PDXL 1.1 Version
Now with half the bytes and more finger tips! I was a bit bothered by the generations from the first PDXL version coming from something overcooked. I'm a bit happier with my settings now so I can finally start retraining my other LoRAs. Enjoy!
PDXL Version
Hey all! It's been a while but I've decided I might start working on making PDXL versions of my LoRAs. This first one is just a test run but I think it turned out well enough. Let me know what you think and if there are any of my old LoRAs you want me to prioritize updating let me know.
Please note that all of the included images had adetailer used as part of their generation.
Also, no I have no idea what that last image is. I just did a bunch of wildcard gens and thought that one was hilarious.
Old SD 1.5 Version
I made this LoRA a while back, but it by far the best of the ones I've created. There used to be an embed or hypernet for aaaa, but since there doesn't seem to be a LoRA yet I made this.
It was trained on Anything V3 if I remember correctly, but all the example images are from v4.5.
You'll likely want nsfw in the negatives if you are concerned at all. It was trained on all of the non-censored/uncensored aaaa works so if you're interested in what the model can do take a look at their art for inspiration. The artist covers a wide range of body shapes.
Description
Retrained to work on PDXL and its derivatives
FAQ
Comments (8)
Can you do Anyannko and Neneneji next?
Someone requested the Tsukino LoRA so I just finished that one and am doing testing now. I'll start one of the two you requested as soon as I'm done with these tests!
Sorry for the delay I spent way to long testing the retrains of both the Tsukino LoRA and this one. I'm gonna flip a coin and start the training on one of your requests now haha
No problem, thank you!
It's a nice lora... but man it's insanely large file size
Oh I didn't even notice! Sorry about that I forgot to adjust the settings. I'll probably end up retraining it at a smaller size. I was just more focused on the quality than the size. When I'm finished with the retrain of them all I'll go back and make smaller versions.
It's still rather large, but here it's half the size now. I definitely can work on reducing it more, but I don't want to keep retraining the same LoRA so I'm sorry but I'll be coming back to this in the future!
@thesauceer For reference though, for your other loras you'll be working on, I do highly suggest on working to get them even lower, cause not only will they be smaller file size, but it will train in less time too, from my experience with them anyway.