Get early access to build and test build, be able to try all epochs and test them by yourself on Patreon or contact me for support on Discord
Kohya GUI, 150 Images, 30 steps.
You are responsible for the images created by you. Unethical usage of this LORA is prohibited.
I strongly advise against uploading any NSFW generated by this model. I warmly welcome you to communicate with me here on how to improve the working method to make the model more realistic. I am currently considering how to generate a better version 3.0. If you have any suggestions, such as choosing different base models, training with more images, increasing the number of training steps, or changing the training size to achieve a more realistic overall effect, please feel free to leave a comment below.
Description
FAQ
Comments (4)
I can suggest removing any watermarks on images you use to train it with.. will really help with that
actually i just checked and there was no water watermarks on the trained images, btw this surelly isnt my best model. not sure about what went wrong in this training, or maybe my prompts are not suited for it, if you have any good prompt for her please post it :)
oh wow.. thats actually fascinating then.. so SD somehow edits and adds random watermarks onto pics that didnt even have it.. wonder how that can be erased. negative prompting doesnt always work with that.. thx for the feedback, that was super interesting
@DukeNukem47 also i can tell you that i did a few models that had many trained images with simple watermarks and they were not appearing in the generated images. so i really have no idea how sd decides what to generate xD