有哪些好的 Stable Diffusion 提示词（Prompt）可以参考？

看完这八种“提示词”种类，你也会Ai高质量出图！

作为一个老法师，Prompt需要我们自己搜集，然后总结。很多人习惯于通过教程和模型来学习绘画，沿着别人预设的轨道前行，这样一开始没有错，但是长久缺乏自主思考和创造的能力，机械式的学习方式只会限制个人的表达，接下来分享几个写实类和动漫类的咒语，以及一些Prompt的使用技巧，出图都是自己去一遍一遍实验的，效果亲测好用！！

一、写实类

1.冷艳御姐

正面tag:

photographic,Cinematic Lighting,high-end fashion photoshoot,wide shot,at night,moon,outdoors,Tall building,street,street lamp,stairs,
Chinese,strapless top,navel cutout,cj_hs,solo,lower body,legs,realistic,fine fabric emphasis,stiletto heels,student,acute face,medium breasts,wearing a miniskirt,skirt,black thigh-highs,high heels,fair skin,black hair,light blush,solo,sitting,from below,(looking at viewer:1.4),povs,
负面tag:

bad_pictures,negative_hand-neg.FastNegativeV2,(bad-artist:1),(loli:1.2),(worst quality, low quality:1.4),(bad_prompt_version2:0.8),bad-hands-5,lowres,bad anatomy,bad hands,((text)),(watermark),error,missing fingers,extra digit,fewer digits,cropped,worst quality,low quality,normal quality,((username)),blurry,(extra limbs),bad-artist-anime,badhandv4,EasyNegative,ng_deepnegative_v1_75t,verybadimagenegative_v1.3,BadDream,(three hands:1.1),(three legs:1.1),(more than two hands:1.4),(more than two legs,:1.2),nsfw,
参数设置：

Steps: 25, Sampler: DPM++ 2M Karras, CFG scale: 7.0, Image CFG scale: 1.5, Seed: 4085590446, Size: 1536x1920, Model hash: d8fd60692a, Model: LEOSAM HelloWorld 新世界 | SDXL真实感大模型_v5.0.safetensors, Denoising strength: 0.4, Clip skip: 2, RNG: CPU, Mask blur: 4, vae_name: automatic

2.性感女骑

正面tag:

1girl,brown hair,outdoor,boots,Bomber Jacket,Wide-Leg Cropped Pants,Seaside View,femalerider, (masterpiece:1.2), best quality, highres,extremely detailed CG,perfect lighting,8k wallpaper, photograph, photorealistic,(sexy:1.4),rider_silk,
负面tag:

NSFW, (worst quality:2), (low quality:2), (normal quality:2), lowres, normal quality, ((monochrome)),((grayscale)), skin spots, acnes, skin blemishes, age spot, (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), (missing arms:1.331), (extra legs:1.331), (fused fingers:1.5), (too many fingers:1.5), (unclear eyes:1.331), lowers, bad hands, missing fingers, extra digit,bad hands, missing fingers, (((extra arms and legs))),nsfw
参数设置：

Steps: 25, Sampler: Euler, CFG scale: 7.0, Image CFG scale: 1.5, Seed: 1967564703, Size: 512x768, Model hash: 7c819b6d13, Model: majicMIX realistic _v7.safetensors, Denoising strength: 0.4, Clip skip: 2, RNG: CPU, Mask blur: 4, Lora 1: 性感丨女骑2.0_性感, Lora Hash 1: 0e673bd5ce, Lora Weight 1: 0.7, vae_name: automatic

3.沙滩丽影

正面tag:

(masterpiece:1.2), best quality, highres,extremely detailed CG,perfect lighting,8k wallpaper, photograph, photorealistic,(full body:1.4),long black hair, cold, serious, white skin, long legs, white bikini, medium breasts,transparent silk scarf, bare feet, sandy beach, excellent weather,film grain,
负面tag:

NSFW, (worst quality:2), (low quality:2), (normal quality:2), lowres, normal quality, ((monochrome)),((grayscale)), skin spots, acnes, skin blemishes, age spot, (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), (missing arms:1.331), (extra legs:1.331), (fused fingers:1.5), (too many fingers:1.5), (unclear eyes:1.331), lowers, bad hands, missing fingers, extra digit,bad hands, missing fingers, (((extra arms and legs))),
参数设置：

DPM++ 2S a Karras, CFG scale: 7.0, Seed: 3265480054, Size: 768x1024, Model hash: 7c819b6d13, Model: majicMIX realistic 麦橘写实_v7.safetensors, Denoising strength: 0.15, Clip skip: 2, ENSD: 31337, RNG: CPU, vae_name: automatic, Hires resize: 1536x2048, Hires steps: 15, Hires upscaler: R-ESRGAN_4x+

4.蕾丝女友

正面tag:

1girl,moyou,upper body,full body,black pantyhose,full body black silk,indoor,tight fitting,nsfw,(lie in bed:1.1),underwear,
负面tag:

