In this paper, we present the Directly Denoising Diffusion Model (DDDM): a
simple and generic approach for generating realistic images with few-step
sampling, while multistep sampling is still preserved for better performance.
DDDMs require no delicately designed samplers nor distillation on pre-trained
distillation models. DDDMs train the diffusion model conditioned on an
estimated target that was generated from previous training iterations of its
own. To generate images, samples generated from the previous time step are also
taken into consideration, guiding the generation process iteratively. We
further propose Pseudo-LPIPS, a novel metric loss that is more robust to
various values of hyperparameter. Despite its simplicity, the proposed approach
can achieve strong performance in benchmark datasets. Our model achieves FID
scores of 2.57 and 2.33 on CIFAR-10 in one-step and two-step sampling
respectively, surpassing those obtained from GANs and distillation-based
models. By extending the sampling to 1000 steps, we further reduce FID score to
1.79, aligning with state-of-the-art methods in the literature. For ImageNet
64x64, our approach stands as a competitive contender against leading models.

本文介绍了直接去噪扩散模型（DDDM）：一种生成逼真图像的简单且通用方法，具有少步采样的特点，同时保留了多步采样以获得更好的性能。DDDM 不需要精心设计的采样器或预训练的蒸馏模型，而是以自身的先前训练迭代生成的估计目标为条件进行扩散模型的训练，通过考虑前一时间步生成的样本来引导迭代生成过程。我们还提出了 Pseudo-LPIPS，一种对各种超参数值更鲁棒的新型度量损失。尽管简单，该方法在基准数据集上表现出强大的性能。我们的模型在 CIFAR-10 上分别以一步采样和两步采样的 FID 分数为 2.57 和 2.33，超越了 GAN 和蒸馏基模型获得的分数。通过将采样扩展到 1000 步，我们将 FID 分数进一步降低到 1.79，与文献中的最先进方法相一致。在 ImageNet 64x64 上，我们的方法与主要模型相当。