AI Paper Atlas

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

Rethinking Prompt Design for Inference-time Scaling in Text-to-Visual Generation

Subin Kim, Sangwoo Mo, Mamshad Nayeem Rizve, Yiran Xu, Difan Liu, Jinwoo Shin, Tobias Hinz

CVPR 2026 · Image generation › Text-to-image diffusion · approach: T2I evaluation and controllable adapters · 1 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
revise prompt from common failures instead of per-sample failures (ablation)PRIS · T2V prompt revision ablationscore: 0.619 -> 0.754Improved
apply per-sample prompt revision instead of no revision (ablation)PRIS · T2V prompt revision ablationscore: 0.711 -> 0.619Hurt
revise prompt from common failures instead of per-sample failures (ablation)PRIS · T2I prompt revision ablationscore: 0.853 -> 0.854No clear effect
add element-level factual correction instead of decomposed binary VQA (ablation)EFC verifier · Verifier benchmark (410 prompts)accuracy: 0.700 -> 0.763Improved
add PRIS prompt redesign to Best-of-NFLUX.1-dev · GenAI-BenchVQAScore: 0.783 -> 0.854Improved
add PRIS prompt redesign to Best-of-N* on Wan2.1-14BWan2.1-14B · VBench-2.0 Controllability & Creativityaverage score: 58.51 -> 68.85Improved
add PRIS to Wan2.1-1.3B vs BoN* on commonsense and physicsWan2.1-1.3B · VBench-2.0 Commonsense & Physicsaverage score: 62.61 -> 66.07Improved
add PRIS on Thermotics dimensionWan2.1-1.3B · VBench-2.0 Thermoticsscore: 86.25 -> 70.37Hurt
use PRIS with fewer NFEs than BoN under matched wall-clock timeFLUX.1-dev · GenAI-Bench matched computescore: 0.790 -> 0.834Improved
use PRIS with fewer NFEs than BoN under matched wall-clock timeWan2.1-1.3B · T2V matched computescore: 0.935 -> 0.964Improved
add standard prompt expansion to FLUX.1-devFLUX.1-dev · GenAI-BenchVQAScore: 0.718 -> 0.769Improved
add Best-of-N to Wan2.1-1.3BWan2.1-1.3B · VBench-2.0 Commonsense & Physicsaverage score: 63.22 -> 62.61Hurt

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