Research report · updated 2026-08-04 · profile-level analysis
There is no universal passport-photo format. This report analyses 236 source-cited passport, visa and identity-document profiles across 67 countries and jurisdictions, using the open Anfas.Pro dataset.
82 of 128applicant-photo profiles with stored physical dimensions use 35 × 45 mm (64.1%).
15 sizesappear in that applicant-photo cohort. A familiar size should never be assumed without checking the exact document.
89.8% portrait115 applicant-photo physical contracts are taller than wide; 13 are square and 0 are landscape.
30 digital contractspublish pixel dimensions. Physical and digital groups can overlap, so this is not a separate-document total.
56 capture-ledprofiles use authority capture or a hybrid flow, while 139 ask for an applicant-provided photo and 41 do not specify the mode.
18 of 23non-ICAO-derived passport/visa pairs use the same stored primary size (78.3%); 5 differ.
The 35 × 45 mm format is the clear mode among profiles that explicitly require an applicant-provided photo, accounting for 82 of 128. The next most frequent formats are 51 × 51 mm (10), 30 × 40 mm (5), 33 × 48 mm (5). Those differences are operationally important: changing only the crop without checking the required aspect ratio can make an otherwise acceptable portrait fail.
The table counts one stored primary size per applicant-photo profile. It excludes authority-capture, hybrid and unspecified workflows, does not count optional accepted formats, and does not treat a country as one observation when that country issues several documents.
| Primary size | Profiles | Share of 128 |
|---|---|---|
| 35 × 45 mm | 82 | 64.1% |
| 51 × 51 mm | 10 | 7.8% |
| 30 × 40 mm | 5 | 3.9% |
| 33 × 48 mm | 5 | 3.9% |
| 40 × 60 mm | 4 | 3.1% |
| 50 × 60 mm | 4 | 3.1% |
| 50 × 70 mm | 4 | 3.1% |
| 26 × 32 mm | 3 | 2.3% |
| 20 × 30 mm | 2 | 1.6% |
| 36 × 47 mm | 2 | 1.6% |
| 40 × 40 mm | 2 | 1.6% |
| 40 × 50 mm | 2 | 1.6% |
| 24 × 30 mm | 1 | 0.8% |
| 25 × 35 mm | 1 | 0.8% |
| 50 × 50 mm | 1 | 0.8% |
For 23 jurisdictions, both the passport profile and generic visa profile have a stored primary physical size and neither record is ICAO-derived. 18 pairs match. 5 do not: Brazil, Canada, India, Serbia, Vietnam. This comparison describes registry dimensions, not whether the applicant supplies the photo; submission mode must still be checked separately.
| Jurisdiction | Passport | Visa |
|---|---|---|
| Brazil | 50 × 70 mm | 35 × 45 mm |
| Canada | 50 × 70 mm | 35 × 45 mm |
| India | 35 × 45 mm | 51 × 51 mm |
| Serbia | 50 × 50 mm | 35 × 45 mm |
| Vietnam | 51 × 51 mm | 40 × 60 mm |
“Visa” means the dataset’s generic visa profile. A named route such as a Schengen visa, eVisa or Diversity Visa can have a separate contract and must be checked independently.
The dataset records 139 applicant-photo profiles, 56 authority-capture or hybrid profiles, and 41 profiles where the mode is not stated. An authority-capture document may still publish portrait rules, but that does not necessarily mean an applicant should upload or print a photo.
This distinction matters for search intent. A page can accurately describe biometric portrait standards while the actual application workflow requires an in-person camera. The report therefore keeps the photo specification and the submission process as separate fields.
96 profiles have an exact official specification, 117 have official general guidance, 4 have partial official coverage, and 19 use an ICAO-derived baseline. These labels measure how precisely the public source describes the contract. They do not rank a country as more or less strict.
The 236 records contain 321 source links: 236 unique URLs across 154 hosts. 77 profiles cite more than one source. These are profile-source links, not a claim that every URL is a separate government publication.
We deliberately do not publish a “strictest countries” ranking. The dataset has no defensible strictness score, and several operational framing fields belong to the separate anfasProcessingPreset. Treating those implementation values as government rules would overstate the evidence.
Unit of analysis. One row is one document profile, not one person, application or country. Results are not population-weighted.
Dimensions. The headline size distribution uses only records where submissionMode is applicant-photo and photoContract.printSizeMm is present. That produces 128 observations from 224 stored physical dimensions. Accepted alternatives are excluded.
Paired comparison. Passport-versus-visa results include only jurisdictions with both generic profiles, known primary physical dimensions and no ICAO-derived record.
Missing values. A missing digital field or submission mode means the registry does not state it; it is not evidence that the authority has no such rule.
Authority boundary. Findings exclude anfasProcessingPreset. Every dataset record links to its public source and carries a verification level and last verification date.
Change over time. Issuing authorities can change requirements. The authority linked by each record remains the final source of truth.
The underlying data is available as JSON and CSV under CC BY 4.0. Cite the dataset version and landing page, state any transformations, and keep the distinction between source-backed fields and the Anfas.Pro processing preset.
Kravchenko, Y. (2026). ICAO and Government Document Photo Specifications (Version 1.0.0) [Data set]. Anfas.Pro. https://www.anfas.pro/datasets/icao-document-photo-specs/
JSON distribution · CSV distribution · Editorial policy · How we verify