Files
SnapOtter/apps/docs/zh-TW/tools/image/transparency-fixer.md
T
SnapOtterandGitHub 4963ab3bbd feat(docs-i18n): translate all documentation into 20 languages
All 181 docs markdown files translated into 20 languages (apps/docs/<locale>/**). Companion to the i18n code PR; admin-merged because the file count exceeds GitHub's per-PR CI trigger limit. Validated by pnpm i18n:check (all surfaces, 0 stale/missing) and a clean all-locale docs build.
2026-07-11 13:52:47 +08:00

79 lines
2.7 KiB
Markdown
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
---
description: "使用 AI 去背(BiRefNet)修正假透明 PNG 以產生真正的 alpha 通道,並附帶去邊緣清理。"
i18n_source_hash: 7eb748b80f93
i18n_provenance: human
i18n_output_hash: 48d860859f01
---
# PNG 透明度修正 {#png-transparency-fixer}
一鍵修正假透明 PNG。使用 AI 去背(BiRefNet HR Matting 模型)產生真正的 alpha 透明度,並以去邊緣後處理清理邊緣。
## API 端點 {#api-endpoint}
`POST /api/v1/tools/image/transparency-fixer`
**處理方式:** 非同步(回傳 202,透過 SSE 輪詢 `/api/v1/jobs/{jobId}/progress` 以取得狀態)
**模型套件:** `background-removal`4-5 GB
## 參數 {#parameters}
| 參數 | 型別 | 必填 | 預設值 | 說明 |
|-----------|------|----------|---------|-------------|
| file | file | 是 | - | 影像檔案(multipart |
| defringe | number | 否 | `30` | 去邊緣強度(0-100)。移除邊緣周圍的半透明邊緣像素 |
| outputFormat | string | 否 | `"png"` | 輸出格式:`png``webp` |
| removeWatermark | boolean | 否 | `false` | 套用浮水印移除前處理(中值濾波) |
## 範例請求 {#example-request}
```bash
curl -X POST http://localhost:1349/api/v1/tools/image/transparency-fixer \
-F "file=@fake-transparent.png" \
-F 'settings={"defringe":40,"outputFormat":"png"}'
```
## 回應 {#response}
### 初始回應(202 Accepted {#initial-response-202-accepted}
```json
{
"jobId": "a1b2c3d4-e5f6-7890-abcd-ef1234567890",
"async": true
}
```
### 進度(SSE 位於 `/api/v1/jobs/{jobId}/progress` {#progress-sse-at-api-v1-jobs-jobid-progress}
```
event: progress
data: {"phase":"processing","stage":"Processing transparency...","percent":50}
```
### 最終結果(透過 SSE {#final-result-via-sse}
```json
{
"phase": "complete",
"percent": 100,
"result": {
"jobId": "a1b2c3d4-e5f6-7890-abcd-ef1234567890",
"downloadUrl": "/api/v1/download/{jobId}/fake-transparent_fixed.png",
"originalSize": 180000,
"processedSize": 150000,
"filename": "fake-transparent.png"
}
}
```
## 注意事項 {#notes}
- 需要安裝 `background-removal` 模型套件(4-5 GB)。
- 使用 `birefnet-hr-matting` 作為高品質 alpha 去背的主要模型。若 HR 模型記憶體不足,則回退到 `birefnet-general`
- `defringe` 選項會移除 AI 去背有時在頭髮、毛皮與細緻邊緣周圍留下的半透明邊緣像素。其運作方式為模糊 alpha 通道並將低信心像素歸零。
- `removeWatermark` 選項會套用中值濾波前處理步驟。這是基本的浮水印減弱,並非專用的浮水印移除工具。
- 只輸出 PNG 或無損 WebP(兩者皆支援 alpha 透明度)。
- 透過自動解碼支援 HEIC/HEIF、RAW、TGA、PSD、EXR 與 HDR 輸入格式。