This model is a blend of 90s anime, gothic aesthetic and cyberpunk mixed into one. I used some screengrabs from some anime classics mostly from director Yoshiaki Kawajiri's works and some images I made myself and then I ran all the images through photoshop and fixed them up a bit and added my own filters (color grading/saturation/grain/etc...). Let me know if you encounter any problems. As always have fun generating!
If you want to support me financially or request a custom LoRa you can do so here: KoFi
Usage:
IllustriousXL
Trigger word: g0th1cpnk
Strength: I mostly use 0.8-1 but feel free to experiment
FLUX
Trigger word: g0th1cpnk
Strength: I mostly use 0.8-1 but feel free to experiment
PonyXL
Trigger word: g0th1cpnk
Strength: I mostly use 0.8-1 but feel free to experiment
Description
FAQ
Comments (5)
The general prompt I use works on many different things the hundreds of times I've used it, so it's easy to tell when there's outliers, good or bad. This one gives me extreme neck closeups lol. Adding "closeup" the the negative, adding "portrait" to the positive and reducing weight to 0.85 did not help. Unless there's a secret phrase to get it to not neck focus (I see you had "long neck" in the negs) then I'm not sure what's going on.
Edit "long neck" in the negative didn't help. It still kept the neck long though 1 image out of 21 did almost turn out usable. Lol. I may revisit this later and try again.
@Kaladae I see yes I looked into it seems the problem is that some words trigger the long neck and looking up, further checking I found the problem is in one of my models I merged into this one. Will fix it in V2 thanks for pointing it out.
@NeoWave Mo problem, glad I could be of assistance.
@Kaladae If It's not any hassle could you tell me what your prompt was? It'll help me find the issue.
@NeoWave My "standard" positive prompt is
"digital_media_(artwork) hi_res, score_9, score_8_up, score_7_up, score_6_up, score_5_up, score_4_up, rating_safe, beautiful, intricate, high quality details, highly detailed, amazing quality, anthro, furry, female, detailed textured fur, fur tufts, fluffy, slim, slender, cute, small, detailed sexy eyes, solo, SFW, leopard, long sweater, baggy denim jeans, long hair, multicolored hair, streaked hair, absurdres, highres, goth, heavy eyeliner, heavy eye shadow,"
neg
"score_1, score_2, score_3, male, 3d, amputee, bald,(masculine, close-up, muscular), exposed midriff, bare midriff, navel, crop top, shorts, shirt lift, stomach, human, patreon logo, logo, signature, watermark, user name, artist name, text, traditional media, sketch, painting, breast focus, cleavage, side boob, under boob, simple background, unbuttoned pants, (fat, chubby, obese, plump, overweight, wide hips, thick thighs), NSFW, nipple outline, nipples, unguligrade, red eyes, black eyes, black sclera, swirly pupils,"
That's 98% of everything I run through when I gen images just to test style LoRas. If I notice anomolies, I will change samplers or some texts to do a short bit of troubleshooting, but after 21 gens with this one I gave up xD