Type | |
Stats | 291 409 122 |
Reviews | (72) |
Published | Feb 24, 2025 |
Base Model | |
Training | Steps: 1,600 Epochs: 13 |
Usage Tips | Clip Skip: 2 Strength: 1 |
Trigger Words | wfus3 |
Hash | AutoV2 5193E3D544 |
v.2:
Trigger: wfus3
Negative (optional): anime
Now that I know more what I'm doing with illustrious, I created >250 images for the dataset, concentrating on expressive faces and dynamic poses. I really like how those aspects turned out in this one.
v.1:
Trigger: artist:schmedefus3
Some checkpoints, will require negative prompt of "anime" or increasing strength, the listed checkpoints are the ones that do not.
This is just a LoRa I made to align more with my preferences, which is slightly more realistically proportioned faces that are more typical of western art, with outlines and soft coloring. It might not be that much of a shift to some, but it prefer it.
The entire dataset I generated running through multiple detailer nodes and I included many images of correct hands and close-ups with different expressions. The goal was to increase facial detail and to get hands correct more often while shifting the art styles closer to my preferences.