AI Paper Atlas

Research map / Image generation / Fast diffusion sampling

What kills $v$-prediction? A Patch-wise PCA Perspective on Pixel-Space Flow Matching

Tianze Luo, Haofeng Huang, Yuan Yao

NeurIPS 2026 · Image generation › Fast diffusion sampling · approach: Flow matching distillation · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Analyzes why v-prediction fails in pixel-space flow matching through a patch-wise PCA lens and proposes a remedy for the resulting training instability.

Problem

Pixel-space flow matching with the standard v-prediction objective often suffers training instabilities, while latent diffusion requires two-stage training and is limited by autoencoder reconstruction.

Approach

Attributes v-prediction failure to a patch-wise PCA structure of pixel data rather than only to the off-manifold nature of noised targets, offering a perspective that explains the instability in pixel space.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
restrict data and noise to low-dimensional principal subspace (PCA 128) (ablation)v-prediction JiT-B · ImageNet 256 pixel spaceFID-50K: 149.07 -> 33.99Improved
increase PCA dimension from 128 to 384 (ablation)v-prediction JiT-B · ImageNet 256FID-50K: 33.99 -> 47.41Hurt
increase PCA dimension to full 768 (ablation)x-prediction JiT-B · ImageNet 256FID-50K: 35.01 -> 40.39No clear effect
increase backbone width to 1024 (JiT-L) (ablation)vanilla v-prediction · ImageNet 256FID-50K: 149.07 -> 33.43Improved
use patch size 8 in backbone (ablation)vanilla v-prediction B · ImageNet 256FID-50K: 149.07 -> 28.70Improved
give decoder direct raw-input access (ablation)width-384 decoder · ImageNet 256FID-50K: 186.16 -> 31.85Improved
increase decoder width from 128 to 256 (ablation)raw-conditioned decoder · ImageNet 256FID-50K: 220.78 -> 34.68Improved
add output-side shortcut with anchored fractional gains (ablation)RiT-B decoder · ImageNet 256FID-50K: 31.85 -> 27.97Improved
use separate denominators for raw and semantic branches (ablation)RiT shortcut · ImageNet 256FID-50K: 27.97 -> 28.51No clear effect
use x-prediction objective with RiT architecture (ablation)RiT-B · ImageNet 256FID-50K: 27.97 -> 28.49Hurt
reduce backbone input bottleneck width from 768 to 128 (ablation)RiT-B · ImageNet 256FID-50K: 30.45 -> 27.97Improved
shift time sampling to higher noise and sigma 0.8 (ablation)RiT-B · ImageNet 256FID-50K: 33.69 -> 27.97Improved

Experimental setup

Builds on: flow matching v-prediction

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