Seedream 5.5 Layers API for Editable Layer Splits

Turn one flat png or jpeg into an editable layered file with POST /api/v1/images/generations and model: seedream-5-5-layers. You get a base image and up to 99 transparent PNG layers, each with a z_index, pixel and normalized bounding boxes, a name and a description. Turn on text_recognition to also get the text and font styles of clean text layers.
Get API Key
Poster separated into stacked translucent panels on a studio table

A Layered File You Can Edit and Localize

Up to 99 layers, you set the cap

max_layers (1 to 99, default 16) caps how many layers come back, not counting the base. Elements beyond the cap may be merged or dropped. Optional prompt names what to split, in plain language or with <bbox> coordinates.
Sheets of tracing paper each holding one element of a drawing, fanned out

Text recognition for text layers

Set text_recognition: true and clean text layers return the recognized text, each line's box and font styles. Stylized or handwritten lettering is not recognized. It is free.
Printed sale banner with its headline letters lifted off as separate pieces

Ready-to-place layers

The base image has z_index 0; layers ascend from 1. bounding_box.absolute gives [left, top, right, bottom] in base pixels and bounding_box.normalized uses a 0-1000 scale. Layers are always transparent PNG; output_format only sets the base.
Theater stage with flat painted scenery panels at different depths
model and image_list are required; everything else tunes the split.

Seedream 5.5 Layers Request Fields and Limits

model

Always seedream-5-5-layers.

image_list (required)

Exactly one image as [{"url": "https://…"}]: png or jpeg, up to 30MB, at least 512x512 pixels in total.

prompt

Optional, up to 4000 characters, naming elements to split in plain language or with <bbox> coordinates.

image_size

auto (default, follows the input), 1K, 1.5K or 2K. Pixel values such as 2048x2048 are rejected.

max_layers and text_recognition

max_layers: 1 to 99, default 16, not counting the base. text_recognition: false (default) or true.

output_format

jpeg (default) or png for the base image. Every layer is a transparent PNG.
Charged per request: one input image plus max_layers + 1 output images.

Seedream 5.5 Layers API Pricing

31.36 tokens ($0.6272) with defaults

At auto or 2K with the default 16 layers: one input at 0.42 tokens ($0.0084) plus 17 outputs at 1.82 tokens ($0.0364) each.

18.10 tokens ($0.3620) at 1K or 1.5K

With the default 16 layers: one input at 0.42 tokens ($0.0084) plus 17 outputs at 1.04 tokens ($0.0208) each.

Lower max_layers costs less

Each output image is 1.04 tokens ($0.0208) at 1K or 1.5K and 1.82 tokens ($0.0364) at 2K or auto, plus 0.42 tokens ($0.0084) for the input. 2K with 5 layers costs 11.34 tokens ($0.2268).

Billed on the cap, refunded on failure

The charge is the same however many layers actually come back. Failed requests are refunded automatically; text_recognition is free.

Integrate the Seedream 5.5 Layers API

1

1. Create an API key

Generate a key at /keys and send it as Authorization: Bearer <key>.
2

2. Build the request

Start from {"model": "seedream-5-5-layers", "image_list": [{"url": "…"}], "max_layers": 16}; add text_recognition: true for posters with text.
3

3. Submit and wait

POST to /api/v1/images/generations, save img_uuid, then poll GET /api/v1/jobs/detail or use custom_callback_url. A split takes about two to three minutes, longer with more layers.
4

4. Rebuild the stack

Sort the results by z_index and place each layer with its bounding box. Download files before link_expired_at, 24 hours after completion.
Create an API Key
Workflows that need editable files from flat images.

What Developers Build with Seedream 5.5 Layers

Poster localization

Split a finished poster, read its text with text_recognition, and swap in translated copy.

Design editor import

Turn a flat image into movable layers inside a canvas editor.

Ad variants

Reuse the same background and product layers while changing headlines or layout.

Asset extraction

Pull individual objects out of an image as transparent PNGs for reuse.
Layers, text and pricing.

Seedream 5.5 Layers API FAQ

What is the Seedream 5.5 Layers API endpoint?

POST https://api.goenhance.ai/api/v1/images/generations with model: seedream-5-5-layers.

How many layers can I get?

Up to 99 plus the base image. max_layers sets the cap and defaults to 16.

Why is the price fixed per request?

The number of layers is only known after the split, so a request is billed as max_layers + 1 output images plus one input. Lower max_layers to pay less.

Which text does text_recognition read?

Clean text layers only. Stylized, decorative or handwritten lettering returns no recognition result.

Can I pass ratio or background?

No. Layers follow the input image's aspect ratio and are always transparent PNG. Use seedream-5-5 for normal or transparent generation.

What input does it accept?

Exactly one png or jpeg, up to 30MB and at least 512x512 pixels in total.

Start Building with the Seedream 5.5 Layers API