Papers › Wavelet Diffusion Models are fast and scalable Image Generators

Wavelet Diffusion Models are fast and scalable Image Generators

29 Nov 2022CVPR 2023 1arXiv:2211.16152archive 2025-07-28

Hao Phung, Quan Dao, Anh Tran

Diffusion models are rising as a powerful solution for high-fidelity image generation, which exceeds GANs in quality in many circumstances. However, their slow training and inference speed is a huge bottleneck, blocking them from being used in real-time applications. A recent DiffusionGAN method significantly decreases the models' running time by reducing the number of sampling steps from thousands to several, but their speeds still largely lag behind the GAN counterparts. This paper aims to reduce the speed gap by proposing a novel wavelet-based diffusion scheme. We extract low-and-high frequency components from both image and feature levels via wavelet decomposition and adaptively handle these components for faster processing while maintaining good generation quality. Furthermore, we propose to use a reconstruction term, which effectively boosts the model training convergence. Experimental results on CelebA-HQ, CIFAR-10, LSUN-Church, and STL-10 datasets prove our solution is a stepping-stone to offering real-time and high-fidelity diffusion models. Our code and pre-trained checkpoints are available at \url{https://github.com/VinAIResearch/WaveDiff.git}.

PaperPDFConference PDFCode

In Syntology Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

Code

vinairesearch/wavediff officialmentioned in papermentioned on GitHubpytorchAGPL-3.0 report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

BlockingImage Generation

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Image Generation CelebA-HQ 1024x1024 WaveDiff FID 5.98 #4 of 10 Archive leaderboard report
Image Generation CelebA-HQ 1024x1024 WaveDiff NFE 2 #4 of 10 Archive leaderboard report
Image Generation CelebA-HQ 256x256 WaveDiff FID 5.94 #6 of 19 Archive leaderboard report
Image Generation CelebA-HQ 256x256 WaveDiff NFE 2 #6 of 19 Archive leaderboard report
Image Generation CelebA-HQ 256x256 WaveDiff Recall 0.37 #6 of 19 Archive leaderboard report
Image Generation CelebA-HQ 512x512 WaveDiff FID 6.40 #1 of 1 Archive leaderboard report
Image Generation CelebA-HQ 512x512 WaveDiff NFE 2 #1 of 1 Archive leaderboard report
Image Generation CelebA-HQ 512x512 WaveDiff Recall 0.35 #1 of 1 Archive leaderboard report
Image Generation LSUN Churches 256 x 256 WaveDiff FID 5.06 #17 of 27 Archive leaderboard report
Image Generation LSUN Churches 256 x 256 WaveDiff NFE 4 #17 of 27 Archive leaderboard report
Image Generation LSUN Churches 256 x 256 WaveDiff Recall 0.40 #17 of 27 Archive leaderboard report
Image Generation STL-10 WaveDiff FID 12.93 #7 of 31 Archive leaderboard report
Image Generation STL-10 WaveDiff NFE 4 #7 of 31 Archive leaderboard report
Image Generation STL-10 WaveDiff Recall 0.41 #7 of 31 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

DiffusionSPEED

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections