| Prompt | score_9, score_8_up, score_7_up, source_anime, 5 fingers, 1girl, black hair, straight hair, very long hair, (red eyes:1.2), (blush:1.1), smile, smug, horse girl, horse ears, horse tail, nude, thighhighs, red legwear, elbow gloves, reverse cowgirl position, girl on top, sitting on lap, sitting on someones lap, spread legs, cum in pussy, creampie, after sex, spread pussy, pussy open, 1boy, white man, faceless male, big penis, <lora:v6:0.5> couch, indoors |
|---|---|
| Negative prompt | score_6, score_5, score_4, source_furry, source_pony, text, patreon username, signature, watermark, fisheye, bar censor, mosaic censoring, censored, greyscale, monochrome, simple background, white background, dark-skinned male |
| Sampler | Euler a |
| Seed | 3886502199 |
| Steps | 26 |
| Cfg Scale | 6 |
| Model Hash | 67ab2fd8ec |
| Clip Skip | 2 |
| Model | ponyDiffusionV6XL_v6StartWithThisOne |
| Size | 768x1280 |
| VAE | sdxl_vae.safetensors |
| VAE Hash | 235745af8d |
| Version | f0.0.12-latest-114-g8316773c |
- ‹ prev Search: has:metadata next ›
Anonymous 06/03/24(Mon)02:07:13 No.8047392
00034-20240602201308.png (1.12 MB, 768x1280)
I got so coomfortable with Counterfeit V3.0 that I didn't even try SDXL until yesterday. Figured I'd try PonyXL with a single style LoRA.
It took very little to get semi-decent gens and it naturally understands all conventional poses, even after sex pull out which was tough to do on SD1.5 without very baked LoRA.
However, I noticed that it is inferior to Counterfeit V3.0 when it comes to simpler 1girl prompts. I assume it needs style LoRA at higher strengths or is that an inherent flaw of the vanilla Pony 6 checkpoint?