Diffusion models: DDPM, DDIM and classifier-free guidance
Diffusion models solved what GANs could not: stable training, high diversity, and fine-grained controllability. The core idea is elegant — learn to reverse a noise-adding process. Add gaussian noise to an image 1000 times until nothing recognisable remains, then train a neural network to undo one noising step at a time. At inference, start from pure noise and let the network denoise its way to a coherent image.
Content is available with subscription.
Get full access to all courses on the platform for one year with a single payment.
▼
Unlike other platforms that charge per course, here you get everything for one price, and after one year of use there will be no automatic charge for the following year.