Type | |
Stats | 255 |
Reviews | (37) |
Published | Aug 10, 2023 |
Base Model | |
Training | Steps: 1,220 Epochs: 10 |
Usage Tips | Clip Skip: 2 |
Trigger Words | Queen Gna |
Hash | AutoV2 B8AD320252 |
Hello you beautiful nerds, thought I was gone for good didn't you?
Note: I am working on a pc that can only handle resolutions that are 512 or under, I do not know how will the lora will work in higher resolutions.
I plan on making more of these. To be completely honest this dataset was trash, I have no idea how the AI managed to pull this one off but I'm very grateful I was able to get something decent out of it. Anyway, I have a lot more to make and I would very much appriciate your support through my Machine Learning Journey, what I'm trying to say is give me money.
My Ko-Fi: https://ko-fi.com/keyennorman
You can recommend me some LORA ideas but I'd be more likely to go through with them if the bag is right. I am considering doing commissions so if you're interested you know where to find me. And hey, maybe I'll be able to afford me a new pc so I could create better LORAs, that's some wishful thinking though.
Lets go over some prerequisites:
1. I'd prefer a front view above all else as well as the tag: "Highly detailed face: if you are using this LORA with a photorealistic model as the AI will fail to accurately depict the face in most cases. If you generate a side or back view the wings will most likey not be layered correctly or just fail to generate properly entirely.
2. Activation Tag: Queen Gna | Please take a look at the images for a reference prompt.
3. Weights could range from 0.6-0.8 (I use 7 mostly). The fluctuation really depends on the model's style, for cartoon it may work on 0.6 but for realistic models 0.7 could do fine. Again, it depends.
That should be about it, see you later nerds, peace!