Looking for the Provyn app? Go to provyn.network → ×

Why Metadata Fails &
pHash Wins

C2PA promised to solve image provenance. Instead, it created a false sense of security. Here's why perceptual hashing is the only approach that actually works.

Why Metadata-Based Verification Fails

The Content Authenticity Initiative (C2PA) embeds cryptographic signatures directly into image files. It's a solid idea — in theory. In practice, the moment you share your image, the verification breaks.

BROKEN

Metadata gets stripped on upload

Every major platform — Instagram, Twitter/X, TikTok, Facebook, Threads — removes all metadata during upload. Your C2PA credentials vanish before anyone sees them.

BROKEN

Reprocessing destroys signatures

Platforms resize, compress, and re-encode images. Cryptographic signatures that depend on exact byte content break after any modification.

BROKEN

Requires platform cooperation

C2PA only works if platforms choose to read and preserve the metadata. They don't. There's no enforcement, no incentive, no consequence.

BROKEN

No backward compatibility

Existing photos can't be retroactively certified. You can only verify images that were C2PA-signed from the moment of creation — excluding millions of existing photographs.

Who strips C2PA/EXIF data?

Every major platform removes metadata during image processing:

Instagram Facebook Twitter / X TikTok Threads Snapchat WhatsApp Discord Slack iMessage

How Perceptual Hashing Works

Perceptual hashing (pHash) creates a fingerprint based on what an image looks like, not its exact byte content. This fundamental difference is why it survives transformations that destroy metadata-based approaches.

📷
Upload Image
🎨
Resize to 9x8
🧬
Grayscale + Compare
🔒
64-bit Hash
Result: dHash fingerprint (64-bit) a4f2e8c1d7b30956
SURVIVES

Compression

JPEG re-encoding slightly changes byte values but preserves visual structure. pHash remains stable because it measures relative brightness, not absolute values.

SURVIVES

Resizing

Downscaling to thumbnail doesn't erase the perceptual fingerprint — the gradient directions that pHash captures are preserved at any resolution.

SURVIVES

Screenshots

Screenshot of your image? The perceptual fingerprint survives. It's still recognizably derived from your original — which is exactly what you want to prove.

SURVIVES

Brightness/Contrast

Hash compares adjacent pixel relationships, not absolute values. A brighter or darker version produces the same (or very similar) fingerprint.

pHash vs C2PA: Side by Side

Here's how the two approaches compare across the criteria that actually matter for photographers.

Capability Provyn (pHash) C2PA (Metadata)
Survives metadata stripping
Works after Instagram/Twitter upload
✓ Yes ✗ No
Survives image compression
Works after JPEG re-encoding
✓ Yes ✗ No
Survives resizing
Works on thumbnails
✓ Yes ✗ No
Works on existing photos
No camera hardware required
✓ Yes ✗ No
Platform-agnostic
No platform cooperation needed
✓ Yes ✗ No
Detects similar images
Finds modified copies
✓ Yes ✗ No
Backward compatible
Works with old images
✓ Yes ✗ No

pHash Fingerprint + Blockchain Certificate

Provyn combines perceptual hashing with blockchain timestamping. The hash survives everything. The timestamp proves when.

Step 01
📷

Generate pHash

Upload any image. We compute a perceptual hash that captures visual structure, not file bytes.

Step 02
📜

Anchor to Blockchain

Your pHash is committed to a public blockchain with a timestamp. Immutable. Public. Verifiable.

Step 03

Verify Any Time

Anyone can check if an image matches your registered fingerprint — no matter where it's been.

Ready to protect your work?

Try the interactive demo to see pHash in action, or join the waitlist for early access.

Try the Demo