stable-diffusion 模型效果+prompt

摘自个人印象笔记，图不完整可查看原笔记：https://app.yinxiang.com/fx/55cda0c6-2af5-4d66-bd86-85da79c5574e

Prompt运用规则及技巧：

1. https://publicprompts.art/（最适用于OpenArt 线上模型 https://openart.ai/）

2. https://docs.qq.com/doc/DWFdSTHJtQWRzYk9k（转自元素法典，含各类风格用词）

3. https://www.ptsearch.info/home/（实例查询）

4. http://prompttool.com/NovelAI（体验还不错）

使用：

1. 通用Tag起手式：正向：

masterpiece, best quality, 更多画质词，画面描述

反向：

nsfw, lowres, bad anatomy, bad hands, text, error, missing fingers,extra digit, fewer digits, cropped, worst quality, low quality, normal quality, jpeg artifacts, signature, watermark, username, blurry,根据画面产出加不想出现的画面。nsfw（not safe for work＝涩图）

2. 越靠前的Tag权重越大；比如景色Tag在前，人物就会小，相反的人物会变大或半身。

3. 生成图片的大小会影响Prompt的效果，图片越大需要的Prompt越多，不然Prompt会相互污染。

4. 在Stable Diffusion 中使用()英文括号可增加括号中Tag在画面中的权重 x1.1，[]可减小Tag权重x0.91。在NovelAi官网中，使用{}增加权重x1.05。

5. Prompt支持使用emoji，且表现力较好，可通过添加emoji达到表现效果。如?形容表情，?可修手。

6.“+” ， “ AND” ， “|” 用法：

“+”和“ AND ”都是用于连接短Tag，但AND两端要加空格。"+"约等于" and "

“|” 为循环绘制符号（融合符号）

(Prompt A: w1)|(Prompt B: w2)

以上表达适用于WebUI，w1、w2为权重。AI会对A、 B两Prompt进行循环绘制。可往后无限加入Prompt。

Prompt格式优化：

1. 简易换行三段式表达：

第一段：画质tag，画风tag

第二段：画面主体，主体强调，主体细节概括。（主体可以是人、事、物、景）画面核心内容

第三段：画面场景细节，或人物细节，embedding tag。画面细节内容

第二段一般提供人数，人物主要特征，主要动作（一般置于人物之前），物体主要特征，主景或景色框架等

2.元素同典调整版语法：

质量词→前置画风→前置镜头效果→前置光照效果→（带描述的人或物AND人或物的次要描述AND镜头效果和光照）*系数→全局光照效果→全局镜头效果→画风滤镜（embedding）

Tips: 不同模型（ckpt）对Prompt的敏感程度不同，一套完善的prompt在不同的模型中，表达效果会有差异。每个模型都有自己的特色，需根据模型特色，慢慢调试prompt组合。

模型名称

效果图

prompt

备注

unstableinkdream_v6.safetensors

(Gorgeous anime_screencap,limited palette, :1.3) of (Screen print:1.3),(Cosy:1.3) photorealistic, Analog Style photo, (architecture, bird, blue_sky, bridge, (single Medieval Tavern|castle on hill:1.3), cloud, cloudy_sky,, no humans, :1.2)by studio ghibli, oil painting \(medium\), acrycle painting, by (nvinkpunk:0.3),(kuvshinov:0.8),(dreamlikeart:0.3),(samdoesart:0.3),(modelshoot style:0.3),, unreal engin 5,octane render,(2D:1.2), digital painting, artstation, concept art, smooth, sharp focus,art by ross tran and greg rutkowski and alphonse Mucha and Bak Karol,,(colorful:1.16), soft lightingstudio ghibli,Shinkai Makoto,(__Hyperrealism:1.3),(__Photorealism:1.3),(80s Art:1.3),(Low Contrast:1.3)<hypernet:2nvinkpunk:1.0>

