Closed 823863429 closed 6 months ago
Hi, you may have to use your lora's corresponding tokens in the prompt to obtain optimal results? Hard to know based on information provided - hotshot-xl definitely works with LORAs.
Perhaps ask in the Discord and someone can help!
Thank you for your outstanding work. I specified a lora file when generating, but the generated content has nothing to do with the style of lora. Is there any problem when I use it? lora is trained according to sdxl. Here's how I use it: python inference.py \ --prompt="a camel is smoking, hd, high quality" \ --output="output.gif" \ --lora="/content/hotshot-xl/3d_keai_2.0.safetensors"