Prefect Pony (NSFW) 动漫模型

使用介绍

V3更新：

• 添加了一些风格
• 应该给更好的脸和眼睛
• 更好的 NSFW 支持
• 广泛的 Lora 支持

建议设置：

• 我在每张图像上都设置了 CLIP skip 2
• 我有ENSD：全部 31337
• 它们都有更高分辨率的highres.fix或 img2img。
• 我不使用恢restore faces
• 我使用 aftertailer 来处理脸部细节
• 用于 img2img upscaling 的平铺扩散，有时用于一些更多额外细节的噪声反转。
• 4x-Ultrasharp upscaler

注意：如果您发现预览图像的任何提示看起来很熟悉，那是因为我从其他模型评论图像中获取了它们，感谢原作者。感谢基准测试。

作品参照

score_9,score_8_up,score_7_up, 1girl, aqua eyes, blonde hair, blush, eyelashes, hair ornament, heart, heart-shaped pupils, heart background, heart hair ornament, long hair, looking at viewer, one side up, open mouth, pink background, portrait, red ribbon, ribbon, saliva, shirt, sidelocks, smile, solo, symbol-shaped pupils, teeth, tongue, tongue out, upper teeth only
Negative prompt: realistic,monochrome,greyscale, artist name, signature, watermark,
Steps: 20, CFG scale: 7, Sampler: DPM++ 2M, Seed: 1709835910, : [object Object], Size: 1248x1824, Model: prefect_pony_xl_v3.fp16, Version: v1.10.1, {Method: [object Object], Upscaler: [object Object], Model hash: e31a2563f0, Tile Overlap: 48, Schedule type: Karras, Upscale factor: 1.5, ADetailer model: face_yolov8n.pt, Keep input size: true}, Tile batch size: 6, Tile tile width: 160, Tiled Diffusion: [object Object], Tile tile height: 160, ADetailer version: 24.8.0, Denoising strength: 0.2, ADetailer mask blur: 4, ADetailer confidence: 0.3, ADetailer dilate erode: 4, Tiled Diffusion upscaler: 4x-UltraSharp, ADetailer inpaint padding: 32, ADetailer denoising strength: 0.4, Tiled Diffusion scale factor: 1.5, ADetailer inpaint only masked: True

score_9,score_8_up,score_7_up, 1girl, alternate costume, beanie, black choker, black hair, black hat, black sweater, bow, brown eyes, buttons, choker, collared shirt, covered mouth, cross, cross earrings, earrings, floral background, hair between eyes, hat, hat bow, jewelry, long hair, looking at viewer, mask, mouth mask, pink background, portrait, prototype design, red bow, shirt, sidelocks, solo, spiked ear piercing, spikes, studded choker, sweater, twintails, white shirt
Negative prompt: realistic,monochrome,greyscale, artist name, signature, watermark,
Steps: 20, CFG scale: 7, Sampler: DPM++ 2M, Seed: 1709835912, : [object Object], Size: 1248x1824, Model: prefect_pony_xl_v3.fp16, Version: v1.10.1, {Method: [object Object], Upscaler: [object Object], Model hash: e31a2563f0, Tile Overlap: 48, Schedule type: Karras, Upscale factor: 1.5, ADetailer model: face_yolov8n.pt, Keep input size: true}, Tile batch size: 6, Tile tile width: 160, Tiled Diffusion: [object Object], Tile tile height: 160, ADetailer version: 24.8.0, Denoising strength: 0.2, ADetailer mask blur: 4, ADetailer confidence: 0.3, ADetailer dilate erode: 4, Tiled Diffusion upscaler: 4x-UltraSharp, ADetailer inpaint padding: 32, ADetailer denoising strength: 0.4, Tiled Diffusion scale factor: 1.5, ADetailer inpaint only masked: True

score_9, score_8_up, score_7_up, score_6_up, 1girl, black dress, weiss schnee, (from below:0.85), crossed legs, looking at viewer, source_anime, blue eyes, beautiful, perfect eyes, white legwear, disdain, blush, tight clothes, rating_safe, (black garter:1.3), white ponytail, sitting, throne room
Negative prompt: text, watermark, source_pony, source_furry, rating_explicit
Steps: 4, CFG scale: 1, Sampler: DPM++ SDE, Seed: 1, Model: prefectPonyXL_v3, width: 512, height: 768

