AI Paper Atlas

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

Enhancing Reward Models for High-quality Image Generation: Beyond Text-Image Alignment

Ying Ba, Tianyu Zhang, Yalong Bai, Wenyi Mo, Tao Liang, Bing Su, Ji-Rong Wen

ICCV 2025 · Image generation › Text-to-image diffusion · approach: T2I evaluation and controllable adapters · 9 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
use ICT reward instead of CLIP reward to optimize SD3.5-TurboSD3.5-Turbo · GenEvalMean: 0.13 -> 0.71Improved
optimize SD3.5-Turbo with ICT rewardSD3.5-Turbo · GenEvalMean: 0.69 -> 0.71Improved
optimize SD3.5-Turbo with PickScore rewardSD3.5-Turbo · GenEvalMean: 0.69 -> 0.66Hurt
optimize SD3.5-Turbo with HP rewardSD3.5-Turbo · Parti-PromptsAesthetic Score: 6.293 -> 6.448Improved
optimize SD3.5-Turbo with HP rewardSD3.5-Turbo · Parti-PromptsJPEG compressibility: 313.10 -> 334.86Improved
optimize SD3.5-Turbo with ICT rewardSD3.5-Turbo · Parti-PromptsAesthetic Score: 6.293 -> 6.285No clear effect
use CLIP as reward to optimize SD3.5-TurboSD3.5-Turbo · Parti-PromptsAesthetic Score: 6.293 -> 5.175Hurt
combine ICT and HP rewards multiplicatively instead of ICT alone (ablation)SD3.5-Turbo · Parti-PromptsAesthetic Score: 6.285 -> 6.300Improved
use ICT-HP model instead of PickScore for preference predictionreward model · Pick-High + Pickapic_v2 testaverage accuracy: 79.04 -> 88.84Improved
add HP image-only stage on top of ICT model (ablation)reward model · Pick-High + Pickapic_v2 testaverage accuracy: 87.58 -> 88.47Improved
use ICT-HP model instead of PickScore on I2 > I1 pairsreward model · Pick-High + Pickapic_v2 testpairwise accuracy I2>I1: 74.80 -> 66.42Hurt
optimize SD3.5-Turbo with HPS_v2, PickScore, ImageReward rewardsSD3.5-Turbo · Table 5 multi-metricmultiple reward metricsMixed

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