随着人工智能技术的不断发展，越来越多的AIGC技术被应用到了人们的日常生活中。其中，stable diffusion(SD)在交互中的应用越来越受到人们的关注。SD可以快速生成大量创意图片，从而提高互联网产品的图片产出效率，提高用户体验和市场竞争力。本文将分享SD在交互中的10个应用案例，希望能够为读者带来一些启发和思考。
例1：横幅图片景深交互
我们以b站首页为例，制作这样的带景深交互的横幅图片。
b站首页效果：
我们实现的效果：
实现步骤：
- 通过sd的controlnet depth功能生成深度图
- 人为介入修改，调整边缘模糊、局部细节修正（人物必须留出足够的安全区）
- 通过fake-3d-image的示例，导入原图和深度图
例2：定制门户网站首页大图
效果：
步骤：
- 通过人为调整原图在画面中所占位置
- 配置controlnet 线稿像素级别控制效果
- 通过科幻城市+汽车等lora模型，生产图像
- 通过现有的ai放大工具进一步放大生成超清图
生成信息：
a white car , ,dark orange background,black background,dark orange theme,pitch black,total darkness,night scene,Front viewing angle,Neon orange Light,Cyberpunk,Science Fantasy,
Negative prompt: light,bright,(worse quality:1.4),(low quality:1.5), sunshine,sunlight,white,blue
Steps: 100, Sampler: DPM++ 3M SDE Karras, CFG scale: 6, Seed: 2409581708, Size: 1920x264, Model hash: 3e06af9e61, Model: 25DWorld_25dv5, Denoising strength: 1, ControlNet 0: "Module: lineart_standard (from white bg & black line), Model: control_v11p_sd15_lineart [43d4be0d], Weight: 1.5, Resize Mode: Crop and Resize, Low Vram: False, Processor Res: 512, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: ControlNet is more important", Lora hashes: "qicheV2: 8ef2fb8f1463, chuansongzhan: 4c962f193aa4, _20230804165634: 28c437ce1c4c", Version: v1.6.0 例3：移动端竖屏互联网插画风格图
一种是通过 现有的互联网插画风格lora模型 用提示词生成；
一种是有线稿+sd人为限定风格生成；
效果：
生成信息：
an artwork, Van, in the style of 2d game art, characterized girl, dmitry vishnevsky, orange and emerald, cute and dreamy, editorial illustrations, zeiss batis 18mm f/2.8
Negative prompt: lowres, bad anatomy, ((bad hands)), (worst quality:2), (low quality:2), (normal quality:2), paintings, sketches, lowres, bad anatomy, bad hands, text, error, missing fingers,
Steps: 30, Size: 832x1536, Seed: 137767811, Model: IDillustration_SDXL_Ckpt_V1.0 , Sampler: DPM++ 3M SDE Karras, CFG scale: 7例4：游戏（背景或人物）原型/素材/草稿
背景素材效果：
生成信息：
pixel art,(new_york),((sakura)),(morden_tall_building),(((streets)))(((tram))),(((shops))),((bridge)),(cars), (((mountain))),forest,christmas,((sky_cloud)),summer,shining,blue lake,nature light and shadows,Fantastic light,scenery,landscape,(aerial_view)
Negative prompt: (simple background),(((temple))),monochrome ,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, lowres, bad anatomy, bad hands, text, error, extra digit, fewer digits, cropped, worst quality, low quality, normal quality, jpeg artifacts, signature, watermark, username, blurry, ugly,pregnant,vore,duplicate,morbid,mut ilated,tran nsexual, hermaphrodite,long neck,mutated hands,poorly drawn hands,poorly drawn face,mutation,deformed,blurry,bad anatomy,bad proportions,malformed limbs,extra limbs,cloned face,disfigured,gross proportions, (((missing arms))),(((missing legs))), (((extra arms))),(((extra legs))),pubic hair, plump,bad legs,error legs,username,blurry,bad feet,dark shadows
Steps: 100, Sampler: DPM++ 2M Karras, CFG scale: 5, Seed: 2715271919, Size: 2048x1024, Model hash: 0eca235400, Model: SDXLPixelArtBeta_v10, Clip skip: 2, Version: v1.6.0人物素材效果：
生成信息：
(masterpiece),pixcel art,(8k),hdr,character sheet,woman warrior,pixcel art style,((multiple views)),((highly detailed)), digital photography,run,walk,jump,swim,get down,stand,4 piece,4 diffrent angle,simple background,clean white background,white clothes,red hair
Negative prompt: (worst quality,low quality:1.4),(depth of field,blurry:1.2),(greyscale,monochrome:1.1),3d face,cropped,lowres,text,jpeg artifacts,signature,ng_deepnegative_v1_75t, (badhandv4:1.2), (worst quality:2), (low quality:2), (normal quality:2), lowres, bad anatomy, bad hands, normal quality, ((monochrome)), ((grayscale)) watermark,username,blurry,artist name,trademark,fat
Steps: 145, Sampler: DPM++ 3M SDE Exponential, CFG scale: 22.5, Seed: 3523505705, Size: 1280x640, Model hash: 0eca235400, Model: SDXLPixelArtBeta_v10, Denoising strength: 1, Clip skip: 2, ControlNet 0: "Module: none, Model: None, Weight: 2, Resize Mode: Crop and Resize, Low Vram: False, Guidance Start: 0, Guidance End: 0.94, Pixel Perfect: False, Control Mode: ControlNet is more important", Version: v1.6.0例5：插画线稿上色
