AI Paper Atlas

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

Scale Your Instructions: Enhance the Instruction-Following Fidelity of Unified Image Generation Model by Self-Adaptive Attention Scaling

Chao Zhou, Tianyi Wei, Nenghai Yu

ICCV 2025 · Image generation › Text-to-image diffusion · approach: Layout-to-image control · 3 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
apply self-adaptive attention scaling (SaaS) to OmniGen for multi-sub-instruction editingOmniGen-v1 · PIE-Bench subset (multiple sub-instructions)CLIP-T: 0.276 -> 0.315Improved
apply SaaS to OmniGen for multi-sub-instruction editingOmniGen-v1 · PIE-Bench subset (multiple sub-instructions)PickScore: 0.244 -> 0.513Improved
apply SaaS to OmniGen for single-instruction editingOmniGen-v1 · EMU-Edit (single instruction)PickScore: 0.233 -> 0.462Improved
apply SaaS to OmniGen for single-instruction editingOmniGen-v1 · EMU-Edit (single instruction)CLIP-I: 0.915 -> 0.900Hurt
apply SaaS to OmniGen for depth-conditioned generationOmniGen-v1 · MultiGen-20M depth (2,000 samples)PickScore: 0.342 -> 0.658Improved
apply SaaS to OmniGen for segmentation-conditioned generationOmniGen-v1 · ADE20K segmentation (2,000 samples)CLIP-T: 0.274 -> 0.317Improved
use adaptive scaling factors instead of fixed scaling factors (2 or 5) (ablation)SaaS · image editing (qualitative)instruction followingImproved
use SaaS instead of simply increasing the guidance scale (ablation)OmniGen-v1 · EMU-Edit (single instruction)PickScore: 0.228 -> 0.397Improved
vary mask threshold between 0.2, 0.4, 0.6, 0.8 and auto (ablation)SaaS · image editingPickScore: 0.195 to 0.203No clear effect
apply SaaS in later denoising steps instead of early steps (ablation)SaaS · image editing (qualitative)instruction followingHurt
apply SaaS to deeper layers instead of shallow layers (ablation)SaaS · image editing (qualitative)instruction followingImproved
add SaaS overhead at inferenceOmniGen-v1 · RTX A6000 latencylatency (s): 29.1 -> 29.4No 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