Negative prompt: (nsfw:1.3),(tatoo, things on face :1.2),(watermark:1.2),(bored photo:1.2),no color, blurry, ugly, poor quality, deformed hands, deformed face, deformed body, extra limbs, low quality, normal quality, text, errors, bad anatomy, mutation, deformed fingers, missing fingers, low res, bad hands, cropped, deformed hands, (deformed legs:1.2), (deformed arms:1.2), (multiple arms:1.2), (signature:1.2),bad_bad02, long neck, out of frame, extra fingers, mutated hands, monochrome,Low quality,Photo,Artifacts,Table,Paper,Pencils,Pages,Wall,watermark,signature

ENSD: 31337, Size: 512x1024, Seed: 2545151719, Steps: 15, Sampler: DPM++ SDE Karras, CFG scale: 7, Model hash: d855c1d2ba

疑似需要在models\hypernetworks目录下放置2nvinkpunk模型

(Gorgeous anime_screencap,limited palette, :1.3) of (Screen print:1.3),(Cosy:1.3) photorealistic,,best quality,masterpiece,highly detailed,ultra-detailed, (a castle no no humans, :1.5)by studio ghibli, oil painting \(medium\), acrycle painting, by (nvinkpunk:0.3),(kuvshinov:0.8),(dreamlikeart:0.3),(samdoesart:0.3),(modelshoot style:0.3),, unreal engin 5,octane render,(2D:1.2), digital painting, artstation, concept art, smooth, sharp focus,art by ross tran and greg rutkowski and alphonse Mucha and Bak Karol,,(colorful:1.16), soft lightingstudio ghibli,Shinkai Makoto,(__Hyperrealism:1.3),(__Photorealism:1.3),(80s Art:1.3),(Provia:1.3),naturalism,land Art,regionalism,shutterstock contest winner,trending on unsplash,featured on Flickr

ENSD: 31337, Size: 1088x512, Seed: 4113261453, Steps: 15, Sampler: DPM++ 2S a Karras, Hypernet: 2nvinkpunk, CFG scale: 7, Model hash: d855c1d2ba, Hypernet hash: 1cca8b1232

(anime_screencap,limited palette, :1.3) of by (nvinkpunk:0.3),(kuvshinov:0.8),(dreamlikeart:0.3),(samdoesart:0.3),(modelshoot style:0.3), photorealistic,,best quality,masterpiece,highly detailed,ultra-detailed, east asian architecture, banners, rolling green hills, path, stairs, forest, sunny day, clouds, sharp focus, by studio ghibli,, unreal engin 5,octane render,(2D:1.2), limited palette flat color,, digital painting, artstation, concept art, smooth, sharp focus,art by ross tran and greg rutkowski and alphonse Mucha and Bak Karol,, (oil painting:1.16),(acrylic paint:1.16),, by (photorealistic:1.0),(analog style:1.0),studio ghibli,Shinkai Makoto,(__chinese ink painting:1.3),(__Ukiyo-E:1.3)

Negative prompt: (nsfw:1.3),(tatoo, things on face :1.2),(watermark:1.2),(bored photo:1.2),no color, blurry, ugly, poor quality, deformed hands, deformed face, deformed body, extra limbs, low quality, normal quality, text, errors, bad anatomy, mutation, deformed fingers, missing fingers, low res, bad hands, cropped, deformed hands, (deformed legs:1.2), (deformed arms:1.2), (multiple arms:1.2), (signature:1.2),bad_bad04,Low quality,Photo,Artifacts,Table,Paper,Pencils,Pages,Wall,watermark,signature

ENSD: 31337, Size: 768x512, Seed: 3685826003, Steps: 20, Sampler: DPM++ 2M Karras, Hypernet: 2nvinkpunk, CFG scale: 7, Model hash: d855c1d2ba, Hypernet hash: 1cca8b1232

1: A steampunk town, highly detailed oil painting by Thomas Kinkade and James Gurney. Trending on artstation 8k HD octane render photorealistic concept Art matte digital illustration grainy cgsociety contest winner 3d behance h 1024 w- 768 W&D unreal engine 5 quality cinematic dramatic lighting chiaroscuro volumetric light foggy lens flare red color scheme dark green black purple colors vibrant tones atmospheric photograph

