I wanted to create a Bluey style with more flexibility. This was trained on 250 images ranging from on-model character sheets, 3D renders, backgrounds and rendered art styles to make the LoRA more compatible with more creative prompts, while maintaining character likeness.
Main trigger:
Blueyxl
Style trigger:
boxchibi
Character trigger:
bluey heeler
bingo heeler
bandit heeler
chili heeler
muffin (bluey)
Description
FAQ
Comments (7)
Hi. Would I be able to get the dataset with tags so I can produce a YiffyMix XL & Yiffymix v34 version of this model? I'd leave a credit & link to your profile & model. I love this bluey model, one of my favorite series of characters to use and I'd like more model versatility.
Sorry for the late reply! Sure thing! I haven't uploaded training data before, but I have seen them under the drop down on certain models.
If you can direct me as to how to upload it here, I'll add it! It wouldn't be a bad idea to take some of the best generations below, and add them to your new LoRA.
@jasonafex Oh wow, sorry. I missed your response. Came back today to actually ask about training an Illustrious model instead. I'm absolutely still up for it.
I don't think you can add training data later after uploading the model, you'd have to upload a new version of the model and just call it "dataset"
@BenjiLORA Gotcha, I'll add a [Dataset] now, cheers!
@jasonafex Training!
Could you separate out the characters? I try to generate Bingo but it keeps generating Bluey, Chilli or a weird mishmash of all the characters features, creating random spots on the fur, misplaced colors, etc.
Works very well, generally. However, lots of characters get mish-mashed together.
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.