MyHeritage Photo Enhancer

• MyHeritage warns that AI-simulated facial detail may be inaccurate. The same caution applies when any enhancer invents pixels in an old scan.
• For an archival scan, “improvement” includes preserving the record. MyHeritage’s face-oriented path may appeal for group portraits; WaveSpeed’s general enhancer may suit enlarging the whole scan.

MyHeritage’s help article says its Photo Enhancer enlarges small faces, keeps the original photo intact, and marks enhanced versions with a magic-wand symbol. It explicitly warns that simulated facial detail can be inaccurate. WaveSpeed’s browser enhancer enlarges the whole image at 2×, 3× or 4×. These are meaningful workflow differences, not evidence of which result is more historically faithful.

For a family group scan, first name the original file with its scan date and keep it read-only. In MyHeritage, inspect each detected face individually. In WaveSpeed, inspect the whole frame, especially handwriting and clothing patterns. Save derived files with a suffix such as -enhanced, and keep a short note with four fields: original filename, tool used, date processed, and uncertain details. If a newly clear badge number or handwritten name cannot be read in the source, mark it as an inference rather than transcribing it into the archive.

Choose the face-oriented route when individual relatives are the focus. Choose whole-frame enlargement when the document border, labels and background matter equally. Confirm account and download conditions in the product before processing a collection.

• WaveSpeed’s browser enhancer is free and needs no account. MyHeritage’s subscription guide says Photo Enhancer is included in the Photo subscription; Omni includes all Photo features, while Complete includes photo features but not Photo Scanner. Confirm the plan linked to your family site before a large archive job.
• No complete same-workload price comparison is established; choose based on the exact archive task and current access terms.

• Scan at the best available source quality first. Then test 2× on the whole image, inspect each face, handwriting and damaged border, and only increase scale if it helps the intended display.
• At the intended display size and at 100% zoom, look for invented facial features, changed handwriting and repeated clothing texture. Keep uncertain details out of archive captions.

• For many scans, define a consistent review and recording process before evaluating any repeat model route. Upscaling alone will not mend scratches, tears or missing content.

Next step: Preserve the scan, then make a derivative with the free browser enhancer and inspect every face and word.