Coder Social home page Coder Social logo

sdwebuiapi's Introduction

sdwebuiapi

基于gin+gorm+mysql+sd模型的Golang API客户端，可二开

流程图

https://cdn.staticaly.com/gh/4927525/images@master/20230628/whiteboard_exported_image.xu44suwx6i8.jpg

后端项目架构

├── api                 (api层)
    │   └── v1          (v1版本接口)
    ├── cache           (缓存包)
    ├── config          (配置包)
    ├── dao             (数据库操作层)
    ├── define          (全局常量)
    ├── docs            (swagger文档目录)
    ├── e               (错误代码)                    
    ├── helper          (助手方法包)                    
    ├── logs            (日志文件)                    
    ├── middleware      (中间件层)                        
    ├── model           (模型层)             
    ├── resource        (静态资源库)             
    ├── router          (路由层)                    
    ├── serializer      (序列化请求相应参数)                    
    ├── server          (系统启动)                    
    ├── service         (service层)     
    ├── sql             (sql表结构文件)                    
    ├── test            (单元测试)                    
    └── utils           (工具包)                    
        ├── jwt         (JWT)     
        ├── logger      (日志包)                      
        └── translate   (阿里云机翻翻译)

项目运行

普通运行

go run main.go

以二进制文件运行

go mod tidy
go build -o main
./main

主要功能

  • 文生图
  • 图生图
  • 局部重绘
  • 切换pt风格
  • 切换服务器模型
  • 阿里云机翻
  • OSS存储
  • 异步队列
  • 定时任务
  • 失败重试
  • 插队功能
  • swagger

项目规划

  • 加入kafka或是rabbitmq，出图结果异步回调通知服务器
  • 抽离 service 的结构体到 types，引入 sync.Once 模块，重构 service 层
  • 加入docker-compose
  • 加入 Jaeger 进行链路追踪
  • 加入 Skywalking 监控中间件
  • 加入ELK体系，方便日志查看和管理
  • 加入前端代码
  • base642url方法补充

生命周期

  • 初始化redis、mysql、env配置类
  • 初始化排队队列
  • 初始化路由
  • 绑定端口启动服务

主要依赖

名称 版本
golang 1.20v
gin v1.9.0
gorm v1.23.9
mysql v1.3.6
redis v6.15.9
jwt-go v3.2.0
crypto v0.8.0
logrus v1.9.0
gin-swagger v1.6.0
cron v1.2.0
darabonba-openapi v2.0.1

配置文件

server:
  app: dev
  port: :8009
database:
  host: 127.0.0.1
  port: 3306
  user: root
  dbname: ces
  pwd: root
redis:
  addr: 127.0.0.1:6379
  pwd: root
  dbname: 0

Postman导入

{
	"info": {
		"_postman_id": "13f1f635-8514-4ec7-b68d-64d9cc628951",
		"name": "sddemo",
		"schema": "https://schema.getpostman.com/json/collection/v2.1.0/collection.json",
		"_exporter_id": "5152396"
	},
	"item": [
		{
			"name": "txt2img",
			"request": {
				"method": "POST",
				"header": [],
				"body": {
					"mode": "raw",
					"raw": "{\r\n    \"prompts\": \"((masterpiece)), (((best quality))), ((ultra-detailed)), ((illustration)), dusk, in the cyberpunk city, girl, veil, hair_ornament,  artbook, real, realistic, earrings, black choker ,shackles,  crop top,\",\r\n    \"size\": 2,\r\n    \"model_id\":1\r\n}",
					"options": {
						"raw": {
							"language": "json"
						}
					}
				},
				"url": {
					"raw": "http://127.0.0.1:8009/api/v1/sd/txt2img",
					"protocol": "http",
					"host": [
						"127",
						"0",
						"0",
						"1"
					],
					"port": "8009",
					"path": [
						"api",
						"v1",
						"sd",
						"txt2img"
					]
				}
			},
			"response": []
		},
		{
			"name": "img2img",
			"request": {
				"method": "POST",
				"header": [],
				"body": {
					"mode": "raw",
					"raw": "{\r\n    \"prompts\": \"color purple\",\r\n    \"model_id\": 1,\r\n    \"denoising_strength\":0.25,\r\n    \"inpainting_fill\":0,\r\n    \"size\": 3,\r\n    // \"mask_url\":\"https://ossstatic.leiting.com/static/wd/home/202207/pc/role11.png\",\r\n    \"original_url\":\"https://ossstatic.leiting.com/static/wd/home/202207/pc/role11.png\"\r\n}",
					"options": {
						"raw": {
							"language": "json"
						}
					}
				},
				"url": {
					"raw": "http://127.0.0.1:8009/api/v1/sd/img2img",
					"protocol": "http",
					"host": [
						"127",
						"0",
						"0",
						"1"
					],
					"port": "8009",
					"path": [
						"api",
						"v1",
						"sd",
						"img2img"
					]
				}
			},
			"response": []
		}
	]
}