使用介绍

V2更新：

• 风格影响力较小
• 可以与多种 loras 配合使用
• 更好的解剖学
• 颜色更好

将 pony 与 animagine 以及我自己的风格 loras 合并。应该可以提供良好的眼睛和良好的图像，无需任何升级。

该模型内置了 VAE。

建议设置：

• 我在每张图像上都设置了 CLIP skip 2
• 我有ENSD：全部 31337
• 它们都有更高分辨率的highres.fix或 img2img。
• 我不使用恢restore faces
• 我使用 aftertailer 来处理脸部细节
• 用于 img2img upscaling 的平铺扩散，有时用于一些更多额外细节的噪声反转。
• 4x-Ultrasharp upscaler

注意：如果您发现预览图像的任何提示看起来很熟悉，那是因为我从其他模型评论图像中获取了它们，感谢原作者。感谢基准测试。

作品参照

score_9,score_8_up,score_7_up,score_6_up,hadou nejire, 1girl, ahoge, blue eyes, blue hair, blue sailor collar, blue skirt, drill hair, ghost pose, hitodama, neckerchief, official alternate costume, official alternate hairstyle, open mouth, pleated skirt, red neckerchief, sailor collar, school uniform, serafuku, simple background, skirt, solo, twin drills, white background
Negative prompt: greyscale, monochrome, motion blur, emphasis lines, text, title, logo, signature,censored, 3d,patreon username, patreon logo, artist name, watermark
Steps: 20, CFG scale: 7, Sampler: DPM++ 2M, Seed: 690217838, : [object Object], Size: 1248x1824, Model: prefect_pony_xl_v2.fp16, Version: v1.9.4, {Method: [object Object], Upscaler: [object Object], Mask blur: 4, Resharpen: 0.3, Model hash: 517caf19d7, Inpaint area: Only masked, Tile Overlap: 48, Schedule type: Karras, Upscale factor: 1.5, ADetailer model: face_yolov8n.pt, Keep input size: true}, Tile batch size: 6, Tile tile width: 160, Tiled Diffusion: [object Object], Tile tile height: 160, ADetailer version: 24.3.0, Denoising strength: 0.4, ADetailer mask blur: 4, Masked area padding: 32, ADetailer confidence: 0.3, ADetailer dilate erode: 4, Tiled Diffusion upscaler: 4x-UltraSharp, ADetailer inpaint padding: 32, ADetailer denoising strength: 0.4, Tiled Diffusion scale factor: 1.5, ADetailer inpaint only masked: True

score_9, score_8_up, score_7_up, score_6_up, 1girl, alternate costume, alternate hairstyle, artist name, blonde hair, blue background, blush, butterfly-shaped pupils, chestnut mouth, cowboy shot, double bun, food, hair bun, hair ribbon, holding, holding ice cream cone, ice cream, ice cream cone, long sleeves, open mouth, pink pupils, pink ribbon, red bag, ribbon, shirt, simple background, solo, symbol-shaped pupils, too many, too many scoops, white shirt
Negative prompt: 3d,realistic,monochrome,greyscale, artist name, signature, watermark,
Steps: 30, CFG scale: 7, Sampler: DPM++ 2M, Seed: 2396149213, : [object Object], Size: 1248x1824, Model: prefect_pony_xl_v2.fp16, Version: v1.10.1, {Method: [object Object], Upscaler: [object Object], Model hash: 517caf19d7, Tile Overlap: 48, Schedule type: Karras, Upscale factor: 1.5, ADetailer model: face_yolov8n.pt, Keep input size: true}, Tile batch size: 6, Tile tile width: 160, Tiled Diffusion: [object Object], Tile tile height: 160, ADetailer version: 24.6.0, Denoising strength: 0.15, ADetailer mask blur: 4, ADetailer confidence: 0.3, ADetailer dilate erode: 4, Tiled Diffusion upscaler: 4x-UltraSharp, ADetailer inpaint padding: 32, ADetailer denoising strength: 0.4, Tiled Diffusion scale factor: 1.5, ADetailer inpaint only masked: True

