Every file format, opened up

Thirty-four formats have a full reference page here, each with a working inspector that opens your own file in the browser: signatures, structure, limits, version history and primary sources. The directory lists every extension the converters handle, and the identifier below works out what a file really is from its bytes.

11Image

4Video

5Audio

7Document

1PDF

1Ebook

4Archive

1Font

. 34 pages

identify · reads the bytes, ignores the name

Not sure what a file really is?

Extensions are just part of the name. This reads the first and last 64 KB of the file in your browser and checks them against the signatures on the reference pages: magic numbers, ISO BMFF and RIFF brands, the file list inside ZIP containers (which tells DOCX from EPUB from JAR), TIFF tags that mark a DNG, and text patterns for CSV, XML, iCalendar, vCard, subtitles and email. It then sends you to the page that can open it up.

frame_inspectDrop any file, or click to chooseOnly its first and last 64 KB are read, and it all stays on your device.

or try a file built in this tab

holiday-photo.jpga PNG with the wrong namequarterly-report.zipa Word document renamedbackup-log.txtactually gzip-compressedmeeting.data calendar invite, no clue in the name

holiday-photo.jpgPNG imagesignature match · name is wrong

The 8-byte PNG signature: a high-bit byte, “PNG”, CR LF, Ctrl-Z, LF, designed to break if a transfer mangles line endings. First chunk IHDR: 120 × 80 px. The name says .jpg, so apps that trust extensions will open it with the wrong program or refuse it. Renaming to .png fixes that without changing a byte.

00000089·50P4EN47G0D·0A·1A·0A·00·00·00·0D·49I48H44D52R

00001000·00·00·78x00·00·00·50P08·06·00·00·00·D2·9B·B1·

00002089·00·00·01·94·49I44D41A54T78x01·EC·D3·3110A·02·

00003041A10·44DD1·C1·C3·99·7Az45E13·0D·BC·97·5E^63c85·

479 Bsize

.jpgextension in the name

image/jpegMIME your browser guessed from the name

image/pngMIME from the contents

Open it on the PNG pageChunk inspector with CRC-32, per-row filter strip, re-filter lab, optimised copy checked pixel for pixel.arrow_forward

34 of 34 pages

Reference pages with a working inspector

Each one opens your own file (or a sample built in the page) and shows its structure byte by byte, checks it against the specification, and lists MIME types, limits, history and primary sources.

