Denoising diffusion models, also known as score-based generative models, are a class of generative models that can produce high-fidelity images, videos and 3D models from noise. They define a forward diffusion process that gradually adds noise to the input data, and a reverse diffusion process that iteratively denoises the noise using a learned score function. In this talk, we will introduce the foundations and applications of denoising diffusion-based generative models. We will cover the theoretical background, the training and sampling methods, the recent advances and challenges, and the practical use cases of these models. We will also demonstrate some examples of image/video/3D generation using denoising diffusion-based generative models and discuss their advantages and limitations compared to other generative models.