A peek into the DiffEdit paper
Fastai 2022 part 2
We’ll be looking at the DiffEdit paper and implement it with Stable Diffusion.
DiffEdit uses test-conditioned diffusion models to automatically generate semantic maps. This means that instead of generating the mask(semantic maps) manually we can generate them by only providing a text prompt of what we want to edit. Let’s look at an image from the paper to understand what this means.
DiffEdit uses test-conditioned diffusion models to automatically generate semantic maps. This means that instead of generating the mask(semantic maps) manually we can generate them by only providing a text prompt of what we want to edit. Let’s look at an image from the paper to understand what this means.
No matching items