Skip to main content
AI image processing model
ByteDance
Seedream

Seedream 5.0 Pro Layer Decomposition

Seedream 5.0 Pro Layer Decomposition is designed for one existing image. It produces a base image and exports detected elements as transparent PNG layers. Kyeo validates the true file type, total pixel count, and aspect ratio before the asset enters the task pipeline.

Input
Exactly 1 image
Output
Base image + transparent layers
Tiers
auto/1K/1.5K/2K
Credits
7–14
Accepts exactly 1 image and stays separate from standard multi-image editing
Supports PNG, JPEG, WebP, BMP, TIFF, and GIF up to 30 MB
Requires 262,144–36,000,000 total pixels and a 1:16–16:1 aspect ratio
Lets the base image use JPEG/PNG while extracted layers remain transparent PNG
30-second overview
Seedream
7–14 credits
What it does best
A structured image-processing workflow for post-production, intended to produce visual elements that can be edited, moved, or reused separately.
Best for
Advertising composites, product-image revisions, reusing illustration elements, breaking down layouts, and assets that need further work in design software.
Popular searches
how to use Seedream layer decompositionhow to split an image into layers with AISeedream Layer Decomposition supported formats

At a glance

What is this model like?

Leave the prompt empty to let the model detect prominent elements automatically, describe the objects to extract, or use a bounding box to identify a more precise region. `output_format` controls only whether the base image is JPEG or PNG. Extracted layers are always transparent PNG files, so the option does not set every output's format.

What it does best
A structured image-processing workflow for post-production, intended to produce visual elements that can be edited, moved, or reused separately.
Best for
Advertising composites, product-image revisions, reusing illustration elements, breaking down layouts, and assets that need further work in design software.
Why use it on Kyeo AI
Kyeo binds the single-image, true MIME, 30 MB, pixel, and aspect-ratio constraints to the upload signature and validates them again before charging. auto uses a conservative 14-credit preauthorization.

Key facts

Quickly assess whether this model fits your use case.

Model category
AI image layer decomposition
Vendor
ByteDance
Workflow
Single-image layering, separate from standard image to image
File formats
PNG/JPEG/WebP/BMP/TIFF/GIF
Pixel range
262,144–36,000,000
Aspect ratio
1:16–16:1

Also known as

The same model may appear under different names across documentation and community discussions; this list keeps them easy to search and compare.

Seedream Layer Decomposition
Seedream layer separation
Seedream 5 Pro image layering

Common questions

These practical questions focus on the task, input conditions, and result requirements you should confirm before choosing.

how to use Seedream layer decomposition
how to split an image into layers with AI
Seedream Layer Decomposition supported formats
Seedream layer decomposition pixel limits
Seedream layer decomposition 1K 2K pricing
what format are the separated image layers

Selection guide

Use these decision points when choosing a model.

1
Choose layer decomposition for editable elements

If people, products, decorations, or backgrounds need separate adjustment, layer decomposition fits post-production better than a standard style redraw.

2
Use a narrower tool for background removal only

If you need only an isolated subject rather than several layers, compare Recraft Remove Background for a simpler workflow.

3
Budget 14 credits for auto

The actual auto tier is unknown before task creation, so the system preauthorizes at the 2K ceiling. Choose 1K, 1.5K, or 2K explicitly when a fixed budget matters more.

Popular comparisons

Compare common alternatives on the same task to clarify differences in inputs, controls, and cost.

How does layer decomposition differ from standard image editing?
How should I choose between auto and 1K/1.5K/2K?
Can Seedream Layer Decomposition process multiple images at once?

Model comparison

Compare the current model with alternatives at a glance.

Workflow
Seedream 5.0 Pro Layer Decomposition
Splits one image into a base image and separate layers
Seedream 5.0 Pro
Text generation and multi-image editing
Recraft Remove Background
Removes only the background; does not separate multiple elements
Input count
Seedream 5.0 Pro Layer Decomposition
Exactly 1 image
Seedream 5.0 Pro
Up to 10 images for editing
Seedream 5.0 Lite
Up to 14 images for editing
Output uses
Seedream 5.0 Pro Layer Decomposition
Compositing, local edits, and element reuse
Topaz Image Upscale
Upscaling and clarity enhancement
Recraft Crisp Upscale
Light upscaling and sharpening

Practical usage insights

Practical guidance based on public sources, current on-site limits, and representative tasks.

