ComputingComputer ScienceComputer VisionFeature EngineeringMachine LearningDeep LearningGenerative ModelingImage ProcessingImage SynthesisDiffusion ModelsGenerative Adversarial NetworksGenerative Image SynthesisWavelet Analysis
Diffusion models generate high‑fidelity images but suffer from slow training and inference, and even recent DiffusionGAN reductions in sampling steps still lag behind GANs. This work proposes a wavelet‑based diffusion scheme to close the speed gap. By decomposing images and feature maps into low‑ and high‑frequency components with wavelets and processing them adaptively, and adding a reconstruction term to accelerate convergence, the method achieves faster generation without sacrificing quality. Experiments on CelebA‑HQ, CIFAR‑10, LSUN‑Church, and STL‑10 demonstrate real‑time, high‑fidelity generation. Code and pretrained checkpoints are released at https://github.com/VinAIResearch/WaveDiff.git.
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 https://github.com/VinAIResearch/WaveDiff.git.
24
Advances in neural information processing systems 7
Computers & Mathematics with Applications · 1996 · 14.4K citations
Hierarchical Text-Conditional Image Generation with CLIP Latents
Aditya Ramesh, Prafulla Dhariwal, Alex Nichol et al. · arXiv (Cornell University) · 2022 · 2.3K citations · Full text
Diffusion Models Beat GANs on Image Synthesis
Prafulla Dhariwal, Alex Nichol · arXiv (Cornell University) · 2021 · 2.2K citations · Full text