V2 notes:
Prompts: jpakari, brown hair, green eyes, long hair, high side ponytail,
Casual outfit: blue skirt, pink jacket, pink shirt, belt,
school uniform: school uniform, bowtie, brown skirt,
Lora strength: I think 1.2 gives the best results but lowering it to 1 or lower can also work
I always wanted to go back and remake some of my older loras, and thanks to an user I decided to give Akari another try. My goal was to reduce the number of tags needed to generate her and increase the accuracy, personally I think I succeed but you guys are the judges.
V1 notes
Prompts: Akar1, human, female
Optional: skirt, pink jacket, shirt / school uniform, long sleeves, brown jacket, bow, ribbon
Akari is a great protagonist, I love how much she grows and matures during the show, I haven't watch every season but Akari is definitely my favorite protagonist, her dynamic with Ruby and Labra (I don't think I will make a lora of her) are near perfect.
I had a few issues with Akari but I think she is good enough now. The hair on the sides can be a bit inconsistent and her outfit is also not accurate but I think the included optional prompts should generate her casual and school uniform close enough.
Description
FAQ
Comments (9)
This lora is very good and has a very positive reaction... but how come there aren't any images at all?!
That's interesting, I can see the images on my PC and phone in all my browsers and somebody commented on one of the images. The lora was uploaded while Civitai had some issues so maybe that's the reason why you can't see the images.
Can you see the images on other posts?
And thank you, I'm glad to see people like it.
@rotechacon667 No I mean there are sample images there, just not any of someone else using the lora to make an image
@KirbyGodMode my bad, Honestly I don't know. Hopefully somebody do because they also help me detect things I wasn't able to see with my initial testing.
I followed your suggestion for Prompts, but most of the generated characters do not have side ponytail or green eyes. To be precise, this Lora did not react.
After multiple experiments, I am certain that the weight needs to be set to more than 1,
@4P_lolicon interesting, I want to go back and remake a bunch of my old Loras so thank you for pointing that out. I'm going to look into it.
@rotechacon667 At present, setting the weight above 1.5 has a significant effect, while setting the weight to 2 is the most stable
@4P_lolicon Thanks, I see what you mean. I will try to fix it.