AI Paper Atlas

Research map / Image generation / Text-to-image diffusion

Restoring Initial Noise Sensitivity in Text-to-Image Distillation through Geometric Alignment

Huayang Huang, Ruoyu Wang, Jinhui Zhao, Wei Deng, Daiguo Zhou, Jian Luan, Yu Wu, Ye Zhu

ICML 2026 · Image generation › Text-to-image diffusion · approach: T2I evaluation and controllable adapters

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 GAD (JVP matching regularizer) to LADDLADD, SD2.1 · Seed identifiabilityself-identifiability accuracy (%): 87.60% -> 92.40%Improved
add GAD to LADDLADD, SD2.1 · Layout control (800 COCO prompts)AP: 5.0 -> 5.8Improved
add GAD to SiDSiD, SANA · MS-COCO 1000 promptsCLIP score: 32.75 -> 34.40Improved
add GAD to TDMTDM, PixArt-alpha · MS-COCO 1000 promptsPickScore: 22.1007 -> 22.2704Improved
add GAD to TDMTDM, PixArt-alpha · zero-shot FID vs COCO-30kFID: 25.3909 -> 26.3235Hurt
add GAD to LADDLADD, SD2 · FID vs TeacherFID: 11.2920 -> 7.8463Improved
add GAD to SiDSiD, SANA · DrawBench zero-shot NoiseQuery transferHPSv2: 0.249 -> 0.289Improved
add GAD to TDMTDM, PixArt-alpha · Trajectory deviation, 200 unseen promptsfinal cumulative deviation: 0.491 -> 0.427Improved
replace Jacobian matching with noise augmentation (perturbed inputs, no matching target) (ablation)TDM, PixArt-alpha · 1k COCO promptsVendi score: 2.7187 -> 2.6355Hurt
replace Jacobian matching with blind diversity repulsion loss (ablation)TDM, PixArt-alpha · 1k COCO promptsCLIP score: Vendi 3.0559 (highest), CLIP 33.42 -> 31.76Mixed
apply GAD only at mid-noise timesteps instead of full steps (ablation)TDM, PixArt-alpha · GAD timesteps ablationVendi score: 2.5967 -> 2.7140 (full steps 2.7187)Mixed
use increasing linear schedule for perturbation scale h instead of fixed h=1e-2 (ablation)GAD on TDM, PixArt-alpha · scale schedule ablationVendi score: 2.7187 -> 2.7202No clear effect

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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