Have you ever experimented with the SDXL baseline instead of the SD 1.5 baseline? #110
Description
Hello! Thanks for sharing the great code.
I'm very interested in the techniques that make anime images dance! There are many other codes, but the Moore is the most accessible and easy pick for me.
However, I'm worried that the SD 1.5 model might be too old compared to the recent Nano Banana and Flux models.
When I look at other codes or papers, it seems like most of them use SD 1.5, except the video diffusion models.
SDXL seems to outperform SD 1.5, so i'm now considering using SDXL instead of SD.
Even if it's not Flux, have you experimented with SDXL?
Also I'm wondering if I can get a more satisfying video by using SDXL!