AI Paper Atlas

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

GASS: Geometry-Aware Spherical Sampling for Disentangled Diversity Enhancement in Text-to-Image Generation

Ye Zhu, Kaleb Newman, Johannes F. Lutzeyer, Adriana Romero-Soriano, Michal Drozdzal, Olga Russakovsky

ICML 2026 · Image generation › Text-to-image diffusion · approach: T2I evaluation and controllable adapters · 0 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 re-normalization of perturbed embeddings (ablation)GASS (SD3-M) · DrawBenchImageReward: 0.732 -> 0.778Improved
add re-normalization of perturbed embeddings (ablation)GASS (SD3-M) · DrawBenchClipScore: 0.313 -> 0.320Improved
use dominant residual basis instead of isotropic perturbation (ablation)GASS (SD3-M) · DrawBenchClipScore: 0.308 -> 0.320Improved
use dominant residual basis instead of random orthogonal direction (ablation)GASS (SD3-M) · DrawBenchClipScore: 0.313 -> 0.320Improved
expand along both prompt-dependent and independent axes instead of only independent (ablation)GASS (SD3-M) · DrawBenchSPP: 0.111 -> 0.114Improved
increase expansion range from 0.02 to 0.05 (ablation)GASS (SD3-M) · DrawBenchSPP: 0.114 -> 0.112Hurt
use uniform instead of consecutive steps (t=10) (ablation)GASS (SD3-M) · DrawBenchImageReward: 0.808 -> 0.757Hurt
increase uniform guidance steps from 10 to 20 (ablation)GASS (SD3-M) · DrawBenchVendi Score: 8.127 -> 8.212Improved
add GASS to CFG on long promptsSD3-M · DrawBench long promptsVendi Score: 7.549 -> 7.935Improved
add GASS to CFG on short promptsSD3-M · DrawBench short promptsImageReward: 0.748 -> 0.698Hurt
add GASS to CFGSD3-M · ImageNet-1KVendi Score: 28.119 -> 28.877Improved
add GASS to CFGSD2.1 · ImageNet-1KDensity: 1.130 -> 1.015Hurt

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