https://rentry.co/ProdiAgy
Uncaptioned, 20 images, 10 minutes. (5 of which probably spent being selective with images found on Google Images)

Try to describe her a bit if there are missing elements.

Should do the outfit quite a bit better, might have needed a bit more training but I'd rather undertrain than overtrain.

Still would like an even better dataset for her if somebody could provide as her outfit is highly dependent on it.

This one should do the outfit better probably, I put both this and the r34 Dataset version of this here so you can see how dataset affects training (duh).