A browser-first way to reduce a green photo cast — and when local correction is not enough

A strong yellow-green photo treatment is not always one problem.

For a mild or moderate cast, ordinary pixel correction may be enough. It is fast, deterministic, and can stay entirely in the browser. But when a heavy effect has compressed colors or changed visible details, moving RGB sliders cannot reconstruct information that is no longer present.

I built Matcha Filter Remover around that distinction:

• Low and Medium run locally in the browser.
• AI-assisted High is optional for difficult photos.
• Every AI result is labeled as an edit, not a recovered original.
• The before/after comparison stays visible before download.
• The tool refuses face reveal, hidden-content recovery, nudification, and non-consensual uses.

The hardest product decision was not model selection. It was deciding when to stop claiming “correction” and start explaining that a generated result may change faces, hands, text, clothing, lighting, or background details.

I would value feedback on two questions:

1. Is the local-versus-AI split clear before upload?
2. Does the comparison make the limitations obvious enough?

The current version supports JPEG, PNG, and WebP up to 10 MB. It is an independent tool and is not affiliated with TikTok or any filter platform.