Negative prompt: bad_bad04,Hands-neg,

ENSD: 31337, Size: 512x640, Seed: 3833685131, Steps: 20, Sampler: DPM++ 2S a Karras, Hypernet: 2nvinkpunk, CFG scale: 7, Model hash: d855c1d2ba, Hypernet hash: 1cca8b1232

prmj_v1.safetensors

a cute lion cub in a comfy wooden basket, photoreal, shot using sony a7riii camera, photography

ENSD: 31337, Size: 960x768, Seed: 546740366, Steps: 16, Sampler: DPM++ SDE, CFG scale: 6, Model hash: e40298941c

rabbitRabbit_v10.safetensors

blue_flower, bunny, bush, daisy, dandelion, day, field, fireflies, flower, flower_field, full_body, garden, grass, green_background, holding_flower, hydrangea, leaf, nature, no_humans, on_grass, outdoors, palm_tree, plant, pokemon_\(creature\), potted_plant, purple_flower, smile, spider_lily, tree, tree_stump, tulip, watering_can, white_flower

Negative prompt: hairless,Blurred background

Seed: 731834947, Steps: 7, Sampler: DPM++ SDE Karras, CFG scale: 4

搭配chilloutmix_NiPrunedFp32Fix或者sunshinemix_sunlightmixPruned来使用

Negative prompt: hairless,Blurred background

Seed: 731834940, Steps: 30, Sampler: DPM++ SDE Karras, CFG scale: 4

animal, rabbit,realistic,hairy, animal_focus, blush, looking_at_viewer, no_humans, open_mouth, realistic, smile, tongue, tongue_out, whiskers

Negative prompt: hairless

Seed: 387474633, Steps: 20, Sampler: DPM++ SDE Karras, CFG scale: 5

:3,rabbit,realistic,hairy,clothed_animal, black_eyes, bunny, cherry_blossoms, flower, ground_vehicle, no_humans, outdoors, petals, rabbit_ears, snow, snowflakes, snowing, solo, watermark, winter

Negative prompt: hairless

Seed: 50802857, Steps: 20, Sampler: DPM++ SDE Karras, CFG scale: 7

:3,rabbit,realistic,hairy,clothed_animal, apple, black_eyes, blush, cherry, food, fruit, full_body, hat, no_humans, strawberry, tomato, watermelon

Negative prompt: hairless

Seed: 4106245977, Steps: 20, Sampler: DPM++ SDE Karras, CFG scale: 4

Moxin_10.safetensors

这是一个和墨心搭配使用的构图Lora，一旦使用并再最前前置提示词后，就会采用较大面积留白的构图风格。可以在版本处找到他

注意事项：

1.）CFG范围将会改变风格

1~3 : 大小写意

3~7 : 逐渐工笔

2.）推荐基础模型为国风3.2（国风3.2 ）

3.）《墨心》的推荐Lora权重为0.85以下

4.）《疏可走马》推荐Lora权重为0.7~1

shuimobysim, girl, woman,bare shoulders, (ecchi0.5), (trees:0.5), (flowers:0.6) ,(birds:0.2), (bamboo0.1), lakes, Hangzhou, wuchangshuo, wuchangshuo in background <lora:shuimobysimV3:0.8>

Negative prompt: mutated hands, (poorly drawn hands:1.331),(fused fingers:1.61051), (too many fingers:1.61051), bad hands, missing fingers, extra digit

Size: 1024x1365, Seed: 2643581225, Model: GuoFeng3.2, Steps: 28, Sampler: DPM++ SDE Karras, CFG scale: 2, Model hash: 0bb33c7041

masterpiece, best quality, official art, 8k wallpaper, highly detailed, illustration, 1 girl, black hair, long hair, detailed eyes, bamboo forrest, bare shoulders, lakes, pure, soft smile <lora:Moxin_10:0.55>

<lora:hipoly3DModelLora_v10:0.0> <lora:fashionGirl_v45:0.4> <lora:Moxin_Shukezouma10:0.5>

