AI Paper Atlas

Research map / Image generation / Super-resolution

Any-Resolution Training for High-Resolution Image Synthesis

Lucy Chai, Michaël Gharbi, Eli Shechtman, Phillip Isola, Richard Zhang

ECCV 2022 · Image generation › Super-resolution · approach: Perceptual and real-world SR · 62 citations (OpenAlex)

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 changedWhereMeasuredEffect
add scale conditioning (ablation)inverse-teacher Anyres-GAN · FFHQFID@1024: 4.88 -> 4.47Improved
add scale conditioning (ablation)inverse-teacher Anyres-GAN · FFHQpFID: 4.67 -> 4.28Improved
use inverse instead of forward teacher loss (ablation)Anyres-GAN · FFHQ6KFID@1024: 5.35 -> 4.88Improved
add teacher regularization (inverse) (ablation)Anyres-GAN · FFHQ6KFID@1024: 7.13 -> 4.88Improved
add teacher regularization (inverse) (ablation)Anyres-GAN · FFHQ6KpFID: 3.17 -> 4.67Hurt
raise teacher weight from 2 to 5 (ablation)Anyres-GAN · FFHQ6KFID@1024 and pFID: FID@1024 5.46 -> 4.47; pFID 3.15 -> 4.28Mixed
multisize training instead of resizing all images down to 512 (ablation)Anyres-GAN · FFHQFID@1024: 19.18 -> 4.47Improved
multisize training instead of upsampling all images to 1024 (ablation)Anyres-GAN · FFHQFID@1024: 4.86 -> 4.47Improved
multisize training instead of 1024 subset only (ablation)Anyres-GAN · FFHQFID@512: 12.41 -> 4.41Improved
increase HR images from 1k to 70k (ablation)Anyres-GAN · FFHQFID: 4.38 -> 4.52No clear effect
sample training scales from (512, 1024) instead of (256, 512) (ablation)Anyres-GAN · FFHQ6KFID@1024: 19.01 -> 4.16Improved
remove base-resolution sampling for the discriminator (ablation)Anyres-GAN · FFHQ6KFID@256: 3.28 -> 9.96Hurt

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy