V2

The second version, I managed to get the model to focus less of the face when images are generated, so it's more likely to give better full body images, though it seems in the process I've lost the details in her eyes, mainly the black sclera and orange pupils generating properly, though it can be worked around. Will probably be a V3 before I'm truly happy with the final model.

V1

The demon merchant from dungeon odyssey, because she isn't really the focus in a lot of chapters, the dataset was really more focused on her shoulders up, so it tends to make images with face focus. I managed to get all her facial details and hair details and intend to use the lore to create some full body images to increase a dataset for a V2 to fix the problem.

I find that removing the waistcoat prompts and them prompting heavily for her bottom half helps me make full body shots. You can also remove alot of the trigger prompts to help generate, as long as you keep the face specific prompt it generates well. If you have any good images generated that have her outfit properly generated, if you don't mind, can you post the image to the model page so I can sort through and use the best for my next dataset.

Recommended strength around 0.7.

If you enjoy my work, please consider making a small buzz donation, or posting images directly to the model page, I can't always train directly on my pc as I use it for work, so I sometimes use the Civitai trainer and every little bit helps me make new models that you can enjoy. If you have any character recommendations, don't be shy and DM me the characters.