Restoration field guide
JPEG vs PNG vs TIFF: Which Is Best for Scanning Old Photos?
Choose the right file format for an old photo scan, from a lossless preservation master to practical editing, restoration, and sharing copies.

By My Restored Photo editorial team
Published September 6, 2026 · Updated September 6, 2026 · 11 min read

On this page
- The quick decision
- First separate the master, working copy, and sharing copy
- What JPEG does well
- What PNG does well
- What TIFF does well and the compression setting people miss
- Bit depth, colour, and resolution are separate decisions
- A practical family-photo workflow
- If you already scanned everything as JPEG
- Preparing a copy for restoration
- Sources
For an important old photograph, save the first good scan as an uncompressed or losslessly compressed TIFF when your scanner offers it and storage is manageable. Keep that file untouched. Use a separate TIFF or PNG for lossless editing, then export a high-quality JPEG when you need a smaller file for sharing, printing, or an online service.
That is a workflow, not a verdict that one format is always "best." A careful JPEG can be a sensible master when TIFF is unavailable or the alternative is never finishing the project. PNG preserves pixel data without lossy compression, but it is not automatically smaller or more archival than a well-specified TIFF. Your capture settings, colour profile, metadata, backups, and whether you overwrite the original matter as much as the extension.
The quick decision
| Need | Usually choose | Why | Important caveat |
|---|---|---|---|
| Preservation master | TIFF with no compression or a stated lossless codec | Widely used for high-quality still-image masters and supports rich image data | A .tif file can use different compression schemes; confirm the setting |
| Lossless working or exchange copy | TIFF or PNG | Both can retain decoded pixel values without JPEG-style loss | Editing operations can still change pixels even when the file format is lossless |
| Small sharing, email, web, or lab copy | High-quality JPEG | Broad support and efficient storage for continuous-tone photographs | Common JPEG encoding is lossy; keep it separate from the master |
| Scanner offers only JPEG | Highest-quality JPEG | A completed, well-exposed scan is more useful than no scan | Preserve the scanner's first output and avoid repeated re-encoding |
The Federal Agencies Digital Guidelines Initiative's 2023 guidance says there is no single correct master format for every application and that every choice balances quality, access, and lifecycle management. It identifies TIFF and JPEG 2000 as master formats for prints and photographs and says best practice is for master files to be lossless.
Those are institutional digitization guidelines, not a rule that every family archive must reproduce a federal imaging program. Their useful principle is simpler: decide what the file is for before choosing its format.
First separate the master, working copy, and sharing copy
Many format arguments become easier once one scan is allowed to have more than one useful version:
- Master: the closest practical record of the scanner's output, kept neutral and never overwritten.
- Working copy: the file used for cropping, dust removal, colour correction, restoration, or layered edits.
- Derivative: a version made for a specific use, such as a JPEG for relatives, a print lab, a website, or an AI restoration upload.
The master is not expected to be convenient for everything. A large TIFF can stay in an archive while smaller JPEG derivatives travel easily. If a new edit or export goes wrong, you can return to the master without rescanning the physical print.
This also prevents an attractive restoration from replacing the historical source. AI tools can generate plausible texture or facial detail that the scan did not record. Preserve the untouched capture and label every restored version as a derivative; the guide to what AI photo restoration can really recover explains why that distinction matters.
What JPEG does well
JPEG is designed for continuous-tone images, so it is practical for photographs with gradual changes in colour and brightness. The Library of Congress description of the JPEG encoding family notes that the familiar baseline and progressive modes use lossy compression. The JPEG family also defines lossless modes, but they have not been widely supported in ordinary applications.
In everyday scanner and photo software, a .jpg generally means the common lossy form. The encoder represents the picture more compactly by approximating some image information. At a high-quality setting, the result may look excellent at normal viewing size and be far easier to store, upload, and share than a lossless scan.
The trade-off is that the file does not contain the exact pixel values that entered the encoder. The Library of Congress's personal archiving guidance explains that copying or simply opening a JPEG does not alter its image data, while editing and re-saving it can apply another round of lossy compression. Block patterns, ringing around sharp edges, and smeared fine texture become more likely as compression grows more aggressive.
Choose JPEG when:
- your scanner only provides JPEG and you can select a high-quality setting;
- you need a convenient access or sharing copy;
- an online service or print workflow prefers it; or
- storage constraints would otherwise stop you from digitizing the collection.
Do not treat an existing JPEG as ruined. Keep the first-generation file, edit a copy, and avoid repeatedly saving over it.
What PNG does well
PNG uses lossless compression. The current W3C PNG specification defines lossless storage for grayscale, indexed-colour, and truecolour raster images, with sample depths up to 16 bits and optional transparency.
That makes PNG useful when you need:
- a broadly readable lossless file;
- crisp edges, text, diagrams, or graphics mixed with an image;
- transparency; or
- a lossless working copy from software that does not offer a suitable TIFF.
For detailed photographic scans, PNG can require substantially more storage than a visually good JPEG because it is preserving rather than approximating the decoded pixels. Exact size depends on pixel dimensions, bit depth, image content, and encoder settings, so a format name alone cannot predict the megabytes.
PNG is a valid practical choice, especially when it is the scanner's only lossless option. It is not automatically the strongest preservation choice for every workflow. FADGI's table for prints and photographs lists TIFF and JPEG 2000 as master formats, while the current National Archives table accepts TIFF 6, JPEG 2000, and PNG 1.2 for digitized permanent photographic prints when the stated codecs are used. Those are institutional requirements that demonstrate several lossless choices, not a household mandate. Software can also discard or rewrite metadata during any conversion, regardless of whether the pixels remain lossless.
What TIFF does well and the compression setting people miss
TIFF is a flexible wrapper for raster image data rather than one fixed encoding. The Library of Congress TIFF 6.0 description explains that different TIFF encodings can use different compression and colour-representation schemes.
That flexibility is why TIFF is common in scanning and preservation workflows, but it creates an important trap: the .tif extension alone does not prove that a file is uncompressed or lossless. A TIFF may be:
- uncompressed;
- compressed with a lossless method such as LZW or Deflate/ZIP; or
- encoded with a lossy method, including JPEG compression in some TIFF workflows.
Check the scanner or export dialog. If the goal is an untouched master, choose uncompressed TIFF or a clearly identified lossless option. The US National Archives file-format tables list TIFF 6 with uncompressed or Deflate/ZIP encoding as one acceptable combination for digitized permanent photographic prints, illustrating why the format and codec need to be considered together.
Uncompressed TIFF is predictable but large. A losslessly compressed TIFF can reduce storage without changing the decoded pixels, although results vary with the image. Either can be a sound master when your software reads it reliably and you can maintain backups.
Bit depth, colour, and resolution are separate decisions
Choosing TIFF does not rescue a poor capture, and choosing JPEG does not automatically make a sharp scan soft. Format is only one part of the chain.
Check these settings independently:
- Pixel dimensions and scanning PPI. Capture enough real detail for the print and intended use. The guide to choosing a scan resolution gives practical 300 and 600 PPI examples.
- Bit depth. If your scanner can produce a genuine higher-bit-depth capture and you expect substantial tonal or colour correction, TIFF or PNG may retain that data. Confirm what the scanner actually outputs; a menu label is not proof.
- Colour profile. Embed or record the profile when the software supports it. A lossless file can still display incorrectly if colour information is missing or mishandled.
- Automatic processing. Turn off sharpening, scratch removal, aggressive noise reduction, and colour "fixes" for the neutral master when possible. Apply changes to a copy.
- Verification. Open the saved file, inspect important areas at 100%, and check that the photo edges, highlights, shadows, and colour were not clipped.
Changing a TIFF into PNG cannot restore information that was discarded earlier. Converting a JPEG to TIFF or PNG stops further JPEG encoding only if you work from that new copy; it does not reverse the first JPEG compression.