txt2img

https://cdn.staticaly.com/gh/4927525/images@master/20230628/image-(1).7d38vap1xzw0.jpg

img2img

https://cdn.staticaly.com/gh/4927525/images@master/20230628/image-(2).6ky5ep82ykk0.jpg

API

官网API介绍：https://github.com/AUTOMATIC1111/stable-diffusion-webui/wiki/API

原生API

txt2img

文生图

URI

/sdapi/v1/txt2img

请求参数(JSON)

{
  "enable_hr": false,
  "denoising_strength": 0,
  "firstphase_width": 0,
  "firstphase_height": 0,
  "hr_scale": 2,
  "hr_upscaler": "string",
  "hr_second_pass_steps": 0,
  "hr_resize_x": 0,
  "hr_resize_y": 0,
  "prompt": "",
  "styles": [
    "string"
  ],
  "seed": -1,
  "subseed": -1,
  "subseed_strength": 0,
  "seed_resize_from_h": -1,
  "seed_resize_from_w": -1,
  "sampler_name": "string",
  "batch_size": 1,
  "n_iter": 1,
  "steps": 50,
  "cfg_scale": 7,
  "width": 512,
  "height": 512,
  "restore_faces": false,
  "tiling": false,
  "negative_prompt": "string",
  "eta": 0,
  "s_churn": 0,
  "s_tmax": 0,
  "s_tmin": 0,
  "s_noise": 1,
  "override_settings": {},
  "override_settings_restore_afterwards": true,
  "script_args": [],
  "sampler_index": "Euler",
  "script_name": "string"
}

请求参数说明

参数 类型 是否必填 备注
prompt string Y 正向提示词
negative_prompt string Y 反向提示词
denoising_strength float Y 重绘幅度(Denoising strength)，取值范围：0.00-1.00图像模仿自由度，越高越自由发挥，越低和参考图像越接近。
seed int Y 种子随机数，默认-1 每次生成的不一样
cfg_scale int Y 提示词相关性，默认7一般默认就好，如果提示词比较具体（又长又臭）就调到8、9试试，越低越有创造性
steps int Y 降噪步骤数，推荐 50更多的去噪步骤通常会导致更高质量的图像 较慢的推理费用
sampler_index string Y 采样器索引
sampler_name string Y 采样器名称
batch_size int Y 生成批次
n_iter int Y 每批数量
tiling boolean Y 用于生成一个可以平铺的图像。
width int Y 宽度
height int Y 高度
restore_faces boolean Y 面部修复

响应参数(JSON)

{
    "images":[
        "iVBORw0KGgoAAAA..." // 不带base64头的图片字符串
    ],
    "info":{
        "all_prompts":null,
        "all_seeds":null,
        "all_subseeds":null,
        "batch_size":0,
        "cfg_scale":0,
        "clip_skip":0,
        "denoising_strength":0,
        "extra_generation_params":{

        },
        "face_restoration_model":null,
        "height":0,
        "index_of_first_image":0,
        "infotexts":null,
        "job_timestamp":"",
        "negative_prompt":"",
        "prompt":"",
        "restore_faces":false,
        "sampler":"",
        "sampler_index":0,
        "sd_model_hash":"",
        "seed":0,
        "seed_resize_from_h":0,
        "seed_resize_from_w":0,
        "steps":0,
        "styles":null,
        "subseed":0,
        "subseed_strength":0,
        "width":0
    },
    "parameters":{
        "batch_size":1,
        "cfg_scale":0,
        "denoising_strength":0,
        "enable_hr":false,
        "eta":0,
        "firstphase_height":0,
        "firstphase_width":0,
        "height":768,
        "n_iter":1,
        "negative_prompt":"",
        "override_settings":{

        },
        "prompt":"8k,((masterpiece)), (((best quality))), ((ultra-detailed)), ((illustration)), dusk, in the cyberpunk city, girl, veil, hair_ornament,  artbook, real, realistic, earrings, black choker ,shackles,  crop top,",
        "restore_faces":false,
        "s_churn":0,
        "s_noise":0,
        "s_tmax":0,
        "s_tmin":0,
        "sampler_index":"Euler",
        "seed":-1,
        "seed_resize_from_h":-1,
        "seed_resize_from_w":-1,
        "steps":45,
        "styles":null,
        "subseed":-1,
        "subseed_strength":0,
        "tiling":false,
        "width":1024
    }
}

响应参数说明

images
参数 类型 是否必填 备注
images list Y 生成的图片list
images.index string Y 生成的图片base64字符串不带base64头
paramenters
参数 类型 是否必填 备注
prompt string Y 正向提示词
negative_prompt string Y 反向提示词
denoising_strength float Y 重绘幅度(Denoising strength)，取值范围：0.00-1.00图像模仿自由度，越高越自由发挥，越低和参考图像越接近。
seed int Y 种子随机数，默认-1 每次生成的不一样
cfg_scale int Y 提示词相关性，默认7一般默认就好，如果提示词比较具体（又长又臭）就调到8、9试试，越低越有创造性
steps int Y 降噪步骤数，推荐 50更多的去噪步骤通常会导致更高质量的图像 较慢的推理费用
sampler_index string Y 采样器索引
sampler_name string Y 采样器名称
batch_size int Y 生成批次
n_iter int Y 每批数量
tiling boolean Y 用于生成一个可以平铺的图像。
width int Y 宽度
height int Y 高度
restore_faces boolean Y 面部修复

img2img（inpaint）

图生图（局部重绘）

URI

/sdapi/v1/img2img

请求参数(JSON)

{
  "init_images": [
    "string"
  ],
  "resize_mode": 0,
  "denoising_strength": 0.75,
  "image_cfg_scale": 0,
  "mask": "string",
  "mask_blur": 4,
  "inpainting_fill": 0, // 是否重绘
  "inpaint_full_res": true,
  "inpaint_full_res_padding": 0,
  "inpainting_mask_invert": 0,
  "initial_noise_multiplier": 0,
  "prompt": "",
  "styles": [
    "string"
  ],
  "seed": -1,
  "subseed": -1,
  "subseed_strength": 0,
  "seed_resize_from_h": -1,
  "seed_resize_from_w": -1,
  "sampler_name": "string",
  "batch_size": 1,
  "n_iter": 1,
  "steps": 50,
  "cfg_scale": 7,
  "width": 512,
  "height": 512,
  "restore_faces": false,
  "tiling": false,
  "negative_prompt": "string",
  "eta": 0,
  "s_churn": 0,
  "s_tmax": 0,
  "s_tmin": 0,
  "s_noise": 1,
  "override_settings": {},
  "override_settings_restore_afterwards": true,
  "script_args": [],
  "sampler_index": "Euler",
  "include_init_images": false,
  "script_name": "string"
}

请求参数说明

参数 类型 是否必填 备注
prompt string Y 正向提示词
negative_prompt string Y 反向提示词
init_images list Y 底图列表
init_images.index string Y 底图base64字符串
resize_mode int Y 缩放模式拉伸｜Just Resize简单的将图片缩放至指定比例，不保证原图尺寸比。裁剪｜Crop And Resize将图片按比例进行缩放，多余的直接裁剪掉。填充｜Resize And Fill将图片按比例进行缩放，缺少的部分填充。隐空间直接缩放｜Latent upScale其实与前面三个不同，这个是常用于图像超分辨率的快捷选项，低显存谨慎使用。
inpainting_fill int Y 是否局部重绘0：否1：是
mask string Y 遮罩层带base64头的base64字符串
denoising_strength float Y 重绘幅度(Denoising strength)，取值范围：0.00-1.00图像模仿自由度，越高越自由发挥，越低和参考图像越接近。
seed int Y 种子随机数，默认-1 每次生成的不一样
cfg_scale int Y 提示词相关性，默认7一般默认就好，如果提示词比较具体（又长又臭）就调到8、9试试，越低越有创造性
steps int Y 降噪步骤数，推荐 50更多的去噪步骤通常会导致更高质量的图像 较慢的推理费用
sampler_index string Y 采样器索引
sampler_name string Y 采样器名称
batch_size int Y 生成批次
n_iter int Y 每批数量
tiling boolean Y 用于生成一个可以平铺的图像。
width int Y 宽度
height int Y 高度
restore_faces boolean Y 面部修复

响应参数(JSON)

{
    "images":[
        "iVBORw0KGgoAAAANSUhEUgAAAwAAAAM..."
    ],
    "info":"{\"prompt\": \"8k,color purple,\", \"all_prompts\": [\"8k,,color purple,\"], \"negative_prompt\": \"\", \"all_negative_prompts\": [\"\"], \"seed\": 1732878691, \"all_seeds\": [1732878691], \"subseed\": 3147614705, \"all_subseeds\": [3147614705], \"subseed_strength\": 0.0, \"width\": 768, \"height\": 768, \"sampler_name\": \"Euler\", \"cfg_scale\": 10.0, \"steps\": 45, \"batch_size\": 1, \"restore_faces\": false, \"face_restoration_model\": null, \"sd_model_hash\": \"d55fbed80e\", \"seed_resize_from_w\": -1, \"seed_resize_from_h\": -1, \"denoising_strength\": 0.25, \"extra_generation_params\": {}, \"index_of_first_image\": 0, \"infotexts\": [\"8k,,color purple,\\nSteps: 45, Sampler: Euler, CFG scale: 10.0, Seed: 1732878691, Size: 768x768, Model hash: d55fbed80e, Seed resize from: -1x-1, Denoising strength: 0.25, Clip skip: 2, ENSD: 31337\"], \"styles\": [], \"job_timestamp\": \"20230615161236\", \"clip_skip\": 2, \"is_using_inpainting_conditioning\": false}",
    "parameters":{
        "batch_size":1,
        "cfg_scale":0,
        "denoising_strength":0.25,
        "eta":0,
        "height":768,
        "include_init_images":false,
        "init_images":null,
        "inpaint_full_res":true,
        "inpaint_full_res_padding":0,
        "inpainting_fill":0,
        "inpainting_mask_invert":0,
        "mask":null,
        "mask_blur":4,
        "n_iter":1,
        "negative_prompt":"",
        "override_settings":{

        },
        "prompt":"8k,color purple,",
        "resize_mode":1,
        "restore_faces":false,
        "s_churn":0,
        "s_noise":0,
        "s_tmax":0,
        "s_tmin":0,
        "sampler_index":"Euler",
        "seed":-1,
        "seed_resize_from_h":-1,
        "seed_resize_from_w":-1,
        "steps":45,
        "styles":[

        ],
        "subseed":-1,
        "subseed_strength":0,
        "tiling":false,
        "width":768
    }
}

响应参数说明

images
参数 类型 是否必填 备注
images list Y 生成的图片list
images.index string Y 生成的图片base64字符串不带base64头
paramenters
参数 类型 是否必填 备注
prompt string Y 正向提示词
negative_prompt string Y 反向提示词
denoising_strength float Y 重绘幅度(Denoising strength)，取值范围：0.00-1.00图像模仿自由度，越高越自由发挥，越低和参考图像越接近。
seed int Y 种子随机数，默认-1 每次生成的不一样
cfg_scale int Y 提示词相关性，默认7一般默认就好，如果提示词比较具体（又长又臭）就调到8、9试试，越低越有创造性
steps int Y 降噪步骤数，推荐 50更多的去噪步骤通常会导致更高质量的图像 较慢的推理费用
sampler_index string Y 采样器索引
sampler_name string Y 采样器名称
batch_size int Y 生成批次
n_iter int Y 每批数量
tiling boolean Y 用于生成一个可以平铺的图像。
width int Y 宽度
height int Y 高度
restore_faces boolean Y 面部修复
inpainting_fill int Y 是否局部重绘0：否1：是

应用API

http://localhost:8009/swagger/index.html

https://cdn.staticaly.com/gh/4927525/images@master/20230628/image.k72oz4u7q4g.jpg

sdwebuiapi's People

Contributors

4927525 avatar

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.