Image Editing

Seedream 4.0 image editing uses /v1/images/edits with the reference image passed in the JSON image field, rather than the OpenAI-style file-upload protocol.

Authorization

BearerAuth

AuthorizationBearer <token>

Model relay interface recognition. Request heading: Autoration: Bearer .

In: header

Request Body

multipart/form-data

model*string

Image editing model name.

Default"doubao-seedream-4-0-250828"

prompt*string

Edits a hint.

image*file

Yes. Original graphic file to edit. Current upstream requests for uploading files do not support the direct uploading of pictures URL with JSON.

Formatbinary

mask?file

Optional. Covers a photo file with a transparent area indicating the area to be edited.

Formatbinary

n?integer

Generates the number of pictures.

Formatint32

Range1 <= value

size?string

Output picture size. Please use a clear < width > x > height, e.g. 1920 x 1920; total pixels may not be less than 3686,400, the actual usable value depends on the model and upstream channel.

Default"1920x1920"

quality?string

Picture quality. Actual available values depend on models and upstream pathways.

Default"auto"

response_format?string

returns the format. __FD_PROTECT_0 or _FD_PROTEC_1; default does not need to transmit this field and different upstream channel support may be different.

Value in"url" | "b64_json"

background?string

Background filler. Support depends on the model.

Default"auto"

Value in"transparent" | "opaque" | "auto"

output_format?string

Output picture format. Support depends on the model.

Default"png"

Value in"png" | "jpeg" | "webp"

output_compression?integer

Output compression level, only applicable to _FD_PROTEC_0 and FD_PROTEC_1.

Default90

Formatint32

Range0 <= value <= 100

moderation?string

Security clearance intensity. Support depends on the model.

Default"auto"

Value in"auto" | "low"

watermark?boolean

Add watermarks. Support depends on the upstream route.

Defaultfalse

user?string

End-user identification, which can be used for abuse surveillance.

Response Body

application/json

curl -X POST "https://api.tokaify.com/v1/images/edits" \
  -F model="doubao-seedream-4-0-250828" \
  -F prompt="Change the picture to a watercolored illustration style and keep the main picture." \
  -F image="@/path/to/image.jpg" \
  -F n="1" \
  -F size="1920x1920" \
  -F quality="auto" \
  -F background="auto" \
  -F output_format="png" \
  -F output_compression="90" \
  -F moderation="auto" \
  -F watermark="false" \
  -F user="user-123"

curl -X POST "https://api.tokaify.com/v1/images/edits" \  -F model="doubao-seedream-4-0-250828" \  -F prompt="Change the picture to a watercolored illustration style and keep the main picture." \  -F image="@/path/to/image.jpg" \  -F n="1" \  -F size="1920x1920" \  -F quality="auto" \  -F background="auto" \  -F output_format="png" \  -F output_compression="90" \  -F moderation="auto" \  -F watermark="false" \  -F user="user-123"

{
  "created": 1,
  "data": {}
}

Request Parameters

Field	Type	Required	Default	Description
`model`	string	Yes	None	Fixed to `doubao-seedream-4-0-250828`.
`prompt`	string	Yes	None	Edit instruction, such as replacing the background, preserving the subject, changing style, or blending multiple reference images.
`image`	string or string[]	Yes	None	Source/reference image URL or Base64. For multi-image blending, up to 14 images is recommended.
`size`	string	No	`1920x1920`	Use an explicit `<width>x<height>` value such as `1920x1920`; the total pixel count must be at least `3686400`.
`response_format`	string	No	None	Supports `url` or `b64_json`. Omit by default unless you need a specific return format.
`watermark`	boolean	No	`true`	Whether to add an AI-generated watermark.
`stream`	boolean	No	`false`	Whether to enable streaming output.

Example Code

curl https://api.tokaify.com/v1/images/edits \
  -H "Authorization: Bearer $TOKAIFY_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "doubao-seedream-4-0-250828",
    "prompt": "Keep the main person unchanged and replace the background with a futuristic office in a realistic photography style.",
    "image": "https://example.com/input.png",
    "size": "1920x1920"
  }'

import requests

response = requests.post(
    "https://api.tokaify.com/v1/images/edits",
    headers={"Authorization": "Bearer YOUR_TOKAIFY_API_KEY"},
    json={
        "model": "doubao-seedream-4-0-250828",
        "prompt": "Keep the main person unchanged and replace the background with a futuristic office in a realistic photography style.",
        "image": "https://example.com/input.png",
        "size": "1920x1920",
    },
)
print(response.json())

const response = await fetch("https://api.tokaify.com/v1/images/edits", {
  method: "POST",
  headers: { Authorization: `Bearer ${process.env.TOKAIFY_API_KEY}`, "Content-Type": "application/json" },
  body: JSON.stringify({
    model: "doubao-seedream-4-0-250828",
    prompt: "Keep the main person unchanged and replace the background with a futuristic office in a realistic photography style.",
    image: "https://example.com/input.png",
    size: "1920x1920",
  }),
});
console.log(await response.json());

Response Example

{
  "created": 1777432320,
  "data": [
    {
      "url": "https://api.tokaify.com/assets/images/seedream-edit.png",
      "b64_json": null,
      "revised_prompt": "Keep the main subject, futuristic office background, realistic photography style."
    }
  ]
}

Notes

Billing follows the configured model and channel ratios. More reference images and higher resolutions increase both latency and cost.

Authorization

Request Body

Response Body

200application/json

Request Parameters

Example Code

Response Example

Notes

application/json