score_9, score_8_up, score_7_up, score_6_up, <lora:luminous_valentine_pdxl_goofy:1>lvslime, 1girl, solo, heterochromia, blue eyes, long hair, red eyes, necklace, jewelry, looking at viewer, hairband, sitting, open mouth, short sleeves, ribbon, upper body, collarbone, smile, black ribbon, grey hair, shirt, chair, black hairband, gothic , black dress, dress, puffy sleeves
Negative prompt: realistic,monochrome,greyscale, artist name, signature, watermark,
Steps: 30, CFG scale: 7, Sampler: DPM++ 2M, Seed: 1921100514, : [object Object], VAE: sdxl_vae_fixed.safetensors, Size: 1248x1824, Model: prefect_pony_xl_v2.fp16, Version: v1.10.1, {Method: [object Object], Upscaler: [object Object], Model hash: 517caf19d7, Tile Overlap: 48, Schedule type: Karras, Upscale factor: 1.5, ADetailer model: mediapipe_face_full, Keep input size: true}, Tile batch size: 6, Tile tile width: 160, Tiled Diffusion: [object Object], Tile tile height: 160, ADetailer version: 24.6.0, Denoising strength: 0.24, ADetailer mask blur: 4, ADetailer confidence: 0.3, ADetailer dilate erode: 4, Tiled Diffusion upscaler: 4x-UltraSharp, ADetailer inpaint padding: 32, ADetailer denoising strength: 0.4, Tiled Diffusion scale factor: 1.5, ADetailer inpaint only masked: True

使用介绍

将 pony 与 animagine 以及我自己的风格 loras 合并。应该可以提供良好的眼睛和良好的图像，无需任何升级。

该模型内置了 VAE。

建议设置：

• 我在每张图像上都设置了 CLIP 跳过 2
• 我有ENSD：全部 31337
• 它们都有更高分辨率的highres.fix或 img2img。
• 我不使用恢restore faces
• 我使用 aftertailer 来处理脸部细节
• 用于 img2img 升级的平铺扩散，有时用于一些更多额外细节的噪声反转。
• 4x-Ultrasharp 升频器

注意：如果您发现预览图像的任何提示看起来很熟悉，那是因为我从其他模型评论图像中获取了它们，感谢原作者。感谢基准测试。

作品参照

Stable Diffusion提示词

scrore_9, score_8_up, score_7_up, source_anime, finger gun, pointing at viewer, upper body, cute, , pink swimsuit, hairband, low twintails, wolf ears,white hair, blue eyes, wink, smile, open mouth, peace, spoken heart,
Negative prompt: score_4,score_5,score_6 ugly, deformed, bad quality, amateur drawing, beginner drawing, deformed, bad anatomy, deformed, white background, deformed hands, deformed feet, ugly eyes, ugly tits, watermark, missing fingers,
Steps: 28, Size: 832x1216, Seed: 1032535588, Sampler: Euler a, CFG scale: 6, Clip skip: 2

Stable Diffusion提示词

score_9, score_8_up, score_7_up, solo, looking at viewer, short hair, shirt, black hair, 1boy, closed mouth, jacket, red eyes, white shirt, upper body, white hair, male focus, multicolored hair, open clothes, red two-tone hair, black jacket, green eyes, blurry background, heterochromia, split-color hair,wet, portrait, bubble, water drop
Negative prompt: score_4,score_5,score_6 ugly, deformed, bad quality, amateur drawing, beginner drawing, deformed, bad anatomy, deformed, white background, deformed hands, deformed feet, ugly eyes, ugly tits, watermark
Steps: 30, : [object Object], Size: 1248x1824, Seed: 1552434314, Model: prefect pony.fp16, Version: v1.9.3, Sampler: Euler, {Method: [object Object], NoiseInv: true, Upscaler: [object Object], CFG scale: 7, Clip skip: 2, Mask blur: 4, Model hash: 91b87bf726, Inpaint area: Only masked, Tile Overlap: 48, Schedule type: Automatic, NoiseInv Steps: 10, Upscale factor: 1.5, ADetailer model: face_yolov8n.pt, Keep input size: true, Tile batch size: 4, Tile tile width: 96, Tiled Diffusion: [object Object], NoiseInv Retouch: 1, Tile tile height: 96, ADetailer version: 24.3.0, Denoising strength: 0.4, ADetailer mask blur: 4, Masked area padding: 32, ADetailer confidence: 0.3, NoiseInv Kernel size: 64}, ADetailer dilate erode: 4, Tiled Diffusion upscaler: 4x-UltraSharp, ADetailer inpaint padding: 32, NoiseInv Renoise strength: 1, ADetailer denoising strength: 0.4, Tiled Diffusion scale factor: 1.5, ADetailer inpaint only masked: True