<lora:drddsp:.9>, drddsp, 1girl, 1boy, nude, pussy, open mouth, black hair,
21 year old,
perfect skin perfect model face,
European supermodel ,
best quality ,
RAW photo, subject, 8k uhd, soft lighting, high quality, film grain, ,
realistic,
Negative prompt: cartoon, sketch, east asian, disfigured, ugly, low quality, CGI, 3d, animated, duplicate, bad anatomy, bad proportions, trans, futanari, full-package_futanari, penis_from_girl, cropped, jpeg artifacts, lowres, out of frame, worst quality, (((deformed))), [blurry], (poorly drawn hands), (poorly drawn feet), child,
Steps: 30, Sampler: Euler a, CFG scale: 6, Seed: 2917139924, Face restoration: GFPGAN, Size: 512x512, Model hash: 15012c538f, Model: realisticVisionV51_v51VAE, Denoising strength: 0.7, Hires upscale: 1.2, Hires upscaler: ESRGAN_4x, Lora hashes: "drddsp: 69608f962920", Version: v1.4.0
<lora:drddsp:.9>, drddsp, {1girl, |}{hetero, |}{1boy, |}{ass, |}{sex, |}{nude, |}{penis, |}{pussy, ||}{lying, ||}{solo focus, |||}{uncensored, |||}{realistic, |||}{thighhighs, ||||}{vaginal, ||||}{dark-skinned male, |||||}{long hair, |||||}{interracial, |||||}{dark skin, |||||}{anal, |||||}{on side, |||||}{lips, |||||}{closed eyes, ||||||}{open mouth, ||||||}{breasts, ||||||}{nipples, |||||||}{black hair, |||||||}{garter straps, |||||||}{pubic hair, |||||||}{makeup, |||||||}{ass grab, |||||||}
21 year old,
perfect skin perfect {face |model face},
{American| European | nationality} {supermodel |model},
{masterpiece | photorealistic | best quality | 4k | 8k },
{RAW photo, subject, 8k uhd, soft lighting, high quality, film grain, },
realistic,
Description
FAQ
Comments (1)
maybe try to edit your training data a bit before you start to get better results.
it often looks like you just take a few screenshots of your vids and autocaption em and pump em straight through the training.
in this case you could for example inpaint the upper part of the girl to get rid of too much fisheye lense effect (if thats what distorts the faces) or rotate the pics a bit and outpaint the resulting empty areas to have the face in a more upward angle most checkpoints can properly render. training data quality is important ;)
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.