AI Paper Atlas

Research map / Image generation / Autoregressive and latent generation

PixelU: A U-Shaped Transformer for Efficient End-to-End Pixel Diffusion

Zipeng Guo, Lichen Ma, Yu He, Xiaolong Fu, Jingling Fu, Junshi Huang, Yan Li

ECCV 2026 · Image generation › Autoregressive and latent generation · approach: High-resolution diffusion synthesis · 7 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 skip connections to JiT-B/16 (x-pred) (ablation)JiT-B/16 · ImageNet 256 160ep cfg=1FID: 43.65 -> 38.53Improved
add DDT decoder (x-pred) (ablation)JiT-B/16 · ImageNet 256 160ep cfg=1FID: 43.65 -> 43.48No clear effect
add PixelDiT decoder instead of skips (ablation)JiT-B/16 · ImageNet 256 160ep cfg=1FID: 43.65 -> 39.20Improved
switch from x-pred to v-pred with skips (ablation)JiT-B/16 + skips · ImageNet 256 160ep cfg=1FID: 38.53 -> 185.54Hurt
add DeCo decoder under v-pred (ablation)JiT-B/16 · ImageNet 256 160ep cfg=1FID: 190.11 -> 42.07Improved
add spatial down-sampling to skip connections (ablation)PixelU-B/16 · ImageNet 256 160ep cfg=1FID: 38.53 -> 34.92Improved
add down-sampling without skip connections (ablation)JiT-B/16 · ImageNet 256 160ep cfg=1FID: 43.65 -> 48.87Hurt
balanced [4,4,4] block allocation vs [3,6,3] (ablation)PixelU-B/16 · ImageNet 256 160ep cfg=1FID: 38.08 -> 34.92Improved
use one down-sampling stage instead of two (ablation)PixelU-B/16 · ImageNet 256 160ep cfg=1FID: 42.70 -> 34.92Improved
add in-context class tokensPixelU-B/16 · ImageNet 256 160ep cfg=1FID: 34.92 -> 30.88Improved
add REPA to class tokensPixelU-B/16 · ImageNet 256 160ep cfg=1FID: 30.88 -> 26.86Improved
add CFG and CFG interval, 600 epochsPixelU-B/16 · ImageNet 256FID: 26.86 -> 3.40Improved

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