Negative prompt: fat, naked, nude, paintings, sketches, (worst quality:2), (low quality:2), (normal quality:2), lowres, ((monochrome)), ((grayscale)), bad anatomy, DeepNegative, lowres, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, worst quality, low quality, normal quality, jpeg artifacts, signature, watermark, username, blurry, bad feet, cropped, poorly drawn hands, poorly drawn face, mutation, deformed, extra fingers, extra limbs, extra arms, extra legs, malformed limbs, fused fingers, too many fingers, long neck, cross-eyed, mutated hands, polar lowres, bad body, bad proportions, gross proportions, text, error, missing fingers, missing arms, missing legs,

Size: 512x768, Seed: 3935949157, Model: chilloutmix_NiPrunedFp32Fix, Steps: 20, Sampler: DPM++ SDE Karras, CFG scale: 2, Model hash: fc2511737a

貌似模型要放到lora目录下

shukezouma, (negative space:1.5),clean background, shuimobysim, hanfu, beard, man, mustaches, playing pipa, blue and green ink, <lora:Moxin_Shukezouma:1> <lora:Moxin_1010:0.7>

Negative prompt: mutated hands, (poorly drawn hands:1.331),(fused fingers:1.61051), (too many fingers:1.61051), bad hands, missing fingers, extra digit

Size: 640x1024, Seed: 856686193, Model: gufeng_v10, Steps: 47, Sampler: DPM++ SDE Karras, CFG scale: 3, Model hash: 49337fdda4

shukezouma, (negative space:1.5),clean background, shuimobysim, hanfu, beard, man, mustaches, playing pipa, blue and green ink, <lora:Moxin_Shukezouma:1> <lora:Moxin_1010:0.7>

Negative prompt: mutated hands, (poorly drawn hands:1.331),(fused fingers:1.61051), (too many fingers:1.61051), bad hands, missing fingers, extra digit

Size: 640x1024, Seed: 1963297257, Model: gufeng_v10, Steps: 47, Sampler: DPM++ SDE Karras, CFG scale: 3, Model hash: 49337fdda4

shuimobysim, swordman,holding a sword(trees:0.5),(eagle:0.5), (bamboo:0.2) fighting stance, Hangzhou, wuchangshuo, wuchangshuo in background <lora:moxin_1010:0.8>

Negative prompt: mutated hands, (poorly drawn hands:1.331),(fused fingers:1.61051), (too many fingers:1.61051), bad hands, missing fingers, extra digit

Eta: 0.667, ENSD: 31337, Size: 512x768, Seed: 2340572146, Model: dalcefoBanmooteAsian_dalcefoBanmoote, Steps: 40, Sampler: DPM++ SDE Karras, CFG scale: 2, Clip skip: 2, Model hash: 9ffb93a408

ChilloutMixss或叫：chilloutmixss_xss10.safetensors

(RAW photo, best quality,high quality,highres), (realistic, photo-realistic), ultra-detailed,(front lighting), closed up portrait photo,physically-based rendering, 1girl,20 years old,(stand,pose,front full body),(highly detailed skin:1.3),laughing out loudly,head tilt, very very long longest hoodie,detailed hoodie,bare legs,hands in hoodie pocket, detailed eyes,air bangs,

<lora:xiaoshasha_xss10:0.70>,

<lora:koreanDollLikeness_v10:0.2>

Negative prompt: (EasyNegative:1.2), paintings, sketches, (worst quality:2), (low quality:2), (normal quality:2), lowres, ((monochrome)), (grayscale:1.2),extra fingers,fewer fingers,(watermark:1.2),(letters:1.2),(nsfw:1.2),(2girls:1.2), see-through clothes, huge breasts,large breasts,pants,pussy,panties,bag,purse,hand,(bad ears,earrings:1.4),

