Short version: Version 2 is preferable over V1 and tends to be a net improvement. V2 alternative is based on some images the model's commissioner gave me. I'd say pick either V2 or V2 (alternative).

This is the Illustrious version of my Miaka Pony LoRA. It uses the same dataset, so similar images should be possible. Obligatory self-promotion: consider supporting or commissioning me at https://www.patreon.com/MetalChromeX or http://www.fiverr.com/s/VY34o4m so I can afford an Nvidia 5090 when I can buy one priced at or near the MSRP (screw you, scalpers!)!

Update (4-9-25): Based on feedback from the model's commissioner, I made some tweaks to the V2 model; I'm calling this version V2 (alternative). V2 was trained only on anime images, while this alternative version was trained on some images the commissioner gave me.

Update (4-6-25): I added an improved V2 with more and better images in the dataset and removed some questionable quality images. I'd recommend V2 over V1.