First Model, I tried my best.
But as the name suggests. It's Lunala from Generation 7 of Pokemon.
Trigger words are "Lunala" and "Legendary_Pokemon"
For more, better results I recommend trying to mix the furry tag, and use yiffymix as the model for more, proper intended results. As, proper as it can be. It's not 100% perfect but it works just enough to make it not human, for whatever reason
Do note, you are entirely allowed to improve on this Lora, as I am new to this, and if any of you can make this better and would like to, then go ahead and be my guest. And I simply made this out of boredom.
(PS. This was trained on PixAi, due to computer limitations, both preventing compiling anything and not meeting requirements to run more powerful generation software. Thank you for reading have a good day or good night)
Description
FAQ
Comments (6)
also, what model was this trained on?
It was trained on PixAi
@ThetaStormer287 yeah but what checkpoint was picked to train on.
@LOS_FORRY_CUSTOM Oh sorry, I trained it using the Anything V5 base model there. But used YiffyMixV34 with generating images, since it seemed to work a bit more better with what I had wanted the actual results to be. Sorry if I didn't understand that easily, I'm new to this. And I'd consider super inconsistent, due to the limitations. Not exactly running the best of computers to train a much better LoRA, so this is not the most accurate. But at the same time, I don't think anyone would be making a more accurate one anytime soon, lunala is a bit, underrated and hasn't been noticed AS much
@ThetaStormer287 Thank You! I usually don't ask this since most model have metadata inside them but for some reason loras made in pixai don't have any metadata.
also i agree, lunala is underrated nowadays
(EDIT: It appears it was a result of the site being janky and not allowing new LoRAs to generate sometimes, and needing to be fully published first in order to use to any extent. When I have no means of fully testing my LoRAs)
It appears that If I had wanted to ever update this with a better version in the future. I would not be able to, due to an unexpected issue, that seemingly popped up out of nowhere, and is preventing me from making new LoRA's by preventing ALL image generation, claiming it "Could not load image" and then give an error in the generation, but no message on what the error was. Meaning any attempts I try would mean absolutely nothing, and only waste hours of my time trying over and over just to get one to work. And I was not having this issue when testing this LoRA. So until there is an explanation, or a fix on Civitai's end, I am entirely unable to make any new LoRA's or update this one for the foreseeable future.