Privacy Policy

The short version

• Your photos are analysed on your own Mac or Android phone and are never uploaded to us.
• There is no account, and the app does not collect analytics or usage data.
• The app goes online only to check GitHub for a new version of itself, to download model files the first time a feature needs them, and to reach a NAS or a backup bucket that you set up yourself.
• This website uses no cookies and no analytics.

What the app does with your photos

When you open a folder, FleetingGleam reads the photos in it and analyses them locally: sharpness, exposure, faces, similarity and a visual score. Original files are not modified. Organising photos moves or copies files only when you ask it to.

If you add a NAS in Settings, the app connects to it with the address and account you give, copies the photos you open into a cache on your device, and analyses them there. Organising on a NAS moves files into a recycle folder on that NAS, and exporting to a NAS writes copies there.

What is stored on your device

To avoid repeating work, the app keeps the following in its own data folder on your Mac or phone:

• Thumbnails and larger previews of your photos, and copies of photos opened from a NAS.
• Analysis results such as scores and numeric descriptions of each photo.
• Your keep, review and reject decisions, your projects, and the taste the app has learned from them.
• Settings, and the model files you have downloaded.
• The NAS passwords and backup keys you enter, encrypted with a key that is created on your device and never leaves it.
• Logs and crash reports. They stay on your device. They are shared only if you export a diagnostics file and send it yourself, and passwords, keys and account names are removed from that file.

What we collect

We do not collect your photos, file names, analysis results, decisions or any personal information. None of the data above is sent to us or to anyone else.

Your NAS and your backups

A NAS you add is reached directly over your network with the SMB protocol. Nothing about it is sent to us.

Cloud backup is off unless you set it up. If you do, the app uploads your settings and the taste it has learned, without any photos, project names or file paths, to a Cloudflare R2 bucket that you own, using the keys you created in your own Cloudflare account. We have no access to that bucket, and Cloudflare's terms apply to it.

Model downloads

Small models come with the app. The larger models that power search, scoring and scene detection are not bundled with the installer. When you first use a feature that needs one, the app asks you and downloads it straight from the public place that publishes it: Hugging Face (huggingface.co) for most models, and the GitHub Releases of this project for a few converted files. We run no model server of our own. Like any download, the request reveals your IP address, the time and the name of the file to that service (Hugging Face or GitHub) and its hosting provider. The app does not send any photo, identifier or usage information with the request.

Update checks

A few seconds after it starts, and when you press “Check for updates”, the app asks the GitHub Releases of this project whether a newer version exists. You can turn the automatic check off in Settings › About. When you open the release notes, their pictures are loaded from GitHub (raw.githubusercontent.com). If you choose to install an update, the app downloads it from GitHub as well. These requests reveal your IP address, the time and the app version to GitHub. The app sends no photo, identifier or usage information with them.

System frameworks and permissions

On macOS the app may use Apple's on-device frameworks, such as Vision, for face and image analysis. On Android phones with a Qualcomm chip, the app may run models on the phone's neural processor through Qualcomm's runtime, which is included in the app. Both run locally.

On Android the app asks for access to your photos so it can analyse them, for notifications so it can show the progress of a long analysis, and to be exempt from battery optimisation so that analysis is not stopped halfway. It uses no Google Play services.

This website

This website is a static site. It sets no cookies and uses no analytics or advertising trackers. Your language and light or dark theme choice are stored in your browser's local storage so they are remembered, and never sent anywhere. Our hosting provider may keep standard server logs, such as IP address and requested page, for security and operation.

Your choices

In Settings you can clear the analysis cache, reset what the app has learned, delete downloaded models, clear the local copies of a NAS, and remove a NAS or a backup target together with its saved password or keys. Removing the app's data folder, or uninstalling the app on Android, deletes everything it has stored. You can also disable photo search and other model features, and the automatic update check, at any time.

Children

FleetingGleam is not directed at children, and because it collects no personal information it does not knowingly collect any from children.

Changes to this policy

If the app or this policy changes in a way that affects your privacy, we will update this page and the date above. Material changes will also be described in the release notes.

Contact

Questions about this policy or these terms, and requests about your data, can be raised in the Issues of the project's GitHub repository. Because the app collects nothing about you, there is no personal data held by us to access or delete. GitHub Issues