Well, today I'm sharing the first "realistic" Lora I've trained.

I gave the girl a generic name, MORENA, and that's what it stuck with xD

Honestly, for the first Lora I've trained with a realistic image dataset, it turned out pretty well. Even though it sometimes looks a bit static, I've added images so you can see that it can adapt to any medium, and it's all about finding the right prompt.

Also, even though it's trained with many images of the white dress as a base, the Lora perfectly accepts other types of clothing without any problem.

On the other hand, the Lora works wonderfully with non-realistic checkpoints and accepts anime or semi-realistic styles without issue, in fact, giving very good results.

The key is finding the prompt, the checkpoint, and the Lora styles that you like and that give you results you're happy with.

This time, there's no Illustrious version. At least, for now, I didn't like the results offered by the different training epochs.

Please leave your feedback and your creations with Morena.

Enjoy!!! ^^