AI Paper Atlas

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

TIPO: Text to Image with Text Pre-sampling for Prompt Optimization

Shih-Ying Yeh, YI LI, SangHyun Park, Giyeong Oh, Xuehai Wang, Min Song, Youngjae Yu, Shang-Hong Lai

ICLR 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
expand tag prompts with TIPOKohaku-XL-Zeta · in-domain tag promptsFDD: 0.3558 -> 0.2282Improved
expand tag prompts with TIPOKohaku-XL-Zeta · in-domain tag promptsAI Corrupt score: 0.5743 -> 0.0805Improved
expand short NL prompts with TIPOSDXL-base-1.0 · in-domain NL-shortFDD: 0.0957 -> 0.1168Hurt
expand short NL prompts with TIPOSDXL-base-1.0 · in-domain NL-shortAesthetic score: 5.8370 -> 5.8531Improved
apply TIPO on out-of-domain promptsSD-3.5-Large · out-of-domainVendi score: 8.9718 -> 21.571Improved
apply TIPO instead of GPT-4o-mini rewriting on out-of-domain promptsSD-3.5-Large · out-of-domainAesthetic score: 6.7125 -> 6.0536Hurt
apply TIPO on models with undisclosed training dataFLUX.1-dev, Lumina-2, HiDream-I1 · COYO/GBC 1000 promptsAesthetic score: 5.2588 -> 5.4105 (Lumina-2)Improved
apply TIPO on Omnigen2Omnigen2 · COYO/GBC 1000 promptsAesthetic score: 5.2629 -> 5.0661Hurt
use TIPO for prompt generation instead of PromptistTIPO · prompt latencyavg. time (s): 1.46 -> 1.03Improved
use TIPO prompts instead of baselines in human studyTIPO · human preferencewin rate (%): 51.3Improved
use TIPO instead of MagicPrompt on NL promptsTIPO · CLIPScore prompt-image, NL-shortCLIPScore: 0.2834 -> 0.2413Hurt
use TIPO prompts for embedding alignment with T2I training textTIPO · prompt distribution alignmentFrechet distance: 0.2578 -> 0.0524 (T5, tag-based vs MagicPrompt)Improved

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