Seed: 1738731153, Steps: 20, Sampler: DPM++ 2M, CFG scale: 8

需要将ChilloutMixss模型复制一份到lora目录下，改名为：xiaoshasha_xss10

(8k, RAW photo, best quality, masterpiece:1.2), (realistic, photo-realistic:1.37), best quality, ultra high res, restaurant, 1girl, tree, pink rose, flower, solo, ((hair blown by air)) , sitting, pink flower, window, white rose, brown eyes,((wink)), short hair, denim, white shirt, blue jeans, red rose, shorts, head rest, table, vase, (ulzzang-6500-v1.1:0.5),(pureerosface_v1:0.5),photographed with a sony A9M2, sigma 85 f1.4 lens, by Laurence Demaison, <lora:chilloutmixss_xss10:0.6> ,

Negative prompt: EasyNegative, paintings, sketches, (worst quality:2), (low quality:2), (normal quality:2), lowres, normal quality, ((monochrome)), ((grayscale)), skin spots, acnes, skin blemishes, age spot, glans, nsfw, (outdoor:1.6), backlight,(ugly:1.331), (duplicate:1.331), (morbid:1.21), (mutilated:1.21), (tranny:1.331), mutated hands, (poorly drawn hands:1.5), blurry, (bad anatomy:1.21), (bad proportions:1.331), extra limbs, (disfigured:1.331), (more than 2 nipples:1.331), (missing arms:1.331), (extra legs:1.331), (fused fingers:1.61051), (too many fingers:1.61051), (unclear eyes:1.331), lowers, bad hands, missing fingers, extra digit, (futa:1.1),bad hands, missing fingers, panties,bad hands,bad finger,long finger,Multiple fingers,hands up,

ENSD: 31337, Size: 640x960, Seed: 172127844, Steps: 20, Sampler: DPM++ 2M Karras, CFG scale: 8, Clip skip: 2, Model hash: a757fe8b3d, ControlNet Model: control_sd15_openpose [fef5e48e], ControlNet Module: openpose, ControlNet Weight: 1, ControlNet Enabled: True, ControlNet Guidance Strength: 1

goodAsianGirlFace_goodAsianGirlFaceV12.safetensors

masterpiece,best quality, upper body protrait of (Wsxff:1.3), brown long wavy hair, bangs, big eyes, parted lips, black eyes,, Sketch, masterpiece, best quality, extremely detailed, 1girl, full body, beautiful detailed eyes, cute anime face, full body, beautiful detailed face, white hair, (Botanical illustration:1.5), white dress

Negative prompt: nsfw,ugly, lowres, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, worst quality, low quality, normal quality, jpeg artifacts, signature, watermark, username, blurry, multiple girls, multiple faces,, canvas frame, cartoon, 3d, ((disfigured)), ((bad art)), ((deformed)),((extra limbs)),((close up)),((b&w)), wierd colors, blurry, (((duplicate))), ((morbid)), ((mutilated)), [out of frame], extra fingers, mutated hands, ((poorly drawn hands)), ((poorly drawn face)), (((mutation))), (((deformed))), ((ugly)), blurry, ((bad anatomy)), (((bad proportions))), ((extra limbs)), cloned face, (((disfigured))), out of frame, ugly, extra limbs, (bad anatomy), gross proportions, (malformed limbs), ((missing arms)), ((missing legs)), (((extra arms))), (((extra legs))), mutated hands, (fused fingers), (too many fingers), (((long neck))), Photoshop, video game, ugly, tiling, poorly drawn hands, poorly drawn feet, poorly drawn face, out of frame, mutation, mutated, extra limbs, extra legs, extra arms, disfigured, deformed, cross-eye, body out of frame, blurry, bad art, bad anatomy, 3d render, nsfw, lower, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, worst quality, low quality, normal quality, jpeg artifacts, signature, watermark, username, blurry, missing fingers, bad hands, missing arms, long neck, Humpbacked, shadow, nude

ENSD: 31337, Size: 512x768, Seed: 104490948, Steps: 28, Sampler: Euler a, CFG scale: 12, Model hash: aff1d22be5, Face restoration: GFPGAN

