---
title: "通过 RunAPI 在龙虾 (OpenClaw) 中使用 Recraft — 超分与背景移除"
url: "https://runapi.ai/zh-CN/openclaw-recraft.md"
canonical: "https://runapi.ai/zh-CN/openclaw-recraft"
locale: "zh-CN"
model: "recraft"
---

# 在 OpenClaw 中使用 Recraft。

Recraft 提供清晰的神经网络图像超分，可在无伪影的情况下恢复精细细节，并提供精确的背景移除，保留发丝和透明边缘。OpenClaw agent 通过与聊天相同的 RunAPI 密钥调用这两个端点——发送源图像 URL，选择超分或背景移除，然后轮询获取处理后的结果。

## API example

```bash
curl -X POST https://runapi.ai/api/v1/recraft/upscale_image \
  -H "Authorization: Bearer $RUNAPI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "recraft-crisp-upscale",
    "source_image_url": "https://example.com/product-photo.jpg"
  }'

```

### Response

```json
{
  "task_id": "tsk_abc123",
  "status": "pending",
  "model": "recraft-crisp-upscale"
}

```

## How it works

1. **Configure RunAPI** — If you have not set up RunAPI in OpenClaw yet, follow the OpenClaw setup guide. Set the RUNAPI_API_KEY environment variable and add RunAPI as an OpenAI-compatible provider with baseUrl https://runapi.ai/v1.
2. **Call Recraft upscale or remove background** — Send a POST request to the upscale_image endpoint with model set to recraft-crisp-upscale and provide the source_image_url. For background removal, POST to the remove_background endpoint with model recraft-remove-background instead. Both require only source_image_url.
3. **Poll for the result** — Both endpoints return a task_id with status pending. Poll the task status endpoint until the status changes to completed, then retrieve the processed image URL from the response. RunAPI SDKs and the CLI handle polling automatically.

## Parameters

| Parameter | Type | Description |
|-----------|------|-------------|
| `model` | `string` | Required. recraft-crisp-upscale for upscaling, recraft-remove-background for background removal. |
| `source_image_url` | `string` | Required. URL of the source image to process. |
| `callback_url` | `string` | Optional. Webhook URL for async completion notification. |

## FAQ

### Can I chain Recraft background removal and upscaling in one OpenClaw workflow?

Yes. Both endpoints are stateless and accept standard image URLs. Instruct the OpenClaw agent to first call remove_background with the original image, then pass the output URL to upscale_image. The same RUNAPI_API_KEY handles both calls.

### What image formats does Recraft accept and output?

Recraft accepts JPEG, PNG, and WebP inputs. Background removal outputs PNG with alpha transparency. Upscale output matches the input format.

### How does Recraft crisp upscale differ from Topaz upscale on RunAPI?

Both use neural networks, but Recraft crisp upscale is optimized for clean edge recovery on design assets and product photos with lower per-task cost. Topaz offers configurable 2x/4x/8x scale factors and is better suited for photographic detail at extreme enlargements. Both are async and share the same RunAPI key.

### Is Recraft processing async or sync?

Both Recraft endpoints are async. The API returns a task_id immediately. Poll the task status endpoint until the result is ready. RunAPI SDKs and the CLI handle polling automatically.

### What does Recraft cost on RunAPI?

Recraft uses per-task billing — upscaling and background removal are priced separately per image processed. Check the RunAPI pricing page for current rates. No subscription or minimum spend required.


## Links

- [OpenClaw 配置指南 →](https://runapi.ai/zh-CN/openclaw)
- [Recraft 模型 →](https://runapi.ai/zh-CN/models/recraft)
- [Model catalog](https://runapi.ai/zh-CN/models)
- [API docs](https://runapi.ai/zh-CN/docs)