Female_Au_Ra-FFXIV

This model is a work in progress. It isn't perfect, but it can generate some nice images. I have trained Xaela (dark horns and scales) and Raen (light horns and scales) as separate concepts, which will allow you to specify which you would like in your image by using the tokens specified below.

I hope you enjoy it as much as I have! I'd love to hear your feedback.

Waifu-Diffusion-v1-3 based StableDiffusion model with Dreambooth training on images from many different artists. The model is trained to 11,000 steps on 80 different images of Au Ra females, a playable race of people with dragon-like horns, and patches of scales from the critically acclaimed MMORPG Final Fantasy XIV (Have you heard about the games free trial btw?).

Usage

Can be used in StableDiffusion, including the extremely popular Web UI by Automatic1111, like any other model by placing the .CKPT file in the correct directory. Please consult the documentation for your installation of StableDiffusion for more specific instructions.

Use "m_arxla" for Xaela clan Au Ra or "m_arrn" for Raen clan Au Ra in your prompt to invoke the style of the desired clan.

Recommended negative prompt

"poorly drawn, bad quality, colored skin" You can also add the following to your negative prompt in order to steer your output towards the WD1.3 default caucasian skintone: "blue skin, purple skin" If you are generating Raen Au Ra I highly recommend also adding "black scales" to your negative prompt as the AI will often draw dark scales without it. I do NOT recommend adding the common negative prompt tags such as "bad anatomy, disfigured, deformed, gross, etc..."

Example prompt

"m_arrn, 1girl, light smile, detailed eyes, extremely detailed face, sidelocks, black hair, grey eyes, long hair, hair clip, collarbone, tank top, shorts, looking to the side, highly detailed face, extremely detailed, intricate, best quality, ultra realistic, cowboy shot, holding shopping bags at the mall, highly detailed background" Negative Prompt: "poorly drawn, bad quality, colored skin, blue skin, purple skin, midriff, black scales"

Sampler: DPM++ 2S a, Sampling Steps: 22, CFG scale: 11, H: 768, W: 512

Xaela example images using "m_arxla"

<table> <tr> <td><img src=https://i.imgur.com/gvgZeT1.png width=100% height=100%/></td> <td><img src=https://i.imgur.com/wWFDxCS.png width=100% height=100%/></td> <td><img src=https://i.imgur.com/yzWhulJ.png width=100% height=100%/></td> <td><img src=https://i.imgur.com/TjTIbIz.png width=100% height=100%/></td> </tr> </table>

Raen example images using "m_arrn"

<table> <tr> <td><img src=https://i.imgur.com/jwoWZWE.png width=100% height=100%/></td> <td><img src=https://i.imgur.com/k1XPAZI.png width=100% height=100%/></td> <td><img src=https://i.imgur.com/MvcSlAd.png width=100% height=100%/></td> <td><img src=https://i.imgur.com/3PLuE1V.png width=100% height=100%/></td> </tr> </table>

License

This model is open access and available to all, with a CreativeML OpenRAIL-M license further specifying rights and usage. The CreativeML OpenRAIL License specifies:

1. You can't use the model to deliberately produce nor share illegal or harmful outputs or content
2. The authors claims no rights on the outputs you generate, you are free to use them and are accountable for their use which must not go against the provisions set in the license
3. You may re-distribute the weights and use the model commercially and/or as a service. If you do, please be aware you have to include the same use restrictions as the ones in the license and share a copy of the CreativeML OpenRAIL-M to all your users (please read the license entirely and carefully) Please read the full license here