v4.0
(The NEW label is misleading, it's an update.)
Trained with the goal of enabling multi-subject capability but didn't turn out well. At least there's a positive probability of multi-char working. Still need to figure out why it fails.
In addition to the character tags, custom style tags were added:
fanart: fan arts from Danbooru (optional, neutral style)u149ani: screenshots from the airing U149 animeapuri: Cinderella Girls and Starlight Stage mobile game art (somehow dominated by early CG style, will probably retag in the future)
As the outfits were not manually tagged, it should not be expected they can be replicated.
Training data can be found in my data repo. Training details will be properly documented when I have time. Would appreciate suggestions or any sort of feedback.
Description
Added anime screenshots and fanarts
Added Yonai P
Added trace amount of multi-char capability
Renamed tag: Yuki Haru -> Yuuki Haru
FAQ
Comments (3)
Added you to my girlpack bookmark page. You rock!! Keep up the great work!!
It shows an error like:
RuntimeError: The size of tensor a (1024) must match the size of tensor b (768) at non-singleton dimension 1
A little late but if you happen to see this
The dataset for something like this is insane. I've been training my own characters(mostly singles) with a few multi concept Loras for about a month and my dataset is usually 50-100 screenshots from shows with some fanart mixed in.
I'm really curious about the process behind making multiple concept models and getting better models overall. Are there any resources, guides, tools, or anything you could point me towards?
Any help would be really appreciated. You do great work, appreciate you and the others that make this so fun :)
I'm really stuck on things like:
How do you go about sorting each of the outfits into categories like "arsfste, arsstrc, etc.)?
How many images for each girl would you typically use for something like this?
How do you keep all the tags from overlapping with each other, when training or is it not really an issue?
Details
Available On (2 platforms)
Same model published on other platforms. May have additional downloads or version variants.