Your images
What happens to your image.
Only the bottom-right 256 × 256 pixel region is sent to our server. Your browser assembles the finished image.
Last updated: .
What is uploaded
When you select an image, the browser reads it locally. HEIC/HEIF files may first be converted on your device; this conversion does not upload the image. When you click Remove watermark, the browser extracts the bottom-right 256 × 256 pixel region. Only that region is uploaded for processing, without the original filename. For an image that is exactly 256 × 256 pixels, the processing region covers the entire image. Your browser combines the restored corner with the original and creates the download.
The corner may contain personal or sensitive details if they are present in that part of your image. Do not upload a region you do not want the service to process.
Processing files and deletion
Our private inference server temporarily stores the uploaded corner and its processed result to complete the job and return it to your browser. These files expire 10 minutes after processing completes or is cancelled and are removed by automatic cleanup. Service interruptions may delay deletion. We do not keep a long-term image library or account history for this tool.
The inference server keeps job metadata for up to 24 hours. The website keeps anonymous operational records, such as job identifier, model version and status, to diagnose failures. Those records become eligible for deletion after 30 days and are removed during hourly scheduled cleanup. Deletion can be delayed if cleanup fails. These records do not contain image files or request bodies.
Once you download an image, that copy is on your own device. Clearing or cancelling the page does not delete copies you have already downloaded.
Uploads are not training data
We do not use images uploaded to this tool to train our models. They are processed to fulfill your request and deleted according to the processing-file schedule above.
Service and usage data
Network providers receive ordinary connection information, including an IP address, when serving a request. Our abuse controls store a hashed IP identifier. These records expire no later than the next UTC day and are removed during hourly scheduled cleanup; deletion can be delayed if cleanup fails. Hashing reduces direct identification but does not make the identifier fully anonymous.
Usage measurement stores aggregate event counts, such as successful processing and downloads. It does not persist filenames, image content or location. Application request logs do not record uploaded request bodies.
Who operates the service
The AuraTuner team maintains this tool and operates its private model server. The website is hosted on Cloudflare infrastructure. Neon PostgreSQL stores the limited operational records described above, separately from AuraTuner’s database. These providers process the requests needed to deliver the website and connect it to the model service.
You do not need an account. If you email us, we receive your email address and anything you choose to include. Avoid attaching a sensitive image unless it is necessary to explain your request.
Questions and requests
Email support@auratuner.com for privacy questions or deletion requests. Include the approximate time of the request and the job identifier if one was shown. We may be unable to locate an anonymous request without that information, and processing files may already have expired.