curl --location "https://randomseed.lol/v1/sync/img2img" \
  --header "Content-Type: application/json" \
  --header "Authorization: Bearer $RANDOMSEED_API_KEY" \
  --data '{
    "img2img": {
        "prompt": "Cute cat",
        "negative_prompt": "ugly, out of frame",
        "width": "512",
        "height": "512",
        "number_of_images": "1",
        "steps": "20",
        "cfg_scale": 7,
        "model_name": "realistic_vision_v3_inpainting",
        "sampler_name": "Euler a",
        "include_init_images": true,
        "image_strength": 0.8,
        "image_url": "...",
        "mask": "...",
        "inpaint_full_res": false,
        "inpaint_full_res_padding": 32,
        "inpainting_fill": 0,
        "inpainting_mask_invert": 0,
        "resize_mode": 0,
        "webhook": "...",
        "track_id": 1,
        "seed": "-1"
    }
  }'

{
  "message": "image generation in progress",
  "result": {
    "id": "...", // request id you're going to use for polling if webhook url is not provided
    "status": "IN_QUEUE"
  }
}

​
API key

Create your account on RandomSeed & grab your api key

​
Headers

​
Authorization
string

Your API key. This is required by most endpoints to access our API programatically. You can get your api key by clicking on ‘API reference’ tab under your profile icon.

​
Body

​
prompt
string
required

This is the text prompt for the image you want generated. Loras can be used by adding this in the prompt <lora:LORA_NAME:weight> where LORA_NAME is the name of the lora model and weight can be from 0 to 1, with decimal values supported.

​
negative_prompt
string
default:""

This is the text prompt describing what you don’t want to see in the image.

​
width
string
default:"512"

This is the width of the image

​
height
string
default:"512"

This is the height of the image

​
number_of_images
string
default:"1"

Number of images you want to generate

​
steps
string
default:"50"

This is the number of the denoising steps used by the AI model when generating an image. A higher number results in a longer generation time, and does not necessarily guarantee higher image quality.

​
cfg_scale
float
default:"7.0"

This is the number of the number you specify to control how closely AI model should follow the text prompt. 7 or 7.5 usually works best. You can increase the value if the generated image doesn’t match your prompt. Increment it by 0.5.

​
model_name
string
required

Specify the name of the model you’d like to use. You can find the model names here

​
sampler_name
string
required

This is the name of the sampler method you will be using to generate an image. Image output varies slightly depending on the method. You can pick one of the following methods

​
include_init_images
boolean
default:"true"

Set it to true for inpainting or basic img2img generation.

​
image_strength
float
default:"0.7"

Determines how close the generated image will be to the provided image.

​
image_url
string
required

Link to the image url you want modified

​
mask
string

Link to the image that will be used for inpainting. You need to specify the region of the starting image that you want modified.

​
inpaint_full_res
boolean
default:"false"

If set to true, inpaint area will be the same resolution as the provided image. If set to false, inpaint area will stretch to fit the provided image. This parameter is required for inpainting.

​
inpaint_full_res_padding
integer
default:"32"

Padding around the masked object to inpaint in full resolution This parameter is required for inpainting.

​
inpainting_fill
integer
default:"1"

Determines how masked content should be displayed. It can be either 0, 1, 2, or 3. 0 -> “fill”, 1 -> “original”, 2 -> “latent noise”, 3 -> “latent nothing” This parameter is required for inpainting.

​
inpainting_mask_invert
integer
default:"0"

Determines whether you want to inpaint the masked object or outside of the masked object. 0 will inpaint the masked object. 1 will inpaint the region outside of the masked object. This parameter is required for inpainting.

​
resize_mode
integer
default:"1"

It can be either 0, 1, 2, or 3. 0 -> “just resize”, 1 -> “crop and resize”, 2 -> “resize”, 3 -> “fill” This parameter is required for inpainting.

​
image_cfg_scale
float
default:"0.7"

Determines how much the result resembles your starting image, so a lower value means a stronger effect. This is the exact opposite of cfg_scale.

​
webhook
string

This is the callback url where you want to receive your response.

​
track_id
string

You can specify any value here to track the request that’s received on your webhook’s end.

​
seed
string
default:"-1"

This is the value that determines the output of a random number generator. You can get ALMOST the same image if you provide the same parameters and the seed.

​
Response

​
message
string

image generation in progress

curl --location "https://randomseed.lol/v1/sync/img2img" \
  --header "Content-Type: application/json" \
  --header "Authorization: Bearer $RANDOMSEED_API_KEY" \
  --data '{
    "img2img": {
        "prompt": "Cute cat",
        "negative_prompt": "ugly, out of frame",
        "width": "512",
        "height": "512",
        "number_of_images": "1",
        "steps": "20",
        "cfg_scale": 7,
        "model_name": "realistic_vision_v3_inpainting",
        "sampler_name": "Euler a",
        "include_init_images": true,
        "image_strength": 0.8,
        "image_url": "...",
        "mask": "...",
        "inpaint_full_res": false,
        "inpaint_full_res_padding": 32,
        "inpainting_fill": 0,
        "inpainting_mask_invert": 0,
        "resize_mode": 0,
        "webhook": "...",
        "track_id": 1,
        "seed": "-1"
    }
  }'

{
  "message": "image generation in progress",
  "result": {
    "id": "...", // request id you're going to use for polling if webhook url is not provided
    "status": "IN_QUEUE"
  }
}