AI Paper Atlas

Research map / Image generation / Fast diffusion sampling

Spectrally-Guided Diffusion Noise Schedules

Carlos Esteves, Ameesh Makadia

ICML 2026 · Image generation › Fast diffusion sampling · approach: Denoising diffusion probabilistic models · 0 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 per-instance spectrally-guided mixed noise schedule instead of SiD2 schedule (with half the steps)SiD2 small · ImageNet 256x256FID: 1.68 -> 1.42Improved
use per-instance spectrally-guided scheduleSiD2 flop heavy · ImageNet 512x512FID: 1.53 -> 1.45Improved
use per-instance spectrally-guided scheduleSiD2 small · MS-COCO 512x512 zero-shotFID: 7.56 -> 7.08Improved
use per-instance schedule instead of a fixed median schedule (ablation)Ours · ImageNet 256x256FID: 1.61 -> 1.42Improved
use spectral min/max-bounded mixed schedule instead of cosine between per-instance min and max (ablation)Ours · ImageNet 256x256FID: 1.98 -> 1.42Improved
average frequency-focused and power-focused schedules (mixed) instead of frequency-focused only (ablation)Ours · ImageNet 256x256FID: 5.16 -> 1.42Improved
average frequency-focused and power-focused schedules (mixed) instead of power-focused only (ablation)Ours · ImageNet 256x256FID: 5.18 -> 1.42Improved
add min/max logSNR FiLM conditioning (ablation)Ours · ImageNet 256x256FID: 1.58 -> 1.42Improved
define guidance interval in t instead of logSNR (ablation)Ours · ImageNet 256x256FID: 1.58 -> 1.42Improved
use ground-truth spectrum instead of trained RAPSD sampler (ablation)Ours · ImageNet 256x256FID: 1.42 -> 1.43No clear effect
use ground-truth spectrum instead of trained RAPSD sampler (ablation)Ours · MS-COCO 512x512 zero-shotFID: 7.08 -> 6.55Improved
increase number of denoising steps at high counts (ablation)Ours · ImageNetFID: 7.08 -> 7.83 (MS-COCO, 96 -> 256 NFE)Hurt

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