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

Files

T

SnapOtterandGitHub ec98dfde9a feat(landing,docs): SEO fixes and query-matched tool pages (#383 )

Landing:
- Preload the Bricolage display font (LCP heading) to cut first paint
- Cap tool-page meta descriptions at ~160 chars via whole-sentence
  truncation, so SERP snippets no longer overflow or cut mid-phrase
- Drop dead HowTo + FAQPage JSON-LD from tool pages (Google removed
  HowTo rich results in 2023 and FAQ rich results in 2026); the visible
  content stays. Breadcrumb + WebApplication schema retained
- Shorten 22 tool titles that exceeded the ~60-char SERP limit
- Add an optional h1 override to tool SEO data
- Rewrite adjust-colors, smart-crop, and ringtone-maker to match real
  Search Console query intent (saturation; object-aware cropping;
  convert audio to ringtone), with query-matched H1s and FAQs

Docs:
- Add meta descriptions to 20 pages missing frontmatter

Repo:
- Ignore local credential JSON files (.secrets/, *client_secret*, *service_account*)

2026-07-01 16:20:05 +08:00

2.2 KiB

Raw Blame History

description

description
Remove unwanted objects from images with AI inpainting (LaMa), guided by a mask of the region to erase.

Object Eraser

Remove unwanted objects from images using AI inpainting (LaMa model). Accepts an image and a mask indicating the region to erase.

API Endpoint

POST /api/v1/tools/image/erase-object

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

Model bundle: object-eraser-colorize (1-2 GB)

Parameters

Parameter	Type	Required	Default	Description
file	file	Yes	-	Source image file (multipart)
mask	file	Yes	-	Mask image (white = area to erase, black = keep). Must be uploaded with fieldname `mask`
format	string	No	`"auto"`	Output format: `auto`, `png`, `jpg`, `jpeg`, `webp`, `tiff`, `gif`, `avif`, `heic`, `heif`, `jxl`
quality	integer	No	`95`	Output quality (1-100)

Example Request

curl -X POST http://localhost:1349/api/v1/tools/image/erase-object \
  -F "file=@photo.jpg" \
  -F "mask=@mask.png" \
  -F "format=png" \
  -F "quality=95"

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":"Inpainting...","percent":70}

Final Result (via SSE)

{
  "phase": "complete",
  "percent": 100,
  "result": {
    "jobId": "a1b2c3d4-e5f6-7890-abcd-ef1234567890",
    "downloadUrl": "/api/v1/download/{jobId}/photo_erased.png",
    "previewUrl": "/api/v1/download/{jobId}/preview.webp",
    "originalSize": 245000,
    "processedSize": 230000
  }
}

Notes

Requires the object-eraser-colorize model bundle to be installed (1-2 GB).
The mask must be the same dimensions as the source image. White pixels indicate areas to erase; the AI fills them with plausible content.
Uses LaMa (Large Mask Inpainting) for high-quality object removal.
For non-browser-previewable output formats, a WebP preview is generated alongside the main output.
Supports HEIC/HEIF, RAW, TGA, PSD, EXR, and HDR input formats via automatic decoding.

2.2 KiB Raw Blame History