Resized to 78% of original (view original)
Prompt | ((POV standing close high angle face)), ((futanari pulling down pants)), solo, stunning ultra high resolution beautifully lit, (black background, photorealistic, realistic lighting:1.35), portrait photo of a beautiful breathtakingly attractive cute sexy erotic stylish horny fit skinny girl with huge soft lips, (((aroused, mesmerized, hot, eye contact, looking at viewer)), big lips, [stylish:((hyper nipples))), huge nipples, NSFW:30], perspective, (((throbbing penis, glans, foreskin, aphrodesiac, pheremones))), skintight white shirt, mesmerized, pleasure, aroused, (standing), embarassed, sly smile, <lora:epiCRealismHelper:1> <lora:hiqcg_body_epoch:0.1@35, 0.3@40> <lora:anteater face:0.3@30, 0.0@35> <lora:puaziStyle_V2:0.1@15, 0.75@25, 0.5@38, 0.25@45> <lora:futanarimix-locon-dim64-alpha32:0.0@30, 0.35@40, 0.1@45> <lora:iqisov:0.6@10, 0.1@25> |
---|---|
Negative prompt | bad-artist-anime, bad-image-v2-39000, bad-artist, bad_prompt_version2, (bleak), EasyNegative, (((fat, Asian, closed eyes))), ((pussy, vagina)), (((drawing, line-art))) |
Sampler | DDIM |
Seed | 246497784 |
Steps | 49 |
Cfg Scale | 7 |
Model Hash | 2d72e3209c |
Anonymous 09/20/23(Wed)20:22:00 No.10765759
catbox_s2r6w3.png (1.5 MB, 1088x2048)
Too many images in the same style, let's spice it up!
I've been having a lot of luck using lora control https://github.com/cheald/sd-webui-loractl
It allows you to specify steps where the lora has different strengths. That means you can get the composition from one lora, then tune it out so the details/style is from another
format is comma separated STRENGTH@STEPNUM
Example:
<lora:hyperballs-v3:0.15@20, 0.99@30, 0.50@45>
starts with low strength so it doesn't change composition, then tries to make balls bigger, then fades near the end so style isn't modified.