EasyNegativeV2,(badhandv4:1.2),nsfw,(medium breasts:1.3),ng_deepnegative_v1_75t,(greyscale:1.2), paintings, sketches, (worst quality:2), (low quality:2), (normal quality:2), lowres, normal quality, ((monochrome)), ((grayscale)), skin spots, acnes, skin blemishes, age spot, glans, badhandsv5-neg, extra legs, broken leg,
参数设置：

Steps: 50, Sampler: DPM++ 2M Karras, CFG scale: 7.0, Seed: 4230811847, Size: 512x768, Model hash: a30fb52420, Model: 墨幽人造人_v1060修复.safetensors, Denoising strength: 0.2, Clip skip: 2, RNG: CPU, Lora 1: 图图的嗨丝（极光5D黑色连裤袜）, Lora Hash 1: 7a47dd78b8, Lora Weight 1: 1.2, vae_name: automatic, Hires resize: 1024x1536, Hires steps: 30, Hires upscaler: ESRGAN_4x, TI hashes: "EasyNegativeV2: 339cc9210f, ng_deepnegative_v1_75t: 54E7E4826D"
以上图片用了不同模型，不同LORA，但是授人以鱼不如授人以渔，看起来prompt很复杂，但风格和把控的提示词是固定的。

总结起来就是：提示词公式=前缀+主体+背景

前缀赋予了提示词特定的权重，在选择前缀时，词汇的精准度和情感色彩是指导模型生成图像的画质。比如“杰作”（masterpiece）、“最佳品质”（best quality）、“官方艺术”（official art）以及“极致细节的CG unity 8k壁纸”（extremely detailed CG unity 8k wallpaper），都能够让Ai明白对作品高度重视，更加强调了图像的质量和细节。为什么要输入这些？因为SD训练时有些图是高清的，有些是模糊的，我们要告诉Ai去倾斜，明白我们想要高品质图。
以下是参考：

best quality, ultra-detailed, masterpiece,hires, 8k
extremely detailed CG unity 8k wallpaper ,unreal engine rendered

其次，主体部分是提示词的核心，它决定了图像所要描述的对象和重点。涉及到外貌、特征、姿态等方面的描述。运用形容词和具体的描述，如a beautiful young girl（一个美丽的年轻女孩）、 stiletto heels（细高跟鞋）,fair skin（白皙的皮肤）,black hair（黑色的头发）可以使Ai为你画出更加生动、迷人的图像。
以下是参考：

服饰穿搭 revealing dress, sweater jacket, pink lucency full dress, 
发型发色 short hair, blonde hair, brown hair, antenna hair, bob cut, spiked hair, 
五官特点 light blush, makeup , cute face, longeyelashes, looking at viewer, 
面部表情grin, laughing, excited, embarrassed , blush, 
肢体动作masterpiece, best quality, leaning forward,  against wall,

最后，背景则是给人物以环境和情景的衬托，增添了画面的层次和深度。通过词汇如at night（夜间）,moon（月亮）,outdoors（户外）,tall building（高层建筑）等，让人物置身于特定的场景中，从而使得整个画面更具真实感。
以下是参考：

白天黑夜 day / night ，in spring, in summer, in autumn, in winter, 
特定时段morning, sunset, moon, cloudy, rain, 
光环境sunlight,bright, dark，strong rim light, intense shadows, fir

根据公式可以得到如下表格：

二、动漫类

1.OL秘书

正面tag:

(best quality), ((masterpiece)), (highres), illustration, original, extremely detailed,1girl, solo, skirt, pantyhose, breasts, black footwear, long hair, shirt, pencil skirt, arms up, black skirt, white shirt, large breasts, white background, full body, looking at viewer, simple background, long sleeves, black pantyhose, drill hair, taut clothes, bangs, miniskirt, high heels, collared shirt, knees together feet apart, yellow eyes, very long hair, standing, shoes, ribbon, ahoge, brown eyes, brown hair, neck ribbon, office lady, closed mouth, bow, parted lips, brown pantyhose, dress shirt, light brown hair, leaning forward, taut shirt, legs, green bow, no background

负面tag:

sketches, (worst quality:2), (low quality:2), (normal quality:2), lowers, normal quality, ((monochrome)), ((grayscale)), skin spots, acnes, skin blemishes, bad anatomy,DeepNegative,(fat:1.2),facing away, looking away,tilted head, 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,worst quality,low quality,normal quality,jpeg artifacts,signature,watermark,extra fingers,fewer digits,extra limbs,extra arms,extra legs,malformed limbs,fused fingers,too many fingers,long neck,cross-eyed,mutated hands,bad body,bad proportions,gross proportions,text,error,missing fingers,missing arms,missing legs,extra digit, extra arms, extra leg, extra foot

