Prospect on integrating Diffusion Models already have a semantic space paper?

[Skimmable-Code-pls]

Their code is provided and implemented on top of Base Diffusion and Elucidated Diffusion setting, they proposed a controllable reverse process less prone to undesirable distortion, which I think greatly beneficial for Imagen trainers developers.
https://github.com/kwonminki/Asyrp_official - Also their paper is top 25% ICLR so they are popular