Hello and thank you for the amazing work you've done on this SOTA text2images. After testing the HF demo I noticed the super-resolution 256 -> 1024 struggle to give good results. Isn't it possible to introduce a middle step like 256 -> 512 -> 1024 instead?
Hello and thank you for the amazing work you've done on this SOTA text2images. After testing the HF demo I noticed the super-resolution 256 -> 1024 struggle to give good results. Isn't it possible to introduce a middle step like 256 -> 512 -> 1024 instead?