参数设置：

Steps: 26, Sampler: Euler a, CFG scale: 5, Seed: 2902768275, Size: 536x960, Model hash: ccc5a563f0, Model: 日式5, VAE hash: 63aeecb90f, VAE: animevae.pt, Clip skip: 2, ENSD: 31337, ADetailer model: face_yolov8n.pt, ADetailer confidence: 0.3, ADetailer dilate erode: 4, ADetailer mask blur: 4, ADetailer denoising strength: 0.4, ADetailer inpaint only masked: True, ADetailer inpaint padding: 32, ADetailer version: 24.1.1, Version: v1.7.0

2.飞天仙女

正面tag:

single girl with single face, In an enchanting setting, an oil painting (1.5) captures a dynamic scene, where a confident girl (1girl) leans against a dreamy background infused with elements of Chinese aesthetics. The backdrop, adorned with mesmerizing Ambilight colors, exudes a touch of China's charm. The dynamic brushwork (1.5) brings her form to life, while the double-exposure background weaves an ethereal atmosphere. Vivid splashes and lively brushstrokes add an artistic flair.
Limited colors burst forth on an ultra-detailed 8k unity wallpaper, seamlessly incorporating elements of Chinese culture, complemented by zentangle and mandala motifs that blend harmoniously with fractal art (1.3). The dynamic angle presents an elegant dance of chaos against a dreamlike atmospheric brutalist design background.
A symphony of renowned artists like James Jean, Roby Dwi Antono, and Ross Tran contributes to an ecstatic musical notes ambiance. Floating colorful sparkles (1) illuminate her glowing skin against the cosmic backdrop, creating a tranquil mosaic reminiscent of blue and green jellyfish (1.4) amidst the photorealistic (1.4) depths.
<lora:anxiang:0.31><lora:GPTS6 Hella-sthetic_424353:0.5> <lora:add_detail:1>

负面tag:

(easynegative:1.2), (badhandv4:1.2), (worst quality:2), (low quality:2), (normal quality:2), lowres, bad anatomy, bad hands, normal quality, ((monochrome)), ((grayscale)) watermark, paintings, sketches, nsfw, ng_deepnegative_v1_75t,multiple faces

参数设置：

teps: 24, Sampler: DPM++ 2M Karras, CFG scale: 3, Seed: 2391792034, Size: 512x768, Model hash: 2de691b494, Model: majicmixReverie_v10, Clip skip: 4, Token merging ratio: 0.2, Lora hashes: "anxiang: 3e95e5f6afaa, GPTS6 Hella-sthetic_424353: 360de1a36997, add_detail: 7c6bad76eb54", TI hashes: "EasyNegative: c74b4e810b03, badhandv4: 5e40d722fc3d, ng_deepnegative_v1_75t: 54e7e4826d53", Version: v1.5.1

3.古风水墨

正面tag:

masterpiece,best quality,ultra high res,raw photo,(photorealistic:1.4),<lora:MengX girl_Mix_V40:0.9>

负面tag：

ng_deepnegative_v1_75t,badhandv4,(worst quality:2),(low quality:2),(normal quality:2),lowres,watermark

参数设置：

Steps: 28, Sampler: DPM++ 2M Karras, CFG scale: 3.5, Image CFG scale: 1.5, Seed: 4214574900, Size: 1024x1664, Model hash: b363551a70, Model: LEOSAM FilmGirl Ultra 写实大模型_FilmGirl Ultra.safetensors, Denoising strength: 0.4, Clip skip: 2, RNG: CPU, Mask blur: 4, Lora 1: 工笔风格 | 流年, Lora Hash 1: daf87afe1e, Lora Weight 1: 0.8, Lora 2: MengX girl_Mix, Lora Hash 2: e413aeda21, Lora Weight 2: 0.8, vae_name: vae-ft-mse-840000-ema-pruned.safetensors, TI hashes: "ng_deepnegative_v1_75t: 54E7E4826D, badhandv4: 5E40D722FC"

4.赛博朋克少女

正面tag:

(masterpiece, best quality, high quality, highres, ultra-detailed), cyber sttyle girl, 1girl, solo, short hair, blurry background, <lora:cyber_style_girl_v1:0.8>,from behind, from below,full body,looking at viewer, back,bare shoulders, <lora:more_details:0.5>

负面tag：

EasyNegative

参数设置：

Steps: 20, Sampler: DPM++ SDE Karras, CFG scale: 7, Seed: 17890705, Size: 512x768, Model hash: bbe83e01df, Model: AWPainting_v1.0, Denoising strength: 0.4, Clip skip: 2, Hires upscale: 2, Hires upscaler: R-ESRGAN 4x+ Anime6B, Lora hashes: "cyber_style_girl_v1: f6fba9a59fd6, more_details: 3b8aa1d351ef", Version: v1.3.2

总结

篇幅有限，这里就不一一展示了，博主这里整理了stablediffusion教学的视频资料和全部的PDF阅读资料，有需要的朋友可以点击下方的卡片进行领取！

有需要的朋友可以点击下方的卡片进行领取！