Update (8-17-25): The person who provided me the original dataset provided me 34 (heh) new images. Since this is about as good as it's going to get to replicating her appearance, this will be the last revision I do (unless they formally commission me to do a V3 and provide me more images; I did this one for free because I thought Shizuku looked super cute).
Update (8-10-25): I noticed some of the test images were outputting black bars on the side. I found that a good chunk of the screencaps given to me were not cropped to remove the black bar, due to an unusual resolution. I manually cropped the black bars out of the images in the dataset, so V1.1 should be less prone to such issues. Otherwise, V1.1 has the same dataset as V1.
A CivitAI user sent me 100 high-quality images of Shizuku. Since she's cute, I made a model of her! Enjoy!
➼Obligatory self-promotion: consider supporting or commissioning me at or so I can continue making cool crap for you all to enjoy for free!