This is a custom LoRA model for the character Frisk from the popular video game Undertale. I trained the model using 84 high-quality images and utilized the AbyssOrangeMix3 model.
For optimal results, I recommend using the Euler a sampler, with the following recommended parameters:
CFG scale of 7
20 steps
hires upscaler set to Latent (nearest-exact)
20 hires steps
Denoising strength of 0.52
For optimal performance, it is recommended that the weight of the LoRA be set within the range of 0.6-0.8.
To accurately depict Frisk's appearance, I suggest including yellow skin, closed eyes, and expressionless in the positive prompt. Additionally, to ensure the correct number of stripes on Frisk's shirt, I recommend placing two two stripes or two purple stripes in the positive prompt and three stripes in the negative prompt.
It should be noted that this model has a tendency to add vignetting in the corners of the output image. To mitigate this, I suggest including vignetting in the negative prompt.
Please feel free to leave a comment below if you have any requests for LoRA models.