Kardia of Rhodes XL - PONY

UPDATE - I've created a handy resource to fix Kardia's hearts on her helmet and tabard. It's best used with photoshop or a similar graphics software tool, but if you don't have access to that you can just use a free program like gimp to make it look roughly correct and then img2img or inpaint it to blend it in.

https://drive.google.com/file/d/12CIt6qxuR4hxuXKucrN3HNnZMfp2rJD7/view?usp=drive_link

I was saddened that no one had yet brought this lovely character to Pony Diffusion. So I decided to fix that. Let's you generate the famous slutty medieval crusader Kardia of Rhodes.

Version 1.0 - A very early version that some freaking how captured Kardia's likeness while still remaining very flexible on only the 3rd epoch. Weak on fine details though. I was unable to replicate it on newer versions (Note I did most of my testing with the tabard tag. After trying without it the lora seems less reliable).

Tags

• Kardia, tabard

• Kardia, {Experiment with adding tags like armor, helmet, plume, plumage, thigh highs, breastplate, and breasts.

• You can also add "thong, panties, underwear" which was an optional detail in the training data.

Version 1.1 - I experimented with using more training tags on this one. It didn't really take until the last epoch, so it's a bit overtrained. Decided the keep it because I didn't know if the extra tags would possibly help anyone with generations or not.

Tags

• Kardia, armor, helmet, plumage, heart cutout, tabard

• Kardia, armor, helmet, plumage, heart cutout, breastplate

• Kardia, armor, helmet, plumage, heart cutout, breasts

• You can also add "thong, panties, underwear" which was an optional detail in the training data.

Version 1.2 - Final version I made, with what I consider to be optimal tags. I want to say this one captures Kardia's details the best. But like 1.1 it's a little overtrained.

Tags

• Kardia, Breastplate, Unconvincing armor

• Kardia, Tabard, Unconvincing armor

• Kardia, Breasts, Unconvincing armor

• You can also add "thong" which was an optional detail in the training data.