AI Paper Atlas

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

UDiffText: A Unified Framework for High-quality Text Synthesis in Arbitrary Images via Character-aware Diffusion Models

Yiming Zhao, Zhouhui Lian

ECCV 2024 · Image generation › Text-to-image diffusion · approach: Text-driven synthesis and semantic guidance · 49 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
replace CLIP text encoder with a character-level text encoder (ablation)SD 2.0 inpainting · SynthText reconstructionSeqAcc-Recon (%): 8.0 -> 40.0Improved
add local attention loss supervised by character segmentation maps (ablation)UDiffText · SynthText reconstructionSeqAcc-Recon (%): 40.0 -> 54.0Improved
add scene text recognition loss (ablation)UDiffText · SynthText reconstructionSeqAcc-Recon (%): 54.0 -> 65.0Improved
add inference-time noised latent refinement (initial noise selection and attention-based latent update) (ablation)UDiffText · SynthText reconstructionSeqAcc-Recon (%): 65.0 -> 76.0Improved
use UDiffText instead of TextDiffuser for text editing on TextSegscene text editing · TextSeg editingSeqAcc-Editing (%): 66.0 -> 84.0Improved
apply UDiffText correction stage to SDXL generated imagesSDXL · SimpleBench promptstext rendering accuracy (%): 8.0% -> 60.0%Improved
use only denoising score matching loss without local attention lossUDiffText · text renderingtext accuracyHurt
use UDiffText instead of MOSTEL on image quality metricsscene text editing · text image qualityFID: 25.09 -> 15.79Improved

Later papers that compared with this method

3 later accepted papers report UDiffText as a baseline in their results tables; 3 of them report a better number on most of the results they share with it.

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