High Detail, Sharp focus, dramatic, photorealistic, hyper realistic, a long bob cut girl [with a warm smile:0.8] working at a cozy coffee shop, wearing a black apron and a white shirt, at dusk, with depth of field, light rays, (light particles:0.9), and soft shadows on her face, looking at viewer, [Paul Hedley:0.5], warm tone

Negative prompt: ugly, fat, obese,(((deformed))), blurry, bad anatomy, disfigured, poorly drawn face, mutation, mutated, (extra_limb), (ugly), (poorly drawn hands fingers), messy drawing, morbid, mutilated, [out of frame], (bad proportions), (poorly drawn body), (poorly drawn legs), worst quality, low quality, normal quality, text, censored,, ugly, fat, obese,(((deformed))), blurry, bad anatomy, disfigured, poorly drawn face, mutation, mutated, (extra_limb), (ugly), (poorly drawn hands fingers), messy drawing, morbid, mutilated, [out of frame], (bad proportions), (poorly drawn body), (poorly drawn legs), worst quality, low quality, normal quality, text, censored, (bad_prompt_version2:0.8)

ENSD: 31337, Size: 768x1024, Seed: 2029189734, Steps: 30, Sampler: Euler a, CFG scale: 7, Batch pos: 1, Clip skip: 2, Mask blur: 4, Batch size: 4, Model hash: 6aa9d14bba, Denoising strength: 0.75

a photo of Katonana, masterpiece, best quality, 4k, illustration,((1girl)), realistic, photo \(medium\), ((masterpiece)), (((best quality))), ((ultra-detailed)), ((illustration)),depth of field,cinematic lighting, volume lighting, bloom effect,light particles, hair ornament, sunlight , morning, lens 135mm,f1.8,

Negative prompt: nsfw, lowres, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, worst quality, low quality, normal quality, jpeg artifacts, signature, watermark, username, blurry,mulitple people, multiple boys, multiple girls,

ENSD: 31337, Size: 512x768, Seed: 3332775436, Model: GoodAsianGirlFace_V1.0, Steps: 50, Sampler: DPM++ 2M Karras, CFG scale: 10.5, Model hash: 206eb7d3, Face restoration: GFPGAN

Xiaorenshu_v20.safetensors

shz,((best quality)), ((masterpiece)), (detailed), realistic, multiple boys,(riding horse),mustache, looking at viewer, outdoors,house, amy,horse

Negative prompt: easynegative bad_prompt_version2 ng_deepnegative_v1_75t

ENSD: 31337, Size: 512x768, Seed: 3836055402, Steps: 25, Sampler: DPM++ SDE Karras, CFG scale: 5, Clip skip: 2, Model hash: 5efb10a9e1, Hires steps: 10, Hires upscale: 2, AddNet Enabled: True, AddNet Model 1: xrs2.0-000016(97413ad0d80d), Hires upscaler: SwinIR_4x, AddNet Module 1: LoRA, AddNet Weight A 1: 0.9, AddNet Weight B 1: 0.9, Denoising strength: 0.45

((best quality)), ((masterpiece)), (detailed), multiple boys, battle, outdoors, fighting stance,amy,horse

Negative prompt: easynegative bad_prompt_version2 ng_deepnegative_v1_75t

ENSD: 31337, Size: 960x640, Seed: 67430860, Steps: 30, Sampler: DPM++ SDE Karras, CFG scale: 6, Clip skip: 2, Model hash: fc2511737a, Hires steps: 10, Hires upscale: 2, AddNet Enabled: True, AddNet Model 1: xrs2.0-000016(97413ad0d80d), Hires upscaler: SwinIR_4x, AddNet Module 1: LoRA, AddNet Weight A 1: 0.9, AddNet Weight B 1: 0.9, Denoising strength: 0.4

藏丹阁-NovelAI模型大全分享-第一层：

https://docs.qq.com/doc/DY0lFeWZuVXRCdUJU

藏丹阁-NovelAI模型大全分享-第二层：

https://docs.qq.com/doc/DY1hydmRGTENGTXdo