高光照片优化_修复旧照片_照片换底色

Upload Images(Multiple images are optional)

上传照片将在24小时内自动删除，请及时下载修复结果

All Delete all Continue uploading images

Enhance color photo 清晰度修复画质修复

Change background color of ID photo Enhance face 尺寸放大

Colorize black&white photo Repair scratched 去雾化

Lossless amplification

Case Display

Enhance color photo

Colorize black&white photo

Enhance color photo

Common Problem

Preparation before repair 1. Rotate the photo to the correct angle and cut off the useless edges, which can greatly improve the AI repair effect.

2. The photo that is already in color does NOT need to be colored again: As AI repair is an algorithmic restoration, it will be different from the original image, and it will be inconsistent with the original image after it is colored again. Therefore, it is recommended that color photos do not need to be colored.
3. After paying, you can download the complete picture without a watermark in high-definition. The payment means that you have agreed to the effect of the comparison picture. The same photo (photo of the same format, size) only needs to pay once. If the photo does not enter your photo album after it is displayed, please manually long-press the photo to save it into your photo album. If you still cannot save it, please add our QQ group (623946846) for assistance.

what is the AI repair ? AI repair of photos is the use of artificial intelligence technology to repair pictures, making old pictures glow with new vitality. For example, before the 1980s, many videos and photos were in black and white. AI repair technology can be used to repair old black and white photos with yellow and blurry pictures, and old black and white photos can be restored into color digital photos. You can also use AI technology to supplement the details of the originally blurred photos to restore high-definition photos.

What is the principle of AI photo repair? AI will automatically learn and remember the color of these key information, such as leaves are green, human faces are skin color, etc. Therefore, it can quickly start to colorize scenes in black and white photos. AI learns through algorithms and training of a large number of existing pictures. First perform image segmentation on the photo to distinguish landmark objects. For example, trees, sky, human faces, clothing... Although it is difficult to be perfect, the restored old photos can still give people a lot of surprises.
PS: Because AI repair is the "experience" of algorithm learning and the result of deduction, it will be different from real photos and real scenes.

What kind of photo is suitable for AI repair? Black and white photos, blurry portrait photos, headshots, half-length photos, etc.

What photos are not suitable for AI? Since blurry photo repairing sharpness is mainly for the training of photos of people, photos of pure scenery without people can be colored, but sharpness cannot be repaired.
The original photo is too blurry, or the portrait of the person in the original photo is too small, or the photo itself is too small to be repaired.
Text-only pictures cannot be repaired. Since no special training has been carried out for text-only pictures, and no targeted algorithms have been designed, it is impossible to repair text-only pictures.

Can the old video be repaired? Of course, this is our corporate service, you can contact huogang@hiliad.com
case show:

Bilibili：https://www.bilibili.com/video/BV13g41157hL
Youtube：https://www.youtube.com/watch?v=XtHbzENmb74

PriceAPI

3 photo = ￥49.8Expiry:one yearPrioritize:YESSize:6000pxOld repair:YESBatch:NOWechatRebate

10 photos = ￥150Expiry:one yearPrioritize:YESSize:6000pxOld repair:YESBatch:YESWechatRebate

50 photos = ￥400Expiry:one yearPrioritize:YESSize:6000pxOld repair:YESBatch:YESWechatRebate

200 photos = ￥1000Expiry:one yearPrioritize:YESSize:6000pxOld repair:YESBatch:YESWechatRebate

For lower prices, please join the QQ group (623946846) for consultation

准备

1、线下付款，获取可上传图片次数
2、申请请求接口所需的third_code和secret_key
3、根据接口文档，接入服务

说明

1、按照图片上传次数扣费，上传失败不扣费
2、获取到的secret_key请妥善保管，不可外传
3、所有修复成功的图片本站将保留1小时后自动删除，如需持久化存储请下载至自有系统

接口说明

1、请求地址: http://sdkphoto.fangtangtv.com
2、code=200固定为成功，code=9000固定为失败
3、code=9000时，如需进一步业务判断，可根据errno参数

1、图片上传接口

URL /api/third/uploads
Method POST
Header token: 加密生成的token，加密规则见附1
Params
field	type	required	commit
third_code	string	Y	线下申请的第三方标识third_code
seq	string	Y	任务参数, 多值以英文逗号连接, 可选参数: repair, colour, sr, bg, scratch, defogging
type	string	N	图像类型, 默认 people, 可选参数: people, cortoon, landscape, portrait, text
file	file	Y	图像文件, 可传多次
rate	int	N	超分倍数, 默认 0, 可选参数: 2, 4, 8
rgb	string	N	底色值, bg in seq 时必传, 可选参数: -1, 1, 235,51,35(红), 85,142,213(蓝), 255,255,255(白)
dpi	int	N	分辨率, 300DPI则传300
platform	string	Y	平台, 可选参数: pc, h5, mp, ios, baidu, zijie, douyin, alipay
clientip	string	N	客户端IP
mdf	string	N	再次处理, 必传, 可传多次

