Made a lora off of some images i saw on tiktok cause i really liked the style of them. overall it came out pretty good results can vary on what model you use and its recommended that you generate bigger pictures then small ones. Its not to good at making small images
use <lora:Fantasy:1> inside the prompt
Update: removed the signature that would show up in bottom right of generations
Description
Retrained the Lora to remove the signature that would show up in the bottom right. Lora might have slight difference in results now compared to the first version.
FAQ
Comments (7)
This is cool, im not going to leave a review yet because it's a bit prone to signature (obviously the same artist, same sig in the dataset... you might wanna either crop out the signatures or photoshop them out then cook it again.
Absolutely in love with this. So freakin cool. Unironically the model that got me hooked on generating art instead of boobies.
I cant use it?
A tensor with all NaNs was produced in Unet. This could be either because there's not enough precision to represent the picture, or because your video card does not support half type. Try setting the "Upcast cross attention layer to float32" option in Settings > Stable Diffusion or using the --no-half commandline argument to fix this. Use --disable-nan-check commandline argument to disable this check.
Tried disabling and enabling this but still doesnt work
I am somehow not getting good results, I cant even remake the preview images with the generation prompt, even tho I have anything midjourney
Hey! Where can I find the model you're using? This looks amazing
as your attorney, I advise you to go watch the high on fire video for "the black plot" then make something with this lora
Do you think we can get a Pony Version ?
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.