This is a Lora of Lusty Lynx.
With the help of the discord community we was able to put together a lot of synthetic data to make and improve the model hopefully.
At this moment the model still needs testing and at least the images that was generated through testing so that there may be some things that can be fixed in the future with better quality data.
With how hard it was for me and others to generate good looking piercings for the nose and lips I don't know how well that will come out when generating images along with producing the snowflake that is in her eye as well.
I hope that for the most part the overall quality is at least better than what it was when I only had 24 images to begin with. Along with that I do not have to flip images to make more training data set of the asymmetrical nature of her design should hopefully come out correctly when prompted.
Along with that there had been some changes to the tags to invoke some of these features as well.
Description
This is a Lora of Lusty Lynx.
I do not have a lot data to use for training or at least I had trouble finding a lot of quality data that I could use so I was limited to about 24 images with only two of them being full-body which I then had to cut up to get a bit more data out of it.
With testing this Lora so far it is not for a beginner you need to be able to know some more techniques to guide the AI model to get to where you are going to and maybe have to finish with using ADetailer so that the details of at least the phase come out well.
When starting out with a small image of 720 x 1280 the details of the face was extremely horrible and it was not until I resized by 1.5 and then eventually using ADetailer within its own prompt did the correct facial details come out.
There was a request to make the lora smaller so I used the res-lora script to do so.
I have not tested to see how this affected the output.