Curl

curl --location --request POST 'http://sdkphoto.fangtangtv.com/api/third/uploads' --header 'token: xxx' --form 'third_code="test"' --form 'seq="bg"' --form 'file=@"/VSCode/image/3c6b4b786c1159441a94e3508c36508b.png"' --form 'rgb="1"' --form 'platform="pc"'

Response

field	commit
img_url	上传图片的url
mdf	图片的唯一标识
{
    "code": 200,
    "data": {
        "mdf": [
	"40b37f58db153e5149f008421797714e"
	]
	}
}

失败时(errno说明)

errno	commit
9999	请求参数token不合法
9101	剩余下载次数为0
9001	传入参数错误
9005	文件上传类型不支持
9006	上传失败
9009	图片异常，处理失败
{
    "code": 9000,
    "errno": 9005,
    "msg": "文件上传类型不支持"
}

2、图片修复状态查询

URL /api/third/status
Method POST
Header token: 加密生成的token，加密规则见附1
Params
field	type	required	commit
third_code	string	Y	线下申请的第三方标识third_code
mdf	string	Y	文件唯一标识
platform	string	Y	平台, 可选参数: pc, h5, mp, ios, baidu, zijie, douyin, alipay

Curl

curl --location --request POST 'http://sdkphoto.fangtangtv.com/api/third/status' --header 'token: xxx' --form 'third_code="test"' --form 'mdf="9804b4cb5b4740fb26afde5f2831c0f5"' --form 'platform="pc"'

Response(正常时)

{
	"code": 200,
	"data": {
	"msg": "获取任务状态成功",
	"status": 200
	"mdfs": [
	{
		"mdf":"9804b4cb5b4740fb26afde5f2831c0f5",
		"state": "任务处理完成",
		"src_url": "",
		"wmk_url":"",
		"icon_url": "",
		"res_size": "1080x1366px 146KB"
	}
	]
}
	}

Response(失败时)errno说明

errno	commit
9002	图片不存在
500	其他错误
{
    "code": 9000,
    "errno": 9002,
    "msg": "图片不存在，请确认图片是否上传成功"
}

3、图片下载

URL /api/third/download
Method POST
Header token: 加密生成的token，加密规则见附1
Params
field	type	required	commit
third_code	string	Y	线下申请的第三方标识third_code
mdf	string	Y	文件唯一标识

Curl

curl --location --request POST 'http://sdkphoto.fangtangtv.com/api/third/download' --header 'token: xxx' --form 'third_code="test"' --form 'mdf="9804b4cb5b4740fb26afde5f2831c0f5"'

Response(正常时)

field	commit
url	图片下载地址
{
	"code": 200,
	"data": {
	"url": "https://sdkphoto.fangtangtv.com/api/third/download/40b37f58db153e5149f008421797714e01100.jpg"
	}
}

Response(失败时)errno说明

errno	commit
9002	图片不存在
500	其他错误
{
    "code": 9000,
    "errno": 9002,
    "msg": "数据已失效，请重新上传"
}

4、剩余上传次数查询

URL /api/third/company/nums
Method POST
Header token: 加密生成的token，加密规则见附1
Params
field	type	required	commit
third_code	string	Y	线下申请的第三方标识third_code

Curl

curl --location --request POST 'XXXXX/api/third/company/nums' \
--header 'token: XXXXX' \
--form 'third_code="XXXXX"'

Response(正常时)

field	commit
code	第三方标识third_code
nums	剩余上传次数
{
    "code": 200,
    "data": {
        "code": "hiliad",
        "nums": 98
    }
}

Response(失败时)errno说明

errno	commit
9002	数据未找到
500	其他错误
{
    "code": 9000,
    "errno": 9002,
    "msg": "数据未找到"
}

附1、token生成规则

1、拿到服务方提供的secret_key，再生成一个当前时间戳time
    secret_key = 'hiliad'
    time = '1629962357'
2、secret_key + time,两字符串拼接，再md5，得到参数mds
    mds = hashlib.md5((secret_key + time).encode("utf8")).hexdigest()
    mds = '5484490c8e797dabe52af90ab8bd1e50'
3、最终的token等于 mds + ',' + time
    token = '5484490c8e797dabe52af90ab8bd1e50,1629962357'

注：生成的时间戳会在2分钟后失效，请每次请求接口都重新生成时间戳