Hello, I would like to share a problem and get advice. I recently decided to learn how to create LoRA, and it seems like I understood the most basic theory, but the practice... not very much.
As for me, my needs are small, just a LoRA of a character with specific features of appearance. In a way, I would even call it a "bundle of prompts" that describe the character.
But in practice, I create ineffective LoRA, which to a large extent does not fulfill the main task, but displays the style of the images of the "raw material" that I used to create LoRA. For training, I use 60-100 images that are generated on this site. If something comes out, then very weakly. Here is an example of the "most successful" (no): https://civitai.com/models/1698576?modelVersionId=1931992;
I would like to know what I am doing wrong and get advice on how to do it better so that LoRAa is viable.