by connecting iFID to recent results on diffusion generalization and hallucination. Theoretically, we show that iFID evaluates decoded interpolations aligned with the ridge set around which diffusion samples concentrate, thereby measuring a quantity closely related to diffusion sample quality. Empirically, iFID is the first metric shown to strongly correlate with diffusion gFID across diverse VAEs, achieving Pearson and Spearman correlations of approximately $0.85$. The project page is available at https://tongdaxu.github.io/pages/ifid.html. |
| title | Making Reconstruction FID Predictive of Diffusion Generation FID |
| topic | Computer Vision and Pattern Recognition Machine Learning |
| url | https://arxiv.org/abs/2603.05630 |