原图+效果：
生成信息：
tianmeng,deers,mountain,sunlight,cloud,pine,Dunhuang
Negative prompt: face,people
Steps: 116, Sampler: Euler a, CFG scale: 7, Seed: 2003434558, Size: 912x540, Model hash: f517a1a289, Model: 首发推荐｜SHMILY梦幻水彩_v1.0, VAE hash: 735e4c3a44, VAE: vae-ft-mse-840000-ema-pruned.safetensors, Denoising strength: 0.81, Clip skip: 2, ENSD: 31337, ControlNet 0: "Module: invert (from white bg & black line), Model: control_v11p_sd15_lineart [43d4be0d], Weight: 1, Resize Mode: Crop and Resize, Low Vram: False, Processor Res: 512, Guidance Start: 0, Guidance End: 0.9, Pixel Perfect: True, Control Mode: Balanced", Version: v1.6.0例6：统一风格图标生成
通过现有的类似BDicon的图标模型生成
效果：
生成信息
Air Force Blue|gold icon,many details, octane render, transparent glass texture,frosted glass, transparent technology sense, industrial design, white background, studio lighting, sunshine, flat, minimal, quasi-object, axisymmetric, blender, C4D, best quality, 4k
Negative prompt: lowres, bad anatomy, ((bad hands)), (worst quality:2), (low quality:2), (normal quality:2), paintings, sketches, lowres, bad anatomy, bad hands, text, error, missing fingers,
Steps: 20, Size: 1920x1920, Seed: 2894304261, Model: Bdicon_V1.0 , Sampler: Euler a, CFG scale: 7例7：用户头像/数字形象(人物或动物)生成
人物效果
生成信息
chahua,a young girl,blue sky,((wooden)),clean water,sunlight,green tube dress,cute,((bikini)),((green mountain)),black hair,
Negative prompt: badhandsv5-neg,EasyNegative,(low quality:2),(normal quality:2),lowers,normal quality,((monochrome)),((grayscale)),teeth,,facing away,looking away,text,error,extra digit,fewer digits,cropped,jpeg artifacts,signature,watermark,username,blurry,skin spots,acnes,skin blemishes,bad anatomy,fat,bad feet,poorly drawn hands,poorly drawn face,mutation,deformedtilted head,bad hands,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,missing fingers,missing arms,missing legs,extra leg,extra foot,
Steps: 73, Sampler: DPM++ 2M Karras, CFG scale: 7, Seed: 19550441, Size: 1280x880, Model hash: 503c83d793, Model: 填色大师Coloring_master_anime_v1.0, Denoising strength: 0.98, Clip skip: 2, ControlNet 0: "Module: lineart_standard (from white bg & black line), Model: control_v11p_sd15_lineart [43d4be0d], Weight: 1, Resize Mode: Crop and Resize, Low Vram: False, Processor Res: 512, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: Balanced", Lora hashes: "MW_小清新插画风格_v1.0: bdea0a1e37ca", TI hashes: "EasyNegative: c74b4e810b03", Version: v1.6.0 动物效果
生成信息：
(Masterpiece:1.2, high quality), orange cat, :3, , simple cat,cake on desk,black ground,blue background,white wall,cake on desk
Negative prompt: EasyNegative , badhandsv5-neg
Steps: 94, Sampler: DPM++ 3M SDE Karras, CFG scale: 8, Seed: 19550441, Size: 512x512, Model hash: 503c83d793, Model: 填色大师Coloring_master_anime_v1.0, Denoising strength: 0.98, Clip skip: 2, ControlNet 0: "Module: lineart_standard (from white bg & black line), Model: control_v11p_sd15_lineart [43d4be0d], Weight: 1, Resize Mode: Crop and Resize, Low Vram: False, Processor Res: 512, Guidance Start: 0, Guidance End: 0.81, Pixel Perfect: True, Control Mode: Balanced", Lora hashes: "cat_20230627113759: 939a01b898c5", TI hashes: "EasyNegative: c74b4e810b03", Version: v1.6.0 例8：VR 360全景图生成
效果：
步骤：
- 通过360 lora模型可以生成全景图
- 通过现有的ai放大工具进一步放大生成超清图
an architectural rendering of a bedroom interior, night, qxj
Steps: 20, Size: 1024x1024, Seed: 4214209492, Tile X: True, Tile Y: False, Sampler: DPM++ SDE Karras, CFG scale: 7, Model hash: 7f16bbcd80, ControlNet Model: control_canny-fp16 [e3fe7712], Seed resize from: 512x512, ControlNet Module: canny, ControlNet Weight: 1, ControlNet Enabled: True, Denoising strength: 0.5, Start Tiling From Step: 9, Stop Tiling After Step: -1, ControlNet Guidance End: 1, ControlNet Guidance Start: 0 例9：AI摄影/婚纱/照相馆 换脸
利用sd controlnet + roop插件 实现妙鸭相机的换脸功能
参考b站教程 sd工作流进阶
例10： 图像多姿势调整出图
利用openpose/poseX控制骨骼图生成来调整360°姿势
效果：
参考b站教程【AI绘画】openpose新插件poseX
最后
目前试下来，生成图像的大小对生成速度影响很大，想要得到高清图需要进一步借助ai放大工具进行调整，否则很影响出图效率。
总之，SD图片生成在交互上具有广泛的应用前景和深远的意义。通过使用预训练的模型来生成图片，从而提高产品的交出产出效率。相信在不久的将来，AIGC将会在更多的领域得到应用，为人们的生活带来更多的便利和创新。