An Exo Stranger Lora model trained on a little over 60 images.
Works pretty well with a CFG somewhere between 3-5. Can be more or less depending on the checkpoint or other Loras you're using it with.
I'd recommend using hires fix to help get her face right. Its more needed when the image is like a full body or one where she is outside the immediate foreground.
If you want a wider variety of detail for her body and clothes, pairing this model with RobotGirlLora https://civitai.com/models/32497/robot-girl-lora can help quite a bit. (Mainly for )
I figure I should also mention that there aren't a ton of quality images depicting her in different poses and clothes, so the image selection is pretty tight. Not going to lie a good number of the training images are thanks to StrangerSFM's work, https://twitter.com/strangersfm ().
Anyway, thanks for giving it a look and I hope it works well for you.
Added/refined new and existing training images with more straight forward captioning
Broadened capabilities and compatibility with other loras/checkpoints thanks to new tagging scheme
Refined training settings to increase realism and detail
Unhealthy amount of testing hours
1. I diritti dei modelli ripubblicati appartengono ai creatori originali.
2. I creatori originali che desiderano reclamare il proprio modello devono contattare il personale di SeaArt AI tramite canali ufficiali. Clicca per rivendicare
