Read about the diffusion model we can use - Paper

[talafek96]

• Model we can use:
  • Paper https://arxiv.org/pdf/2105.05233.pdf
  • Code https://github.com/openai/guided-diffusion
  • Specifically, we want the version which is not “classifier guided”
  • Better to start with the low-resolution version for faster experimentation
  • E.g. “64x64 diffusion”

Related to epic #1

[talafek96]

Take a picture, add noise, and find the maximal time step for which you are still able to reconstruct the noised image.