A 50 MB JPG test file

About this file

A 50 MB test file in JPG format, sized to exactly 52,428,800 bytes. Free to download with no signup, for verifying upload limits and file-handling behaviour.

The MD5, SHA-1 and SHA-256 values listed above are the real digests of this file. Recomputing them after download confirms the transfer was not corrupted, and they double as known-good expected values when testing a hashing library.

What this size means in practice

50 MB is the point where GitHub starts warning you on push (it hard-rejects above 100 MB). In practice it is the decision line for whether a binary belongs in Git directly or should move to Git LFS.

At this scale, an implementation that buffers the whole request in memory burns 50 MB or more per worker process — ten concurrent uploads means 500 MB. It is a good size for verifying that streaming or spill-to-temp-file actually works.

Getting a feel for this size

On a 10 Mbps link the transfer takes roughly 41.9 seconds, and about 4.2 seconds at 100 Mbps. Base64-encoded it grows to roughly 68,267 KB.

Estimating a photographic JPG at roughly 0.2 bytes per pixel, this size works out to around 262,144,000 pixels — about 18,696 × 14,022 px at a 4:3 aspect ratio. Compression varies enormously with image content, so treat this purely as an order-of-magnitude figure.

What to check for this format

JPEG is lossy and has no transparency. Its signature begins with FF D8 FF. The single most common defect in real implementations is ignoring the EXIF Orientation tag, which makes rotated photos appear sideways in the browser.

• Whether you read EXIF Orientation and auto-rotate, or deliberately strip it
• Whether privacy-sensitive data such as GPS coordinates remains — stripping is mandatory for user-submitted images
• Whether both progressive and baseline JPEG decode correctly — older decoders sometimes fail on progressive