/vt/ - Virtual Youtubers

>>62283663
Sure, here's an example:
https://pixeldrain Holy shit why does 4chan hate pixeldrain so much?
com/u/jhhaETej
It's not great and not terrible. You would want as much pose diversity. Full body, upper body, close-ups, from the side, from behind, doing stuff with their hands. But it can work even with worse. I can change the backgrounds if they are too same-y.

Considering it came to this: Announcement! If anyone wants to train loras but doesn't know how to train, feel free to make a dataset, a collection of images of the character like the one I linked, and I will train it for you because fuck Civitai taking training out of their subscription model! I can do the tagging so you can just cover the images.
I will not do styles or concepts though because they require too many images and steps.