README.md · p1atdev/lora at main

metadata

license: creativeml-openrail-m
datasets: p1atdev/noz

LoRA (LierLa)

Not so useful LoRAs. These maybe only works with kohya's sd-scripts or webui extension.

alley-test1-e20.safetensors: Realistic alley backgrounds LoRA for WDv1.4.
alley-test2-e50.safetensors: Better backgrounds LoRA for WDv1.4.

impasto-test1-last.safetensors: Impasto style for WDv1.4 but not good at person.
fluorite-test5-last.safetensors: Photo portrait for SDv2.1 512.
pastel-flavor-test1-e100.safetensors: LoRA trained with PastelMix's images for WD1.4. (bad nose)
pastel-flavor-test2-e100.safetensors: LoRA trained with PastelMix's images for WD1.4. (a little better than test1)

fumo-test1.safetensors: Fumo style for WDv1.4, better than test2 at details.
fumo-test2.safetensors: Fumo style for WDv1.4, better than test1 at backgrounds and resolution.

nurie-test2-e10.safetensors: Good at black and white lineart style.

noz-test3-2-e40.safetensors: NOZ style watch for SDv2.1-768. Dataset.

e.g.

a blue watch
a red pocket watch

pvc-test7-e1.safetensors: PVC figure, figma and nendoroid style LoRA for WD1.5 beta 1. Experimental.

Trigger word: pvc, figma, nendoroid

e.g.

masterpiece, best quality, high quality, exceptional, best aesthetic, newest, pvc, , 1girl, loli, blue hair, short hair, high ponytail, white shirt, dress shirt, collared shirt, frills, ribbon, bow, looking at viewer, upper body, street, day, beautiful detailed,
Negative prompt: nsfw, worst quality, low quality, medium quality, deleted, lowres, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, jpeg artifacts, signature, watermark, username, blurry, oldest, old, bad aesthetic, realistic

acrylic-test2-e32.safetensors: Acrylic figure style LoRA for WD1.5 beta 2.

real life, anime is recommended in positive prompt.

masterpiece, best quality, exceptional, real life, anime, 1girl, aqua eyes, baseball cap, blonde hair, closed mouth, earrings, hat, hoop earrings, jewelry, looking at viewer, shirt, short hair, solo, upper body, yellow shirt, indoors, room, full body, pose, standing
Negative prompt: badquality, bad anatomy, bad legs, extra legs, oldest
Steps: 20, Sampler: DPM++ 2M Karras, CFG scale: 11, Seed: 357295013, Size: 512x768, Model hash: 4bf1855f4d, Model: wd-1-5-beta2-fp16, Denoising strength: 0.7, Clip skip: 2, Hires upscale: 1.5, Hires upscaler: Latent

marker-test3-e20.safetensors: Marker style LoRA for WD1.5 beta 2. Used dataset: waifu, nijijourney.

masterpiece, best quality, exceptional,, 1girl, bangs, bare shoulders, beret, black hair, black shorts, blue hair, bracelet, breasts, buttons, colored inner hair, double-breasted, eyewear removed, green headwear, green jacket, grey eyes, grey sky, hat, jacket, jewelry, long hair, looking at viewer, multicolored hair, neck ring, o-ring, off shoulder, rain, round eyewear, shorts, sidelocks, small breasts, solo, sunglasses, wavy hair, wet, zipper,
Negative prompt: badquality, bad anatomy, nsfw, nude, bad hands
Steps: 20, Sampler: DPM++ 2M Karras, CFG scale: 7, Seed: 3865755286, Size: 512x768, Model hash: 4bf1855f4d, Model: wd-1-5-beta2-fp16, Clip skip: 2

niji-test1-4-e10.safetensors: nijijourney v5's style for WD 1.5 beta 2~.
niji-test1-5-offset.safetensors: continued from test1-4-e10, and trained with noise offset.

test1-4-e10 is more stable than test1-5-offset.

masterpiece, best quality, exceptional, best aesthetic, niji style, green eyes, gloves, solo, hat, black gloves, hair between eyes, looking at viewer, mask, upper body, jacket, short hair, black jacket, cable, male focus, black headwear, 1boy, shirt, long sleeves, bangs, white hair, steampunk, respirator, suit, grey hair, formal, witch hat, gas mask
Negative prompt: worst quality, low quality, bad aesthetic, oldest, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, jpeg artifacts, signature, watermark, username, blurry, colorfixerv1
Steps: 18, Sampler: DPM++ 2M Karras, CFG scale: 7, Seed: 3971604115, Size: 512x768, Model hash: b7c378c76a, Model: wd-15-2.6-186k, Denoising strength: 0.7, Clip skip: 2, Hires upscale: 1.5, Hires steps: 12, Hires upscaler: Latent

ink-painting-1-35.safetensors: Ink painting style LoRA for WD1.5 beta3.

