mirror of https://github.com/snapotter-hq/SnapOtter.git synced 2026-08-03 07:46:42 +02:00

Files

T

SnapOtterandGitHub 00b651c9f8 feat(i18n): 21-language pipeline, landing/docs/API wiring, landing+API translations

Shared Claude Code translation pipeline (scripts/i18n, no API key) plus Astro/VitePress/Scalar i18n wiring. Landing and API reference translated into all 20 languages; docs i18n wiring + English source anchors. The translated docs markdown (apps/docs/<locale>/**, 3,620 files) follows in a companion PR because it exceeds GitHub's per-PR CI file limit.

2026-07-11 13:01:55 +08:00

2.7 KiB

Raw Blame History

description

description
Fix fake transparent PNGs with AI matting (BiRefNet) to produce true alpha, plus defringe edge cleanup.

PNG Transparency Fixer

Fix fake transparent PNGs in one click. Uses AI matting (BiRefNet HR Matting model) to produce true alpha transparency, with defringe post-processing to clean up edges.

API Endpoint

POST /api/v1/tools/image/transparency-fixer

Processing: Asynchronous (returns 202, poll /api/v1/jobs/{jobId}/progress for status via SSE)

Model bundle: background-removal (4-5 GB)

Parameters

Parameter	Type	Required	Default	Description
file	file	Yes	-	Image file (multipart)
defringe	number	No	`30`	Defringe intensity (0-100). Removes semi-transparent fringe pixels around edges
outputFormat	string	No	`"png"`	Output format: `png` or `webp`
removeWatermark	boolean	No	`false`	Apply watermark removal pre-processing (median filter)

Example Request

curl -X POST http://localhost:1349/api/v1/tools/image/transparency-fixer \
  -F "file=@fake-transparent.png" \
  -F 'settings={"defringe":40,"outputFormat":"png"}'

Response

Initial Response (202 Accepted)

{
  "jobId": "a1b2c3d4-e5f6-7890-abcd-ef1234567890",
  "async": true
}

Progress (SSE at `/api/v1/jobs/{jobId}/progress`)

event: progress
data: {"phase":"processing","stage":"Processing transparency...","percent":50}

Final Result (via SSE)

{
  "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"
  }
}

Notes

Requires the background-removal model bundle to be installed (4-5 GB).
Uses birefnet-hr-matting as the primary model for high-quality alpha matting. Falls back to birefnet-general if the HR model runs out of memory.
The defringe option removes semi-transparent fringe pixels that AI matting sometimes leaves around hair, fur, and fine edges. It works by blurring the alpha channel and zeroing out low-confidence pixels.
The removeWatermark option applies a median filter pre-processing step. It is a basic watermark reduction, not a dedicated watermark removal tool.
Only outputs PNG or lossless WebP (both support alpha transparency).
Supports HEIC/HEIF, RAW, TGA, PSD, EXR, and HDR input formats via automatic decoding.

2.7 KiB Raw Blame History