This has become my testing character. As I improve my processes, I will periodically update this model. She is my first Lora and I have things pretty well organized so that I can compare different versions is easily.You may still prefer the original version. (I think I do, to be honest.) However, if you are wanting more realistic images V6 is a good choice.You may have seen my previous <a target="_blank" rel="ugc" href="https://civarchive.com/models/8049/olga-discordia-the-totally-wholesome-dark-elf-updated">Olga Discordia Textual Inversion.</a>I used basically the same process and captions for this. I did add "Olga_Discordia" to the beginning of them all like I saw in a tutorial.

Olga Discordia (LORA)

I'm getting very similar results as my embedding. I guess it trained faster though? I wasn't keeping track of the time. If it is faster, that would be amazing! It's been taking an hour for each of my embeddings.

Files with SHA256: b1e0992f00559f6f2781d3494dc32d00591d3689ed7458c1e4b91e707fb1d531