Use the tag ink painting if it does not work.

masterpiece, best quality, exceptional, best aesthetic, ink painting, 1girl, upper body, looking at viewer, monochrome
Negative prompt: worst quality, low quality, bad aesthetic, oldest, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, jpeg artifacts, signature, watermark, username, blurry
Steps: 20, Sampler: Euler a, CFG scale: 7, Seed: 1805321870, Size: 512x768, Model hash: d38e779546, Model: wd-beta3-base-fp16, Clip skip: 2, Version: v1.2.1, Script: X/Y/Z plot, X Type: AddNet Weight 1, X Values: "0, 1"

masterpiece, best quality, exceptional, best aesthetic, ink painting, 1girl, school uniform, cowboy shot, looking at viewer, blue theme,
Negative prompt: worst quality, low quality, bad aesthetic, oldest, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, jpeg artifacts, signature, watermark, username, blurry
Steps: 20, Sampler: Euler a, CFG scale: 7, Seed: 1622463845, Size: 512x768, Model hash: d38e779546, Model: wd-beta3-base-fp16, Clip skip: 2, Version: v1.2.1, Script: X/Y/Z plot, X Type: AddNet Weight 1, X Values: "0, 1"

LoCon

low_key-test1-e10.safetensors: Low key offset noise LoCon for WD 1.5 beta 2.

maybe good at realistic style rather than anime style in vanilla beta 2 model.

(low key), masterpiece, exceptional, best aesthetic, solo, realistic, photography of a young beautiful girl in the dark, white hair, streaked hair, red hair, black background,
Negative prompt: badquality, bad anatomy, nsfw, nude, iphone, phone, red lips
Steps: 20, Sampler: DPM++ 2M Karras, CFG scale: 11, Seed: 765289347, Size: 512x768, Model hash: 4bf1855f4d, Model: wd-1-5-beta2-fp16, Denoising strength: 0.7, Clip skip: 2, Hires upscale: 1.5, Hires upscaler: Latent

lolita-test2-e20-locon.safetensors: Better realistic LoCon for WD 1.5 beta 2. Trained on lolita photos.

maybe good at dress?

masterpiece, best quality, exceptional, realistic, waifu, 1girl, solo, black hair, hat, dress, long hair, black dress, outdoors, realistic, necktie, black headwear, long sleeves, standing, black footwear, jacket, blurry background, jacket on shoulders, belt, black jacket, boots, facing viewer
Negative prompt: badquality, bad anatomy, oldest, fat, blurry, bad hands, extra digits, fewer digits
Steps: 20, Sampler: DPM++ 2M Karras, CFG scale: 10, Seed: 4146640340, Size: 512x768, Model hash: 4bf1855f4d, Model: wd-1-5-beta2-fp16, Denoising strength: 0.7, Clip skip: 2, Hires upscale: 1.5, Hires upscaler: Latent

fashion-test1-e5.safetensors:　Better realistic LoCon for WD 1.5 beta 2. Trained on lolita and techwear photos.

maybe good at most of the fashions in realistic. More than 1.0 weight is recommended.

masterpiece, exceptional, 1girl, medium hair, blue dress, pose, cowboy shot, looking at viewer
Negative prompt: badquality, bad anatomy, fat, blurry, selfie
Steps: 18, Sampler: DPM++ 2M Karras, CFG scale: 7, Seed: 3389642227, Size: 512x768, Model hash: 4bf1855f4d, Model: wd-1-5-beta2-fp16, Denoising strength: 0.7, Clip skip: 2, Hires upscale: 1.5, Hires steps: 10, Hires upscaler: Latent