After yesterday's embedding attempts, I've retrained Asanagi and Henreader on further curated sets, as well as training Araizumi Rui. Compared to the overtrained stuff I posted before, these embeddings should be more flexible and consistently higher quality.
Pic related uses all three of the new styles.
https://mega.nz/folder/UUgD1TYL#UwpvsiR2X3w-3tm3hck_VQEnjoy!