AllImage 11Video 4Audio 5Document 7PDF 1Ebook 1Archive 4Font 1

formatfamilysignaturewhat the page does

.jpg.jpeg .jfif .jpeImageFF D8 FFMarker-by-marker inspector, every EXIF tag with privacy flags, IJG quality estimate, lossless metadata strip.arrow_forward .png.apngImage89 50 4E 47 0D 0A 1A 0AChunk inspector with CRC-32, per-row filter strip, re-filter lab, optimised copy checked pixel for pixel.arrow_forward .gifImage47 49 46 38 37|39 61 (GIF87a, GIF89a)Own LZW decoder, disposal and delay rules, step through every frame as stored and as shown.arrow_forward .webpImageRIFF …… WEBP (52 49 46 46 at 0, 57 45 42 50 at 8)RIFF and VP8/VP8L inspector, flag-vs-chunk checks, clean container rewrite.arrow_forward .heic.heif .hifImageftyp heic/heix/mif1 at byte 4HEIF box and item explorer: tile grids, HDR gain maps, Exif and GPS found through iloc.arrow_forward .avif.avifsImageftyp avif/avis at byte 4AV1 sequence header bit by bit, AVIF rule checks, in-browser AV1 encoder, colour-label lab.arrow_forward .svg.svgzImage<svg or <?xml (text)Scanner for scripts, external references and entity bombs, plus a 7-step optimiser with pixel diff.arrow_forward .tiff.tifImage49 49 2A 00 (II) or 4D 4D 00 2A (MM)IFD walker with strips and tiles; LZW, PackBits and Deflate written and read in the page.arrow_forward .bmp.dibImage42 4D (BM)Every header version, bitfield masks, RLE and alpha, decoded by the page’s own decoder.arrow_forward .ico.curImage00 00 01 00 (ICO), 00 00 02 00 (CUR)Icon directory with AND masks and PNG entries, plus an ICO/CUR builder.arrow_forward .dng.cr2 .cr3 .nef .arw .raf .orf .rw2 .pef .rawImageTIFF header + DNGVersion tag 50706Raw darkroom: decodes DNG sensor data and develops it step by step; previews from any camera raw.arrow_forward .mp4.m4v .3gpVideoftyp isom/mp42/avc1 at byte 4Stream-readiness inspector, RFC 6381 codec strings, remux to fragmented MP4 played through MSE.arrow_forward .mov.qtVideoftyp qt or moov/mdat/wide at byte 4Rotation matrix, edit lists, timecode and location metadata; rewrite without re-encoding.arrow_forward .mkv.webm .mka .mk3dVideo1A 45 DF A3 (EBML)EBML element tree, clusters and cues; a recorded WebM gets its missing duration fixed.arrow_forward .srt.vtt .sbvVideodigits, then “00:00:00,000 -->” (text)Cue timing, overlaps and reading speed; encoding and numbering repairs; WebVTT too.arrow_forward .mp3Audio49 44 33 (ID3) or FF FB/F3/F2Frame-by-frame bitrate, Xing/LAME tag, gapless delay and padding, three durations compared.arrow_forward .wav.bwf .rf64AudioRIFF …… WAVERIFF/RF64 chunks, Broadcast WAV fields, measured peak, DC and fake 24-bit detection.arrow_forward .flacAudio66 4C 61 43 (fLaC)Metadata blocks, frame headers with CRC-8/16, MD5 of the decoded audio.arrow_forward .ogg.opus .oga .ogv .spxAudio4F 67 67 53 (OggS)Ogg pages with CRC, granule positions and codec headers; damaged-stream repair.arrow_forward .aac.m4a .m4b .adtsAudioFF F1/F9 (ADTS) or ftyp M4AADTS and M4A structure; encoder delay measured by decoding the file again.arrow_forward .pdfPDF25 50 44 46 2D (%PDF-)Revision history: every incremental save, text diffs, what a signature does and doesn’t cover.arrow_forward .docx.docm .dotxDocumentZIP with [Content_Types].xml and word/Formatting cascade for every run, editing sessions from rsids, cleaner copy.arrow_forward .xlsx.xlsmDocumentZIP with [Content_Types].xml and xl/Hidden and very hidden sheets, stale cached values, digits Excel silently drops.arrow_forward .csv.tsvDocumentnone: delimited textStrict RFC 4180 reading vs spreadsheet auto-conversion, encodings, formula injection.arrow_forward .xml.xsd .xsl .rss .kml .gpx .plistDocument<?xml (optional) or BOMEncoding sniff, entity-bomb defusing, XPath namespace lab, Canonical XML hashing.arrow_forward .eml.msg .mimeDocumentnone: RFC 5322 header linesMIME part tree, encodings and attachment names, DKIM checked with Web Crypto, Received hops.arrow_forward .ics.ical .ifbDocumentBEGIN:VCALENDARTime zones, recurrence expansion, line folding; what each calendar does with the file.arrow_forward .vcf.vcardDocumentBEGIN:VCARDvCard 2.1/3.0/4.0 line inspector, quoted-printable and charsets, rewrite with round-trip check.arrow_forward .epub.kepubEbookZIP whose first entry is “mimetype” = application/epub+zipEvery reference resolved, package checks, repaired copy re-checked.arrow_forward .zip.zipxArchive50 4B 03 04 (PK)Local vs central directory compared, extra fields, CRC check, reproducible rebuild.arrow_forward .gz.gzip .tgzArchive1F 8BInstrumented inflater: every Deflate block, Huffman table, literal and match.arrow_forward .tarArchive“ustar” at byte 257ustar, GNU and pax headers, checksums, long names, links and padding.arrow_forward .7zArchive37 7A BC AF 27 1CHeader CRCs, folders and coder chains, the page’s own LZMA decoder, extraction with CRC check.arrow_forward .woff2.woff .ttf .otf .ttcFont77 4F 46 32 (wOF2)sfnt tables and checksums, live specimen with axes, a 5×7 pixel-font builder.arrow_forward

No reference page matches “”. The directory below may still list it.

Directory of 94 extensions

Filled chips open a reference page (several extensions share one, such as .webm on the MKV page). Outlined chips are formats the converters handle that don’t have a page of their own yet; hover for a one-line description.

...

schoolCore conceptsCodecs, containers, lossy vs lossless: the ideas every format page builds on.compareComparisonsHead-to-head pages such as JPG vs PNG, with measured results.checklistGuidesStep-by-step help for the most common conversion jobs.

Every file format, opened up

Not sure what a file really is?

Reference pages with a working inspector

Directory of 94 extensions

All format reference pages 147