Help Center / Supported Formats & Languages

Supported Formats & Languages

Detailed information about file formats, document types, languages, and expected accuracy

Supported File Formats

JPG / JPEG

Most common format

Ideal for scanned photographs of documents. Widely supported and efficient for storage.

PNG

Lossless quality

Best for preserving image quality without compression artifacts. Larger file sizes.

TIFF

Archival quality

Professional archival format. Excellent for high-quality scans from archives and libraries.

WebP

Modern format

Modern image format with excellent compression. Good balance of quality and file size.

File Requirements

  • Maximum file size: 20 MB per image
  • Recommended resolution: 300 DPI or higher for best results
  • Color mode: RGB or Grayscale (color recommended for faded documents)

Language & Script Support

Our AI has been trained and tested on historical documents in various languages. Below is our honest assessment of accuracy levels based on actual processing results.

Tier 1: Extensively Tested

Highest Confidence

These languages have been thoroughly tested with thousands of documents. Expect reliable results on well-preserved scans.

Polish

Latin cursive script

90-95%

Best for 19th-20th century civil records. Extensive testing on Masovia region documents.

5,000+ records processed

Latin

Church records

85-92%

Strong support for Catholic parish registers. Familiar with common abbreviations.

3,000+ records processed

Russian

Cyrillic script

85-90%

Good support for Russian Empire civil records (1850-1920). Handles Cyrillic cursive well.

2,500+ records processed

German

Gothic / Kurrent script

80-90%

Good support for German Gothic script. Accuracy varies more with individual handwriting styles.

2,000+ records processed

Tier 2: Good Support

Reliable

Good results on typical documents. Some challenging scripts or unusual handwriting may reduce accuracy.

Ukrainian

80-88%

Cyrillic script, similar to Russian support

French

82-90%

Latin cursive, civil records

Italian

80-88%

Latin cursive

Spanish

80-88%

Latin cursive

Portuguese

78-86%

Latin cursive

Czech / Slovak

78-86%

Latin cursive with diacritics

Tier 3: Experimental / Beta

Variable Results

Important: These languages have limited testing. Results may be inconsistent. We recommend starting with a small batch to evaluate quality before processing large collections.

Hebrew Script

  • Hebrew
  • Yiddish

60-80% accuracy, highly variable

Other Cyrillic

  • Bulgarian
  • Serbian

70-85% accuracy

Other Scripts

  • Greek
  • Armenian

Limited testing, results vary

Known Limitations

Accuracy can be significantly reduced by the following factors:

Document Condition

  • Faded or water-damaged ink
  • Torn or missing sections
  • Heavy foxing or staining
  • Bleed-through from reverse side
  • Low resolution scans (<200 DPI)

Handwriting Challenges

  • Highly inconsistent handwriting
  • Unusual abbreviations or shorthand
  • Decorative or elaborate scripts
  • Very small or cramped writing
  • Multiple overlapping hands

What To Do

If you receive low-confidence results, you can submit corrections and we'll refine our understanding. For particularly challenging documents, consider our priority tier which includes manual review options. See our Quality Guarantees page for refund policies.

Supported Document Types

👶

Birth Records

Birth certificates, baptismal records, naming ceremonies

💒

Marriage Records

Marriage certificates, banns, wedding records

✝

Death Records

Death certificates, burial records, obituaries

⛪

Church Records

Parish registers, confirmation records, church books

📜

Civil Registration

Government vital records, civil certificates

📋

Other Historical Documents

Census records, wills, property documents

Tips for Best Results

Use high-resolution scans - 300 DPI or higher produces the best results. 600 DPI is ideal for faded documents.

Ensure good lighting - Avoid shadows and uneven illumination during scanning or photography.

Keep documents flat - Minimize curves, folds, and wrinkles when scanning bound volumes.

Include the full page - Don't crop too close to the text. Margins provide helpful context.

Color vs grayscale - Color scans often work better for faded or stained documents.

Test first - For new document types, try a few samples before processing large batches.

Ready to try our service?

Try Interactive Demo Get 5 Free Credits