**Cost**: Model-priced per successful call. See the Credit Matrix for `image_reframe` model and resolution prices.

Image Reframe supports two input modes:

1. **JSON URL mode**: send `application/json` with the existing URL field.
2. **Direct upload mode**: send `multipart/form-data` with a local image file in the `image` field. The compatible field name `file` is also accepted for the primary image. Codia checks credits before reading and uploading the file, so insufficient-credit requests return `402` without uploading anything.

Direct upload example:

```bash
curl 'https://openapi.codia.ai/v2/open/image/reframe' \
  -H 'Authorization: Bearer {codia_api_key}' \
  -F 'image=@./image.png' \
  -F 'resolution=1024x1024'
```

Outpaint / extend an image to fill a larger canvas at a specified target resolution — generate new pixels on the outside of the original frame rather than scaling it up. Synchronous.

## What this API is for

Converting square assets to wide banner formats, extending portraits into landscape, repurposing a 1:1 product photo as a 16:9 hero image. Different from `upscale` (which adds pixels *inside* the same crop) — `reframe` adds pixels *outside* it.

## Request

| Field | Required | Notes |
|---|---|---|
| `image_url` | yes | Public URL of the source. |
| `resolution` | yes | Requested target dimensions as `WIDTHxHEIGHT`, e.g. `1920x1080`. Actual output dimensions may be adjusted to the sizes supported by the selected model. |
| `model` | no | Default `nano_banana_2`. Also: `nano_banana_pro`, `flux_3`, `ideogram_v3`. |

If `resolution` matches the source's aspect ratio, the result is effectively an upscale (use `upscale` instead). The endpoint shines when the **aspect ratio changes** and net new pixels need to be hallucinated.

## Response

Shared `ImageUrlListSuccess` — one URL.

## Billing & limits

- Credits follow the `image_reframe` model/resolution table in the global Credit Matrix.
- Synchronous.

## Errors

Standard set; `400` on malformed `resolution`.

> **Sync vs. async**
> This synchronous endpoint returns the result inline — the simplest path. For production pipelines that need webhook delivery, automatic retry with idempotency, or explicit task cancellation, call `POST /v2/open/tasks` with `operation=image_reframe` instead. Inputs and credit cost are identical.

選填，由用戶自訂的冪等鍵（最多 255 個字元）。設定後，成功（2xx）的回應會快取 7 天。以相同的鍵 + 相同的請求主體重試時，會返回快取的回應（不另外扣除點數）；以相同的鍵但不同的請求主體重試，則返回 409。可讓 POST 請求在網絡故障時安全重試。對於 multipart/form-data 上載，請求會依表單欄位與檔案內容比對，因此即使客戶端產生新的 multipart boundary，重試的上載仍會重播先前的結果。

Idempotency-Key

image_url

目標解像度，格式為 `WIDTHxHEIGHT`。圖片會延伸以填滿此尺寸。FLUX 2 Pro/Max 也接受 1K/2K 尺寸預設值。過大的請求（包括 4K）會縮減至最多約 4MP（正方形圖片為 2048x2048）；不支援原生 4K 輸出。計費依調整後的輸出面積計算：總像素不超過 1,048,576 時按 1K 價格計費，更大的輸出則按 2K 價格計費。

resolution

執行延伸構圖 / 外擴繪製（outpaint）的模型。預設為 `nano_banana_2`。點數依 `image_reframe` 模型/解像度表計算。

| 模型 | 預設 | 1K | 2K | 4K |
| --- | ---: | ---: | ---: | ---: |
| `gpt_image` (medium) | 15 | 15 | 29 | 48 |
| `gpt_image_2_5` (medium) | 4 | 4 | 8 | 12 |
| `nano_banana_2` | 18 | 18 | 27 | 41 |
| `nano_banana_pro` | 36 | 36 | 36 | 65 |
| `seedream_5_pro` | 12 | 12 | 24 | - |
| `seedream_5_flash` | 5 | 5 | 5 | - |
| `seedream_5` | 10 | 10 | 10 | 10 |
| `seedream_4_5` | 11 | 11 | 11 | 11 |
| `flux_3` | 13 | 13 | 27 | 162 |
| `flux_2_pro` | 8 | 8 | 20 | - |
| `flux_2_max` | 19 | 19 | 43 | - |
| `ideogram_v3` | 16 | 16 | - | - |

GPT 列顯示的是 medium 質素。其他質素級別請參閱上方共用的 GPT Image 點數矩陣。

model

`gpt_image` 可使用 `low`、`medium`、`high` 或 `auto`。`gpt_image_2_5`（包括 Flare 與 Sunburst）另支援 `xhigh` 與 `max`。省略 quality 或設為 `auto` 時，按 medium 計費級別計費。請使用 `resolution` 指定輸出尺寸。

quality

image

相容用的來源圖片檔案欄位（已棄用：請改用 `image`）。

file

未上載圖片時使用的來源圖片 URL

code

message

image_urls

data

請求主體格式錯誤或缺少必要參數

Idempotency-Key 已被用於不同的請求主體。

已達速率限制。請等待 `Retry-After` 標頭中指定的秒數後再重試。

在原圖邊緣之外產生新內容，將圖片延伸至新的尺寸或長寬比。

圖片擴展

輸入

輸出

相關 API