A collection of noise designs for diffusion models.
Please find more details on our Eurographics 2025 tutorial / SIGGRAPH 2025 Course: https://diffusion-noise.mpi-inf.mpg.de/ .
-
Denoising diffusion probabilistic models (NeurIPS 2020) [code] [webpage]
-
Denoising Diffusion Implicit Models (ICLR 2021) [code]
-
Improved denoising diffusion probabilistic models (ICML 2021) [code] [webpage]
-
High-Resolution Image Synthesis with Latent Diffusion Models (CVPR 2022) [code]
-
Blue noise for diffusion models (SIGGRAPH 2024) [code] [webpage]
-
Edge-preserving noise for diffusion models (ICLR 2025 DeLTa workshop) [code] [webpage]
-
Multi-Resolution Noise for Diffusion Model Training [blogpost]
-
Common Diffusion Noise Schedules and Sample Steps are Flawed (WACV 2024) [unofficial code]
-
On the Importance of Noise Scheduling for Diffusion Models (Arxiv 2023) [paper]
-
Diffusion with Offset Noise (2023) [blogpost]
-
Relay Diffusion: Unifying diffusion process across resolutions for image synthesis (ICLR 2024) [code]
-
Golden Noise for Diffusion Models: A Learning Framework (2024) [code]
-
Inference-Time Alignment of Diffusion Models with Direct Noise Optimization (ICML 2024 SPIGM workshop) [code] [webpage]
-
ReNO: Enhancing One-step Text-to-Image Models through Reward-based Noise Optimization (NeurIPS 2024) [code]
-
InitNO: Boosting Text-to-Image Diffusion Models via Initial Noise Optimization (CVPR 2024) [code] [webpage]
-
SDEdit: Guided Image Synthesis and Editing with Stochastic Differential Equations (ICLR 2022) [code] [webpage]
-
RePaint: Inpainting using Denoising Diffusion Probabilistic Models (CVPR 2022) [code]
-
Factorized Diffusion: Perceptual Illusions by Noise Decomposition (ECCV 2024) [code] [webpage]
-
An Edit Friendly DDPM Noise Space: Inversion and Manipulations (CVPR 2024) [code] [webpage]
-
A Traveler’s Guide to the Latent Space (section regarding Perlin noise init) [blogpost]
-
NoiseCollage: A Layout-Aware Text-to-Image Diffusion Model Based on Noise Cropping and Merging (ICLR 2024) [code]
-
Optimizing Diffusion Noise Can Serve As Universal Motion Priors (CVPR 2024) [code] [webpage]
-
Nonisotropic Gaussian Diffusion for Realistic 3D Human Motion Prediction (CVPR 2025) [code] [webpage]
-
Video Diffusion Models (NeurIPS 2022) [unofficial code] [webpage]
-
Align your Latents: High-Resolution Video Synthesis with Latent Diffusion Models (CVPR 2023) [webpage]
-
How I Warped Your Noise: a Temporally-Correlated Noise Prior for Diffusion Models (ICLR 2024) [webpage]
-
Preserve Your Own Correlation: A Noise Prior for Video Diffusion Models (ICCV 2023) [webpage]
-
Text2Video-Zero: Text-to-Image Diffusion Models are Zero-Shot Video Generators (ICCV 2023) [code] [webpage]
-
Go-with-the-Flow: Motion-Controllable Video Diffusion Models Using Real-Time Warped Noise (CVPR 2025) [code] [webpage]
-
EquiVDM: Equivariant Video Diffusion Models with Temporally Consistent Noise (Arxiv 2025) [webpage]
-
MultiDiff: Consistent Novel View Synthesis from a Single Image (CVPR 2024) [webpage]
-
Consistent Flow Distillation for Text-to-3D Generation (ICLR 2025) [code] [webpage]
- Noise Synthesis for Low-Light Image Denoising with Diffusion Models (CVPR 2025) [paper]
If you find this useful please consider citing:
@article{singh2025demystifying,
author = {Singh, Gurprit and Huang, Xingchang and Vandersanden, Jente and Öztireli, Cengiz and Mitra, Niloy},
title = {Demystifying noise: The role of randomness in generative AI},
journal = {Computer Graphics Forum},
year = {2025}
}