Split any image into a base plus clean separate layers with the Seedream 5.0 Flash Layer Decomposition API. Prompt or bbox targeting. Pay per generation.
Decompose a composite image into a base image plus clean, separate layer assets with Seedream 5.0 Flash Layer Decomposition. Optionally target elements with prompts or <bbox> regions.
About this model
Seedream 5.0 Flash Layer Decomposition separates a composite image into a base image and independent transparent PNG layers for text, subjects, and graphic elements. Leave the prompt empty to auto-detect the main elements, or describe them and pin exact regions with <bbox>x1 y1 x2 y2</bbox> tags on a 0-1000 normalized grid. It is the fast, low-cost tier alongside Seedream 5.0 Pro Layer Decomposition.
Cost analysis
| Provider | Cost | Notes |
|---|---|---|
| muapiapp | $0.459/generation | Pay-per-generation pricing, no subscription required. |
| Fal.ai | Not available | No dedicated Seedream 5.0 Flash endpoint at this time. |
| Replicate | Not available | No dedicated Seedream 5.0 Flash endpoint at this time. |
Pay-per-generation pricing, no subscription required.
No dedicated Seedream 5.0 Flash endpoint at this time.
No dedicated Seedream 5.0 Flash endpoint at this time.
** Competitor pricing is estimated based on similar model architectures and usage tiers.
Configuration schema
| Parameter | Type | Description | Default |
|---|---|---|---|
| Prompt | string | Optional. Describe which elements to separate; supports <bbox>x1 y1 x2 y2</bbox> (0-1000 normalized). Leave empty to auto-detect main elements. | Separate the title text <bbox>179 58 809 197</bbox> and the parrot <bbox>330 274 641 991</bbox> into independent layers |
| Image URL | string | Source image to decompose (exactly 1; PNG, JPEG, WebP, BMP, TIFF, GIF; max 30MB). | https://static.aiquickdraw.com/tools/example/1786019968051_cKRYLHHu.png |
| Resolution | Enum (4 options) | Target image resolution level. | auto |
| Output Format | Enum (2 options) | Format of the base image only; separated layers are always PNG. | jpeg |
Optional. Describe which elements to separate; supports <bbox>x1 y1 x2 y2</bbox> (0-1000 normalized). Leave empty to auto-detect main elements.
Separate the title text <bbox>179 58 809 197</bbox> and the parrot <bbox>330 274 641 991</bbox> into independent layersSource image to decompose (exactly 1; PNG, JPEG, WebP, BMP, TIFF, GIF; max 30MB).
https://static.aiquickdraw.com/tools/example/1786019968051_cKRYLHHu.pngTarget image resolution level.
autoFormat of the base image only; separated layers are always PNG.
jpegDeveloper documentation
Provide the source image as image_url (one image, max 30MB).
Optionally add a prompt, e.g. with <bbox> regions, or leave empty to auto-detect.
Choose size auto, 1K, 1.5K, or 2K.
Submit and poll request_id; the output contains the base image and each layer.
curl -X POST https://api.muapi.ai/api/v1/seedream-5.0-flash-layerize \
-H "Authorization: Bearer $MUAPI_API_KEY" \
-H "Content-Type: application/json" \
-d '{"image_url":"https://static.aiquickdraw.com/tools/example/1786019968051_cKRYLHHu.png","prompt":"Separate the title text <bbox>179 58 809 197</bbox> and the parrot <bbox>330 274 641 991</bbox> into independent layers","resolution":"2K"}'
Frequently asked
It splits one image into a base image plus separate PNG layers, either auto-detected or targeted by prompt and bbox coordinates.
Pay per generation with no subscription. Pricing is a flat rate per run at every resolution and size.
1K and 2K output through the `resolution` parameter (layer decomposition uses `auto`, `1K`, `1.5K` or `2K`).
Requests are asynchronous: submit a job, then poll `request_id` or pass a `webhook` URL to be notified on completion.
`png` or `jpeg` via `output_format`.
Yes — sandbox API keys return mock results instantly with no credits deducted.