Layer count is not guaranteed

The model identifies elements from the image, but the current contract does not promise a fixed number of layers. Complex occlusion, transparent objects, and fine textures still require manual review.

Layered output is not the original design file

The output contains a model-inferred base image and transparent bitmap layers. It does not recover the source vectors, fonts, masks, blend modes, or application project structure.

Capabilities

Automatic or targeted separation

An empty prompt asks the model to detect prominent elements; a description can focus the task on specified objects.

Location guidance

Normalized bounding-box coordinates can identify a region in the prompt, but the actual separation still needs manual verification.

Multiple tiers with conservative billing

1K/1.5K cost 7 credits and 2K costs 14 credits; auto is preauthorized at 14 credits.

Use cases

Rearranging product images

Separate the product, props, and background, then recombine them in design software for horizontal, vertical, or square ads.

Reusing illustration elements

Extract characters, foreground objects, and decorations for motion work, cover variants, or local revisions.

Layout structure analysis

Break a complex visual into independent bitmap layers to inspect occlusion, without treating the result as the original project file.

Prompt tips

Name the target object first

Describe the object's name, color, location, and relationship to nearby elements to reduce confusion between similar subjects.

Use a bounding box when necessary

For scenes with similar objects, mark the target with normalized 0–1000 coordinates and verify the coordinate order.

Validate structure at 1K

Check layer detection and edges at 1K before deciding whether to process the final asset at 2K.

Why choose it

A dedicated workflow avoids mixing layer controls with standard image-editing parameters.
Six input formats and a 30 MB limit cover a broad range of post-production assets.
Server-signed validation checks both image geometry and true file type.
The output structure is suitable for continued work in design and compositing tools.

What to know first

Each task can process only 1 image.
It does not recover fonts, vectors, masks, or original project information.
The contract does not guarantee layer count, naming, or edge quality.
The actual auto tier is unknown before creation, so it uses the 14-credit upper bound for preauthorization.

Adjustable parameters

Quickly assess whether this model fits your use case.

Output tier
size
Optional
Parameter type: Select
Default: auto
auto
1K
1.5K
2K
Base image format
output_format
Optional
Parameter type: Select
Default: JPEG
JPEG
PNG

Credit usage

7–14 credits

1K and 1.5K cost 7 credits; 2K costs 14 credits. Because the selected tier is not known before creation when size is auto, Kyeo preauthorizes 14 credits, then reads actual consumption and settles any difference after the task completes.

Budget tip

Before batch generation, run an A/B test with the same assets across the current model and alternatives to validate quality and avoid wasting credits.

FAQ

Related models

Compare these similar candidates before deciding.

Seedream 5.0 Pro

Seedream 5.0 Pro combines text-to-image generation with editing from up to 10 reference images in one entry point. basic maps to 1K and high to 2K. Editing preauthorization rises with the number of inputs, making it useful for multi-image workflows that need a known cost before submission.

AI image model
7–19 credits

Recraft Remove Background

Recraft Remove Background / Capabilities — Kyeo accepts one JPG, PNG, or WebP image up to 5 MB and charges a fixed 1 credit. Download the result to review transparency and subject edges. 1 JPG, PNG, or WebP image up to 5 MB and 16 MP; each dimension must be 256–4096 px.

AI image model
1 credit per request

Topaz Image Upscaler

Topaz Image Upscaler / Capabilities — Topaz Image Upscaler accepts one image and one scale factor on Kyeo AI. It does not read a prompt or change the subject or composition. The values 1, 2, and 4 are width-and-height multipliers, not fixed 1K, 2K, or 4K outputs; calculate the output by multiplying the source width and height by the selected factor.

AI image model
Conservative 40-credit preauthorization; refund by actual usage

Sources

Content is based on model documentation, feature references, and the settings available on Kyeo AI.

Source note

This page covers only the single-image layer-decomposition workflow currently integrated into Kyeo. It is a dedicated image-processing entry point, not standard text-to-image or multi-image editing. Input geometry, true file type, and size are validated during upload and again at submission.

Last updated: 2026-08-11
Seedream Layer Decomposition current interface notes
Platform

Used to check the input modes, visible controls, media limits, and availability of the version currently connected to this site. The page and workbench show the actual options.