Awful, awful video quality. Ran dataset through img2img to clean it up.
Recommended prompt: ht_hikaru, glasses, under-rim eyewear
Default outfit: school uniform, blue skirt, brown vest
Description
FAQ
Comments (5)
I got a High Quality version of the images/OVA that you can use for an UPDATED version of this model.
Here is the most HQ versions of the images I could get: https://www.mediafire.com/folder/7kr73fja2aen5/For_Sakajou_Hikaru_LORA
My LORA of Sakajou is inferior to yours anyway, so I figured, why not help yours become better by sharing some of my files?
Thanks for sharing this, I appreciate it. I did think about redoing some of my loras now that I've gained a bit more experience training them. If I redo this one, I'll compare my dataset and source video to yours and perhaps use a mix of both.
@happy_tentacle Cool to hear. It'll be cool to see what you do with both Datasets of Sakajou. Body Transfer is underrated in my opinion. (If you want, I can also provide a dataset for Miho too.)
If you enjoy making loras, I don't think you should give up on improving yours. It's not just about having trained the best lora possible for a character but also learning how to do so, which you can then transfer to your next training. I've had to deal with shitty video quality and very small datasets to make some of my loras, and it only made me stronger :)
@happy_tentacle Thanks for the encouragement. Honestly. But I don't have enough buzz to create my own LORAs at the moment. But this is quite helpful to know about.
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.