Hello there!

I was a bit worried cause it was the very first time training an already existant character/concept (I use to draw or 3D model them). Some people requested if I can train some characters from videogames and TV series and well, here we are. I will be doing styles too but wanted to test my self a bit.

It was trained using internet images, some drawings here and there and polishing and fixing as much as possible (ironically I had to take out the two ribbons of the red bandana cause caused conflict).

Models tested: Faetastic, Dreamshaper 3.32 clipfix, GhostMix 1.2, Colorful v3, Rev Animated 1.22...

Token is: sfr1v and strength recommended is 0.8-1.X

This LORA is good at:
***********************
- Anime style output/semi realistic style outputs/concept illustrations of Ryu.
- Any type of portrait, side portrait, half body shot, face shot you can imagine.
- Create surroundings (if prompted, if not black and white background are designed for it)

- Mixing with other LORAs.

This LORA suffers with:
*************************
- Hands. Yup.
- Standing poses where hands are shown (will ???? the art)

So if you want to do some cool poses I'd extremely recommend to use ControlNet.

How to Prompt
******************
- Portrait/side portrait/half body shot/etc., sfr1v, environment, etc.
- If you use negative embedding CAUTION. This LORA was trained under specific exp. settings so you may need to drop bellow 0.7 or don't use token in the prompt.
- You can always invoke ryu with short black hair, red bandana/head band, white outfit/karate outfit even without using the token. This is specially useful to make advanced combinations in the prompting.

IMPORTANT!

I wrote an article here and it's like my finetunning diary for the ones interested with releases upcoming, trainings done, exclusive patreon models previews, personal thoughts, etc. So if you want to see what I'm planning or what I'm into, feel free to take a look.

That's all! Enjoy and feel free to support me in the Patreon

Thanks!