ToAPIs
home
Log InStart Free

Quick Answer

  • Call the async image task API with model=doubao-seedream-5-5.
  • size accepts 1K/1.5K/2K tiers or explicit WxH pixels within 921,600-4,624,220 total pixels.
  • metadata.background=transparent returns RGBA PNG. For layers, max_images limits total outputs including the base: 2-100, default 30.

Key Parameters

  • model | string | required | doubao-seedream-5-5 | - | Use doubao-seedream-5-5 as the exact model id.
  • prompt | string | optional | - | - | Required for generation and editing; optional when layer_decomposition=true.
  • size | string | optional | - | 1K | 1.5K | 2K | WIDTHxHEIGHT | Uses the model default for generation and auto for layers. Tiers do not fix the aspect ratio; explicit dimensions must total 921,600-4,624,220 pixels.
  • image_urls | string[] | optional | - | 0-10 | Supply 1-10 JPEG, PNG, or WEBP reference images through image_urls. Upload local files first, then use the returned URLs.
  • n | integer | optional | 1 | 1 | Use 1 or omit it. Ordinary generation returns one image; this does not set the layer count.
  • layer_decomposition | boolean | optional | false | - | Keep off for ordinary generation. Requires exactly one reference and no metadata.background. Returns a base and transparent PNG layers; layer_decomposition_options.max_images limits total outputs.
  • layer_decomposition_options.max_images | integer | optional | 30 | 2-100 | Total output limit for layer decomposition, including the base. It is not a guaranteed count; the model determines how many outputs are needed.
  • layer_decomposition_options.text_recognition | boolean | optional | - | - | Controls text recognition during layer decomposition. Task results include text_recognition metadata only when the provider returns it.
  • metadata.output_format | string | optional | - | png | jpeg | When omitted, uses the upstream default; transparency selects png. For layers, controls only the base image; other layers are always PNG.
  • metadata.background | string | optional | - | auto | opaque | transparent | Generation and editing only. Omit when layer_decomposition=true, even opaque. transparent returns RGBA PNG and cannot use jpeg.
  • metadata.watermark | boolean | optional | - | - | Set false to disable the visible watermark.
  • metadata.response_format | string | optional | url | url | b64_json | Result delivery form; url is recommended.

Common Errors

  • 400 invalid_input: trigger=size pixels out of range, more than 10 reference images, or layer_decomposition without exactly one input image.; fix=Validate size, image_urls count, and layer_decomposition input constraints against docs.; retry=Retry only after correcting the payload.
  • 429 rate_limit_exceeded: trigger=Request rate, concurrency, or quota hits rate limiting.; fix=Apply exponential backoff, then review request pacing and quota usage.; retry=Use 1s/2s/4s backoff with jitter.
ByteDance Seed logo

Seedream 5.5

Commercial use

Generate images from text or edit with up to 10 references. Output transparent RGBA directly and preserve untouched pixels across edits. Layer decomposition splits one reference image into a base image and separate transparent PNG layers.

Model Type:
Text To ImageImage To Image
Pricing: 1K / 1.5K $0.016; 2K $0.028
Live
1

Describe your image

Reference images

The prompt is optional for layer decomposition. Leave it empty to decompose the reference image.172 / 1200

Advanced options

Example params
Layer Decomposition
Text recognition

Layer decomposition requires exactly one reference image and does not support background settings. Turn it off for ordinary image generation.

2

Your generated result

Example result
View API DocsGet API Key

Generation takes some time. Please wait.You can leave this page. [ Task Logs ] lets you check the generated content directly.

Request preview

{
  "model": "doubao-seedream-5-5",
  "prompt": "A sheet of embroidered denim patch stickers on a transparent background: teddy bear, daisy, cherries, rainbow, butterfly, hearts and stars, with realistic stitching texture",
  "size": "2K",
  "metadata": {
    "watermark": false,
    "background": "transparent"
  }
}

5.5

Model version

10

Max reference images

100

Max outputs, including base

2K

Max output tier

Generated examples

Explore what Seedream 5.5 can create

Create with this model
Transparent embroidered sticker sheet generated through the ToAPIs Seedream 5.5 API

ToAPIs API output

Seedream 5.5 official showcase of transparent PNG asset generation

Official showcase

Seedream 5.5 official showcase of targeted image editing

Official showcase

Transparent pricing

Seedream 5.5 pricing

Billing is per generated image, at 20% below the ByteDance official list price for each output size tier.

