Where can i find dataset?
did you use anime screenshot or another image like fanart?
Both screenshots and fanarts are used. I used only a small training set for this particular model but the training set for the bocchi-the-rock franchise model is much larger.
would you like to give example at captioning for image that we will train? i really dont understand it
My captioning are very simple. It is basically just the output of a tagger + trigger words that I want to use.
This is how I curate my dataset https://github.com/cyber-meow/anime_screenshot_pipeline
I did upload my dataset for roukin model https://civitai.com/models/17336/roukin8-character-lohaloconfullckpt-8
and for umamusume otherwise https://huggingface.co/datasets/alea31415/umamusume_private
so like: futari, 1girl, uniform, short_sleeves...etc?
and what is trigger word in this ckpt?
Should be hutari according to my model card.