Does anyone know of any good review papers or blogs that explain the various ways to determine how many steps are enough to get a high-quality output image from diffusion models?
Maybe related question is how often is a fixed number of steps good enough when using these models?