AI Paper Atlas

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

FouriScale: A Frequency Perspective on Training-Free High-Resolution Image Synthesis

Linjiang Huang, Rongyao Fang, Aiping Zhang, Guanglu Song, Si Liu, Yu Liu, Hongsheng Li

ECCV 2024 · Image generation › Text-to-image diffusion · approach: Text-driven synthesis and semantic guidance · 40 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
add FouriScale guidance (milder-filter conditional estimate with attention maps from FouriScale estimate) (ablation)FouriScale · LAION-5B, 2048x2048 (SD 2.1)FID_r: -4.26 FID_rImproved
add low-pass filtering to dilated convolution (ablation)FouriScale without guidance · LAION-5B, 2048x2048 (SD 2.1)FID_r: 46.74 without filteringImproved
replace ScaleCrafter kernel enlargement with dilation plus low-pass filteringSDXL 1.0 high-resolution generation · LAION-5B, 16x 1:1FID_r: 84.58 -> 56.66Improved
apply FouriScale instead of vanilla model at 4x resolutionSD 1.5 · LAION-5B, 4x 1:1FID_r: 26.96 -> 23.62Improved
use attention-entropy scaling (Attn-Entro) instead of vanillaSD 1.5 · LAION-5B, 16x 1:1FID_r: 67.90 -> 67.45No clear effect
use a gentler low-pass filter (sigma 0.6) instead of the ideal filter for SDXLFouriScale on SDXL · SDXLvisual qualityImproved
use FouriScale one-stage generation instead of SD followed by diffusion super-resolution cascadeSD 2.1 at 2048x2048 · LAION-5B, 2048x2048FID_r / KID_rHurt
use FouriScale instead of ScaleCrafterSDXL 16x · SDXL 16x 1:1seconds per image: 577 -> 540Improved

Later papers that compared with this method

4 later accepted papers report FouriScale as a baseline in their results tables; 4 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