v1.0.30

GargantuaX/gemini-watermark-removerv1.0.30Jul 16, 2026by GargantuaX

AI Summary

Improves candidate selection for exact-96px watermarks and introduces outline-aware alpha profiles. The release also adds SDK-level signals for quality feedback to help clients avoid meaningless retries.

Key Highlights

Improved same-anchor candidate selection for exact-96px watermarks.
Added conservative outline-aware alpha profiles and contour repair.
Added SDK-level signals for visible residuals and possible content damage.
Expanded regression coverage for candidate ordering and deterministic selection.

New Features

Outline-aware alpha profiles
SDK-level quality signals (residual/content damage)

Full Release Notes

## Image watermark removal

- Improved same-anchor candidate selection for newer exact-96px Gemini watermarks, including light- and dark-outline variants.
- Added conservative outline-aware alpha profiles and contour repair while retaining the canonical result when an alternative is only marginally better.
- Reduced visible dark-star artifacts without broadening candidate position changes.

## Quality feedback

- Added SDK-level signals for visible residuals and possible content damage so clients can expose best-effort quality and avoid encouraging meaningless retries.
- Expanded regression coverage for candidate ordering, outline variants, deterministic selection, release evidence, and package metadata.

## Release scope

- Image defaults and image quality diagnostics are included in this release.
- Video defaults remain unchanged, and this release does not claim new video or V2 capabilities.

The production build, SDK smoke tests, full automated suite, scoped image-quality gate, and GitHub CI all passed for this release.