A photo file has two distinct areas: the compressed image (your actual pixels) and the metadata (everything else). Raza Photos operates exclusively on the metadata area. The compressed image is never decoded, processed, or re-encoded during a metadata save.
This means there is no generation loss, no recompression, and no risk of corrupting the image data. A metadata save is a precise surgical operation on a small part of the file.
How saving works
Changes you make in Raza Photos are queued in memory as you work. They are not written to disk automatically. To save, choose Save Metadata from the Library menu on Mac. You can cancel at any time, and photos without pending changes are never touched.
- 1You make changes — assign a location, fix a date, tag a person
- 2Changes are held in the local database until you are ready
- 3You choose Save Metadata from the Library menu
- 4Raza Photos reads the existing metadata, merges your changes, and writes the updated file atomically to disk
- 5Only files with pending changes are modified — everything else is untouched
Supported file formats
Metadata write-back works with the most common image formats. One exception applies to AVIF.
What gets written
Raza Photos writes to multiple metadata standards simultaneously to ensure maximum compatibility with other photo apps and DAMs.
What is never changed
Raza Photos uses merge semantics — it reads all existing metadata first, then only updates the specific fields you changed. Everything else is preserved exactly as it was.
- Camera settings (aperture, shutter speed, ISO, focal length, lens info)
- Copyright and creator information
- Color profiles and ICC data
- File timestamps (creation and modification dates on disk)
- All metadata fields not managed by Raza Photos
- The compressed image data itself
Metadata writes use Apple's CoreGraphics API, which operates exclusively on the metadata sections of an image file. The compressed image data is streamed directly from the source file without being decoded.
The update process uses merge semantics: existing metadata is read in full, your changes are applied on top, and the result is assembled entirely in memory. A single atomic write then replaces the original file on disk — there is no intermediate temporary file that could leave a partial state if interrupted.
This approach means the write operation is fast, lossless, and safe regardless of file size.