https://files.catbox.moe/n8q94x.png https://files.catbox.moe/kivzcr.png https://files.catbox.moe/4bkxu3.png https://files.catbox.moe/k0350h.png https://files.catbox.moe/p2lhp6.png https://files.catbox.moe/a3co6p.png https://files.catbox.moe/mh3dd0.png https://files.catbox.moe/zbs2mn.png
prompts varied, but similar to: “prompt”: “(four beautiful 18 year old sisters), (pussy), realistic vagina, American, (long neon rainbow hair), cute schoolgirl, (topless), (big boobs), (big tits:1.3), (nude:1.2), very skinny, sweaty, (shaved pussy), shy smile, pale, holding hair, straight hair, laughing, breasts, colorful background, rainbow knee-length socks, barefoot, side by side, no shoes, dreamlike, open eyes, spread legs, sitting, implied sex, indoor, big bright eyes”, “negative_prompt”: “(deformed, distorted, disfigured:1.3), poorly drawn, bad anatomy, wrong anatomy, extra limb, missing limb, floating limbs, (mutated hands and fingers:1.4), disconnected limbs, mutation, mutated, ugly, disgusting, blurry, amputation, asian, black, droopy boobs, poorly Rendered face, poorly drawn face, poor”,
I was always told positive and negative weightings must each add to 1, so for example …
(big tits:0.6), (nude:0.4)… and then the tokens without specified weights get weighted on a gentle curve by order. I’m trying to find my old reference and coming up empty, so thought I’d just ask: are you confident you’re doing it right, and it isn’t just discarding your over-100% weightings? Do you have a link for how your way works and how to choose the numbers?I’m almost certainly doing it wrong; I only started playing with these tools yesterday. I generate an image, then if it doesnt have the desired result then I reorder words or adjust the weightings or add a negative example to try to guide the result towards what I want
Derp. They use different weighting and tell you how to do it right there in the gui.