Restoration vs. Alteration: How to Log Every Change to a Scanned Family Photo
Short answer. Photo restoration and photo alteration can produce similar-looking results, but they make different claims about the historical image. A responsible archive preserves an untouched scan, identifies every edited derivative, and records what changed, which tool or method was used, and why. A simple change log allows relatives and future researchers to distinguish surviving photographic evidence from digital reconstruction.
- Keep an untouched master scan before beginning any restoration.
- Distinguish reversible presentation adjustments from reconstructed image content.
- Record the purpose and extent of every meaningful change.
- Use filenames and metadata to identify original and derivative files.
- Maintain three copies of important digital collections across two storage types, with one copy off-site.
Define restoration and alteration before editing
Restoration generally aims to make an existing photograph easier to view while retaining its historical content. Typical tasks include correcting an overall color cast, adjusting exposure, reducing scanner dust, or improving contrast. These operations can change the appearance of an image, but they do not necessarily introduce new historical details. Their purpose should be documented, especially when the original photograph is faded, damaged, or difficult to interpret.
Alteration changes the image's content or presents reconstructed information as though it were directly recorded by the camera. Examples include generating missing facial features, replacing a background, removing a person, adding color to a monochrome photograph, or inventing details where the original emulsion has disappeared. Some alterations may be appropriate for personal display, but they should be identified clearly. A generated reconstruction is not equivalent to a surviving photographic detail.
- Exposure correction: a presentation adjustment.
- Dust removal: a localized cleanup operation.
- Colorization: an interpretive transformation.
- Generated facial detail: synthetic reconstruction.
- Removing a person or object: content alteration.
Preserve an untouched master before making changes
Begin with a faithful scan of the physical photograph. Record the physical source, scan date, dimensions, resolution, and file format. Save the master as a lossless image, such as TIFF, and make it read-only in the working archive when practical. Do not overwrite this file during editing. A separate working copy should be created for each restoration task, allowing the original digital representation to remain available for comparison.
A straightforward folder structure might contain Originals, Working, Restored, and Logs. For example, PHO-0087_master.tif remains in Originals, while PHO-0087_color-correction_v01.tif is stored in Restored. If an additional version uses AI-generated facial reconstruction, give it a separate descriptive filename. This prevents a heavily modified image from being mistaken for the source and makes it possible to return to an earlier version.
Use a change log that identifies the actual operation
A useful change log should answer five questions: which photograph was edited, which file was used as the input, what operation was performed, which tool or method was involved, and what the output changed. Add the date, operator or responsible archive contributor, software identification when known, and a short explanation of the reason for the edit. If a tool's version or processing settings are available, record them. If information is unavailable, leave it unspecified instead of guessing.
Describe changes in concrete language. "Improved image quality" is too vague to help a future reader. "Adjusted overall contrast and reduced a visible scanner dust mark in the upper-left corner" is more useful. If an AI tool generated missing background texture, say that reconstruction occurred and identify the affected area. The log should also state whether the output was reviewed against the master and whether any remaining uncertainty is known.
- Archive identifier and source filename.
- Input and output filenames.
- Operation and affected image area.
- Tool or method and available settings.
- Date, reason, and responsible contributor.
- Known limitations or remaining uncertainty.
Keep filenames and metadata consistent
A file naming convention provides a durable connection between an image and its editing history. Use a stable identifier that does not depend on a person's uncertain name or an estimated date. For example, PHO-0087_master.tif, PHO-0087_restored_v01.tif, and PHO-0087_reconstruction_v01.tif all refer to one archive item. Avoid vague names such as final, final-new, or best-version, which become confusing when additional edits are made.
Metadata for family photos should describe both the historical item and the digital file. Useful fields include identifier, approximate date, photographer when known, location, people depicted, physical format, source collection, scan resolution, and digitization date. Record uncertainty explicitly. For example, an estimated year should be identified as approximate, and an unidentified person should not be assigned a name simply because a relative thinks the identification is likely. Keep editing information in dedicated fields or an associated log.
Separate historical evidence from presentation copies
A restored photograph can be useful for family albums, exhibitions, and digital sharing, but its presentation should explain the nature of the work. Captions can identify a file as a color-corrected derivative, an AI-assisted restoration, or a colorized interpretation. If missing details were generated, state that the relevant area is reconstructed. This information matters when a photograph is later shared outside the family and viewed without access to its editing history.
For important photographs, retain a comparison copy that shows the untouched master alongside the derivative. Avoid presenting an edited image as proof of a detail that the original does not establish. This is especially important for facial features, uniforms, insignia, objects, and backgrounds that may influence historical interpretation. If a restoration makes a face appear clearer, that visual improvement should not automatically be treated as new evidence about the person's actual appearance.
Protect the archive with a 3-2-1 backup arrangement
The 3-2-1 backup approach provides a useful structure for protecting a photo library: maintain three copies of the data, store them on two different types of storage, and keep one copy off-site. For example, the working archive may reside on a computer, a second copy on an external drive, and a third copy in a separate location or an appropriate cloud storage service. The exact arrangement should reflect collection size, privacy requirements, and available resources.
Include original scans, derivatives, metadata, and change logs in the backup set. A backup that contains only the restored JPEGs cannot preserve the complete editing history. Check that files can actually be opened from backup storage, and periodically confirm that the latest log and metadata files are included. Encryption and access controls may be appropriate for sensitive family material. Begin by backing up one complete photograph project, including its master, derivatives, metadata record, and change log.
- Three copies of important archive data.
- Two different storage types or media.
- One copy in a separate location.
- Include masters, derivatives, metadata, and logs.
- Periodically confirm that backup files open correctly.
Questions readers ask
Should an AI-restored photograph replace the scanned original?
Keep both files. The restored version is a derivative intended for a particular purpose, while the untouched scan preserves the reference against which the restoration can be understood.
How should an AI-generated face be described?
Use explicit language such as AI-generated facial reconstruction. Identify the image area involved and avoid presenting the generated appearance as a verified likeness.
Where should restoration metadata be stored?
Store descriptive metadata in a catalog or supported metadata fields and keep a plain-text or spreadsheet change log alongside the archive. Maintaining both provides portability if software changes.
What happens when an editing application becomes unavailable?
The original image, exported derivative, and plain-language change log should remain understandable without the application. Preserve documented settings when available, but do not make an editable project file the only record.
Can family members disagree about a restoration?
Preserve the source and keep alternative derivatives separately identified. A family may choose different presentation versions without changing the historical record or erasing the reasoning behind earlier edits.