SDXL working but output quality is very poor

[pushstartvr]

Hello, unsure where to post so I just came here. as the title says.
Single image generation is great compared to motion module generation, just like v15 for 512x512, however the output for SDXL is much worse
SDXL video:
https://github.com/continue-revolution/sd-webui-animatediff/assets/25425449/c00bf120-8836-4a12-a90a-9565e8cfea2d
SDXL: single image

its amazing though, centre frame subjects at high res, cat wait til I can get it working properly

[continue-revolution]

Unfortunate, just as what researchers from Shanghai AI Lab have named their model - “beta”, it is not very stable in terms of performance. A good news is that they will release more models in the future, which I expect to be more robust.

[pushstartvr]

I'm no expert though so maybe I am doing something wrong. I've only been on animatediff for a week or so on Auto1111 webUI, haven't attempted comfyUI at all