A model to replicate the style of Snapchat Bitmojis. Works best with DDIM at 20-40 samples with vertical image sizes <512px on the long axis. Works with most models, but the best results may come from SD 1.5 base.
NOTE: The sample images are from a pervious, unreleased full model, but LORA outcomes are similar in structure and appearance.
Description
LORA extract of full model (unreleased)
FAQ
Comments (6)
This is actually genius
It doesn't work for me, what checkpoint do you use with the lora? Also, any chance the checkpoint version of this will be released?
Hmm, that is odd. I used it with the bade model. I might release the checkpoint, but the current version gives very poor results
@lsed22998 I use Google colab so the base model isn't an option, maybe a realistic model will work.
@lsed22998 yeah I tested it on a realistic model and it worked but that means it doesn't work with characters trained in an anime style.
add freddy fazbear to the previews
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.