UPDATE: 20250529 - I didn't know this model was going to take off! The 1.5 version is just a dataset, I believe!
It was originally created to kind of be a "jumping off point" for the style of art that I like to create, but now it's apparently more!(Let me apologize first for anyone who's used this and has gotten inconsistent results - I thought it was just for me!)
Working on updating this model with what I know now as opposed to what I knew 6 months or so ago (a substantial amount, IMHO).
I hope to publish a new version in a few days, but you know how that goes). Anyway, I'm glad it's being used (and at the same time ashamed of myself for publishing too early)
First attempt at a style LoRa. I simply created a dataset of various character individual portraits and group pictures, added in some demi-humans, D&D characters and "monsters" and assorted other fantasy art creations (all created by me - no scraping or blatant thievery involved).
If you use it, let me know how it works with your character art images! I'm especially curious about how it will deal with multiple subjects!
Thanks in advance!
Description
Same dataset as the first, but with better preview pictures
Since this style does really well with fighting scenes (not as simple as it sounds), I'll be updating it regularly. Feel free to post anything you've made with it and I'll toss them in the next training run as well!
FAQ
Comments (2)
I think, you uploaded only the training data for 1.5.
Yeah - I was hoping to get v2 in before the contest deadline, but there just wasn't time.
I've got the Arbiter character v2 ready to publish (I'll do that in just a second) so my GPU is free to begin running the training for v2 of this. Hopefully, it'll be ready by the end of the day (these Flux LoRas take a long time to train!)