Top up credits

1K / 1.5K

ToAPIs

$0.016

image

Save 20%VS

ByteDance official list

$0.02

image

2K

ToAPIs

$0.028

image

Save 20%VS

ByteDance official list

$0.035

image

Ordinary generation returns one image; layer decomposition returns a base image and PNG layers. Billing settles when the task completes.

Why ToAPIs

Why use the Seedream 5.5 API on ToAPIs?

Build production image workflows with transparent output, stable iterative editing, and per-image billing through one endpoint.

Native transparent assets

Set metadata.background to transparent and receive RGBA PNG output directly — stickers, icons, and product cutouts without a separate cutout step.

Pixel-stable multi-round editing

Chain successive edits on the same image: unedited regions stay consistent round after round, guided by up to 10 reference images.

Editable layer output

Split one image into a base and transparent PNG layers. layer_decomposition_options.max_images sets a total limit of 2-100 outputs, including the base, with a default of 30. The actual count is determined by the model.

Call the Seedream 5.5 API in three steps

Seedream 5.5 runs as an asynchronous image task on ToAPIs: submit, poll, and download the result.

  1. 1

    Create an API key

    Generate a key in the ToAPIs dashboard and call POST /v1/images/generations with model set to doubao-seedream-5-5.

  2. 2

    Pick a workflow and parameters

    Choose text-to-image, reference editing with up to 10 image_urls, or layer_decomposition. Set size to 1K, 1.5K, 2K, or an explicit WxH value, and use metadata for output_format, background, and watermark.

  3. 3

    Poll the task and fetch the image

    The submit response returns a task_id. Poll the task endpoint until status is completed, then download the result image URL.

Best-fit Seedream 5.5 use cases

Transparent stickers and assets

Produce RGBA stickers, icons, logos, and product cutouts that drop straight into design files or storefronts.

Iterative retouching workflows

Apply successive edits to portraits, interiors, and product shots while untouched areas stay pixel-consistent.

Layer extraction for design pipelines

Decompose finished artwork into a base image plus transparent PNG layers ready for Figma, Photoshop, or print production.

FAQ

Seedream 5.5 API FAQ

Which model ID calls Seedream 5.5 on ToAPIs?

Submit POST /v1/images/generations with model set to doubao-seedream-5-5. The API is asynchronous: the submit response returns a task_id that you poll until the task completes.

What sizes does Seedream 5.5 support?

Pass size as the 1K, 1.5K, or 2K preset, or an explicit WxH string such as 2048x2048. Total pixels must stay between 921,600 and 4,624,220.

How many reference images can a request include?

Up to 10 images via image_urls (images and reference_images are accepted aliases). JPEG and PNG inputs are supported, and every response returns one result image.

How do I generate a transparent PNG?

Set metadata.background to "transparent". Without an explicit format the API returns an RGBA PNG; you can also set metadata.output_format to "png" explicitly.

What does layer_decomposition return?

Set layer_decomposition=true with exactly one reference image. Results contain a base and transparent PNG layers. layer_decomposition_options.max_images limits the total to 2-100 outputs including the base, default 30; the model determines the actual count. The prompt is optional. Omit metadata.background. layer_decomposition_options.text_recognition controls OCR, and OCR metadata appears when returned by the provider.

Does Seedream 5.5 keep unedited regions stable across edits?

Yes. Seedream 5.5 is built for multi-round editing: regions outside the edit stay consistent across successive changes, so iterative retouching does not drift.

How much does Seedream 5.5 cost on ToAPIs?

Pricing is per generated image at 20% below the official list price: $0.016 per image for 1K and 1.5K outputs (official $0.02) and $0.028 per image at 2K (official $0.035).

How do I turn off the watermark?

Set metadata.watermark to false in the request to receive output without the visible watermark.

Seedream 5.5 API

Build with Seedream 5.5 on ToAPIs

Try Seedream 5.5 in the Playground or read the API reference to integrate transparent output, reference editing, and layer decomposition into your image pipeline.

Try nowView API Docs

Seedream 5.5 API on ToAPIs

One endpoint covers text-to-image, multi-reference editing, transparent RGBA output, and layer decomposition with per-image pricing.

Try Seedream 5.5