A practical family-photo workflow
- Scan one representative print first. Use a stable, ordinary photograph before committing a whole box to one setting.
- Choose the master format. Prefer uncompressed or lossless TIFF when available and manageable. Choose PNG when it is the practical lossless option. If only JPEG is available, select the highest sensible quality.
- Keep the capture neutral. Avoid baking restoration or strong enhancement into the only master.
- Inspect before batching. Check focus, cropping, colour, tonal clipping, and compression artefacts at 100%.
- Record the relationship. A simple pattern such as
1954-picnic-master.tif,1954-picnic-working.tif, and1954-picnic-share.jpgis clearer than names such asfinal-new-2.jpg. - Edit a copy. Never overwrite the only master with cropping, colour correction, dust removal, or AI reconstruction.
- Export for the destination. Use a high-quality JPEG for convenient sharing, or PNG when the destination accepts it and you want a lossless derivative.
- Back up the master. Format choice is not a backup. Keep more than one verified copy on separately managed storage.
For a large collection, test how much space a representative batch actually uses. The right compromise may be TIFF masters for fragile, damaged, small, or especially significant prints and carefully made JPEGs for routine snapshots. Apply the policy consistently enough that another family member can understand it.
If you already scanned everything as JPEG
Do not rescan automatically. First inspect the files at 100% and ask:
- Are the scans sharp enough for their intended use?
- Do gradients and edges show obvious compression artefacts?
- Are the pixel dimensions adequate?
- Are the physical originals safe and available?
- Would a new scan capture materially more information?
Rescan the most important or visibly compromised photographs when the original and equipment make a real improvement possible. Otherwise, preserve the existing JPEGs as first-generation masters, back them up, and make edits on copies. A technically ideal format is not worth losing names, dates, organization, or momentum across a family archive.
Preparing a copy for restoration
My Restored Photo's current home uploader accepts JPEG, PNG, and WebP files up to 10 MB, but not TIFF. Keep the TIFF master in your archive, export a separate JPEG or PNG copy, then restore one photograph and compare the result with the untouched scan before processing a batch.
Use the source with the most genuine detail, not merely the largest file. Converting a small JPEG to a large TIFF does not add historical evidence, and AI enlargement or repair may create plausible detail rather than recover what was truly present.
A practical default:
TIFF for a well-specified preservation master, TIFF or PNG for lossless work, and JPEG for convenient derivatives. Keep the untouched source separate from every edit.
Sources
- Federal Agencies Digital Guidelines Initiative: Technical Guidelines for Digitizing Cultural Heritage Materials, Third Edition
- US National Archives: Appendix A, Tables of File Formats
- Library of Congress: TIFF, Revision 6.0
- Library of Congress: JPEG Image Encoding Family
- Library of Congress: What People Are Asking About Personal Digital Archiving
- W3C: Portable Network Graphics (PNG) Specification, Third Edition
