muapi Logomuapi Logo
muapi Logomuapi Logo

Seedream 5.5 Layer Decomposition: AI 图像编辑器

Split any image into a base image and transparent PNG layers with the Seedream 5.5 Layer Decomposition API. Pay per generation, no subscription. Seedream 5.5 Layer Decomposition splits a single image into a base image plus transparent PNG layers for compositing, asset extraction and layered design workflows.

交互式模型控制

Seedream 5.5 Layer Decomposition splits a single image into a base image plus transparent PNG layers for compositing, asset extraction and layered design workflows.

📝

概览

关于此模型

Seedream 5.5 Layer Decomposition separates a single flat image into a base image and multiple transparent PNG layers, ready for compositing, asset extraction and layered design workflows. Optionally describe the layers you want with a prompt, or leave it empty so the model detects key visual elements automatically. Choose 1K, 1.5K or 2K output and cap the number of returned images with max_images. Generate source images with Seedream 5.5 or revise them with Seedream 5.5 Edit.

1Design Production: Turn flat posters and ads into editable layered components.
2Asset Extraction: Pull reusable objects, people and products out of campaign images.
3Compositing: Separate foreground and background elements for rearranging.
4Creative Editing: Restyle or replace individual layers without regenerating the whole image.
💰

价格与价值

成本分析

muapiapp$0.025 per returned image slot (1K), default 30 = $0.75

Priced by max_images x per-image rate for the chosen resolution.

Fal.ai暂不可用

No dedicated Seedream 5.5 endpoint confirmed at this time.

Replicate暂不可用

No dedicated Seedream 5.5 endpoint confirmed at this time.

** 竞品价格根据相似模型架构和使用层级估算。

⚙️

技术详情

配置参数

图像 URLstring

Source image to decompose (exactly 1; PNG or JPEG).

默认值https://static.aiquickdraw.com/tools/example/1786019968051_cKRYLHHu.png
提示词string

Optional. Describe the layers, targets or grouping you want. Leave empty to auto-detect key elements.

默认值Separate the headline text, the subtitle, the croissant photo and the decorative border into their own layers
分辨率枚举（4 个选项）

Output resolution tier: auto (keeps input size, billed at 2K), 1K, 1.5K or 2K.

默认值1K
Max Imagesinteger

Maximum images to return (base image + layers), 2-100. You are charged for max_images, not the number returned.

默认值30
输出格式枚举（2 个选项）

Format of the base image only; layers are always transparent PNG.

默认值png
📖

实施指南

开发者文档

How to Use the Seedream 5.5 Layer Decomposition API

  1. Provide a source image as image_url (PNG or JPEG).

  2. Optionally describe layers in prompt, e.g. headline text, subject, border.

  3. Set resolution (1K, 1.5K, 2K) and max_images (2-100, default 30, base image included). You are charged for max_images, so set it to the layers you need plus one.

  4. Submit and poll the request_id; the output lists the base image first, then the layers.

curl -X POST https://api.muapi.ai/api/v1/seedream-5.5-layer \
  -H "x-api-key: $MUAPI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"image_url":"https://example.com/poster.png","prompt":"Separate the headline, the photo and the border","max_images":6}'
❓

常见问题

常见问答

What does Seedream 5.5 Layer Decomposition do?

It splits one image into a base image plus transparent PNG layers for compositing and editing.

How is it priced?

Per image up to `max_images`: $0.02 per image at 1K and 1.5K, $0.035 at 2K. The default 30 images costs $0.60 at 1K. You are charged for max_images even if fewer layers are returned.

Do I need a prompt?

No. Without a prompt the model detects key visual elements automatically.

What format are the layers?

Every layer is a PNG with an alpha channel; `output_format` only applies to the base image.

Does Seedream 5.5 support a spicy mode?

The Seedream 5.5 API applies standard content-safety moderation, so explicit adult output is not guaranteed and prompts that trigger the filter are rejected and refunded. If you need a less restricted tier for mature creative work, see the Spicy models on Muapi, such as [Seedance 2.5 Spicy](/seedance-2.5-spicy) for video.

How do I get the result?

Requests are asynchronous: submit a job, then poll the returned `request_id` or pass a `webhook` URL to be notified on completion.

Is there a sandbox for testing?

Yes. Sandbox API keys return mock results instantly with no credits deducted.