Type | |
Stats | 12,901 807,223 |
Reviews | (2,512) |
Published | Feb 21, 2024 |
Base Model | |
Usage Tips | Clip Skip: 2 |
Hash | AutoV2 529C72F6C3 |
That's a lot of consonants.
As this is a merge based off PonDiffusion v6 I HIGHLY recommend you read the description of that model for information regarding prompting, tags, and baked in characters.
Released in collaboration with FallenIncursio. Check his new Meme Diffusion for Pony!
TL;DR Generation Settings:
Sampler - Euler a
Sampling Steps - 25
I upscale using this process and the upscaler below:
Dimensions - 832x1216, 1216x832, or 832x832
CFG Scale - 7
One pass of ADetailer
face_yolo8n with nothing in positive or negative
This might not be needed so much on XL
I use the following positive and negative prompt. I DO NOT USE ANY SOURCE TAGS (source_anime, source_furry etc) :
score_9, score_7_up, BREAK {prompt}
NO NEGATIVE PROMPT OR EMBEDDINGS REQUIRED
3d, realistic in the negative will give a more anime look
watermark, patreon username, artist name, signature, text can help with watermarks generating if you run into that issue
Spiritual successor to my SD 1.5 Doll model. Merged using a LoRA from images generated by myself on my old 1.5 models blended into the exceptional AutismMix_confetti. This is a pretty horny model with the primary goal of generating female characters. As such, it may not be suitable or flexible enough for all scenarios and use cases.
The model has a tendency to draw teeth on characters with tags like smile/grin/smirk. Putting teeth in the negative can help offset this. See sample images for more examples of prompting.