XIFty for Node.js

Read metadata from images, video, and audio — locally, without shelling out, with stable normalized fields you can ship to product code.

npm install @xifty/xifty

const xifty = require("@xifty/xifty");
const result = xifty.extract("photo.jpg", { view: "normalized" });

const fields = Object.fromEntries(
  result.normalized.fields.map((f) => [f.field, f.value.value]),
);

fields["device.make"];      // "Apple"
fields["device.model"];     // "iPhone 15 Pro"
fields["captured_at"];      // "2026-04-21T14:33:08-04:00"
fields["location"];         // { latitude: 40.7922, longitude: -73.9584 }

What you get

XIFty extracts metadata from a wide range of media containers and surfaces it in four views — raw, interpreted, normalized, and report — so you can use clean stable fields in product code without losing access to the underlying values when provenance matters.

Formats supported today

| Container | Decoders | |---|---| | JPEG / TIFF / DNG | EXIF · XMP · ICC · IPTC | | PNG / WebP | EXIF · XMP · ICC · IPTC | | HEIF / HEIC | EXIF · XMP · ICC · IPTC · item-based dimensions | | MP4 / MOV | QuickTime · iTunes · ICC · XMP · Apple/Sony vendor metadata · DJI drone telemetry | | M4A / M4B / M4P | iTunes ilst (title, artist, album, year, genre, track number, cover art …) | | FLAC | Native stream info · Vorbis comments · embedded picture | | OGG (Vorbis / Opus) | Page parsing · ident headers · Vorbis comments | | AIFF / AIFC | Stream info from COMM chunk |

Normalized fields

The normalized view gives you cross-format stable keys so you don't have to know whether a given file was JPEG-EXIF or DJI-udta or Vorbis-comment:

• device.make, device.model, device.serial_number
• captured_at, created_at, modified_at
• location (latitude, longitude, altitude)
• dimensions.width, dimensions.height, orientation
• exposure.iso, exposure.aperture, exposure.shutter_speed, exposure.focal_length_mm
• lens.make, lens.model
• duration, video.framerate, video.bitrate, codec.video, codec.audio
• audio.channels, audio.sample_rate, audio.bit_depth
• drone.flight.{pitch,yaw,roll}_deg, drone.gimbal.{pitch,yaw,roll}_deg, drone.speed.{x,y,z}_mps
• author, copyright, headline, description, keywords
• color.space, color.profile.name, color.profile.class

When fields disagree across sources (e.g. EXIF vs. XMP), the report view tells you the conflict explicitly rather than silently picking one.

Try it

const { extract, probe } = require("@xifty/xifty");

// Lightweight format detection — no full extraction
probe("clip.mp4").input.detected_format;   // "mp4"

// Drone telemetry from a DJI MP4 — same API, same shape
const dji = extract("DJI_0003.MP4", { view: "normalized" });
const fields = Object.fromEntries(
  dji.normalized.fields.map((f) => [f.field, f.value.value])
);
fields["drone.gimbal.pitch_deg"];   // -31.2
fields["drone.flight.yaw_deg"];     // 175.5
fields["device.serial_number"];     // "53HQN4T0M5B7JW"

Common use cases

• Photo / video library ingestion (EXIF, XMP, GPS, capture time)
• Drone footage indexing (DJI flight + gimbal telemetry, GPS, model)
• AWS Lambda media-pipeline upload handlers
• Media-asset deduplication and search
• Audio library tagging (Vorbis comments across FLAC, OGG, Opus)
• Compliance audits (provenance via raw + report views)

More documentation

• API reference — every method, every option, every output field
• AWS Lambda — runtime targets, SAM example, layer assembly
• Supported platforms — what ships, what doesn't, why
• Local development — building from source, running tests
• Releasing — maintainer guide for cutting a new version

License

MIT. Built on the open-source XIFty core.