Lora upload

[Rhumrehtona]

For those of you who play around with sdxl, I have uploaded the first version of my lora.

https://civitai.com/models/264194/quicksand-xl?modelVersionId=297872

It works best at strengths around .7, but will work higher and lower than that.
You can add additional words before or around quicksand to change the look, adding sandy, swamp, etc. It also works without specifying the trigger word (quicksand) if you are using comfyui and have it loaded in your workflow. So you can just use the prompt 'woman sinking in brownie batter' etc.

I did not use any copywrite material, so making it was a slog, and it is more limited than it would be if I had.

g_00078_.png

g_00077_.png

[MadMax359]

well, SHE is worth another look

[cerberus]

That looks pretty good, I haven't investigated LORAs yet and I'm using SD rather than SDXL so can't really try this at the moment. This is an area I intend to look at in the future (both XL and LORA), time and resources allowing. Anyway, well done for achieving this, I suspect there was quite a lot of pain and torn-out hair to get this far.