AI Paper Atlas

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

InstaStyle: Inversion Noise of a Stylized Image is Secretly a Style Adviser

Xing Cui, Zekun Li, Peipei Li, Huaibo Huang, Xuannan Liu, Zhaofeng He

ECCV 2024 · Image generation › Text-to-image diffusion · approach: Text-driven synthesis and semantic guidance · 7 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 the inversion style noise with random noise (ablation)InstaStyle · stylized generation, 100 images per stylestyle consistency scoreHurt
add prompt refinement (learned style token) on top of inversion noise (ablation)InstaStyle · stylized generation, 100 images per stylestyle and content scoresImproved
use DDIM inversion noise as initial noise instead of random noise for generation (ablation)Stable Diffusion · stylized generationstyle detailsImproved
use InstaStyle instead of Custom Diffusionstylized text-to-image generation · 60 styles, 100 CIFAR100 objectsstyle consistency (CLIP): 0.611 -> 0.655Improved
use InstaStyle instead of StyleDropstylized text-to-image generation · 60 styles, 100 CIFAR100 objectscontent fidelity (CLIP): 0.248 -> 0.294Improved
use InstaStyle instead of DreamBoothstylized text-to-image generation · 60 styles, 100 CIFAR100 objectscontent fidelity (CLIP): 0.244 -> 0.294Improved
use InstaStyle (text content) instead of style transfer methods that take content imagesstylized generation · 60 stylesstyle consistency (CLIP): 0.604 -> 0.655 vs AesPA-NetImproved
increase noise mix ratio and prompt mix ratio for style combination (ablation)InstaStyle style combination · two-style combinationqualitative styleImproved

Later papers that compared with this method

1 later accepted paper reports InstaStyle as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.

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