TIFF Format Deep Dive

Header Structure

TIFF is the grandfather of image metadata. When you see EXIF data in a JPEG, that's actually a TIFF structure embedded inside the file. Understanding TIFF means understanding how most image metadata works.

The format starts with an 8-byte header that tells you two critical things: the byte order (Intel vs Motorola) and where to find the first IFD. From there, it's all pointer-chasing.

TIFF File Structure

Header

IFD0

Strip Data

EXIF IFD

GPS IFD

Header

8 bytes

Byte order + magic + IFD offset

IFD0

200 bytes

Main image directory

Strip Data

Variable length

Image pixel data

EXIF IFD

500 bytes

EXIF sub-directory

GPS IFD

100 bytes

GPS sub-directory

TIFF Header

Byte Order (II = Little Endian)

Magic Number (42)

First IFD Offset

Offset	Hex								ASCII
0000	49492A0008000000	II*.....

Byte Order

49 49 (II)

Intel:Little-endian (least significant byte first)

4D 4D (MM)

Motorola:Big-endian (most significant byte first)

IFD Structure

IFD stands for Image File Directory, but think of it as a table of contents. It's a flat list of tags, each pointing to some piece of data. The clever part is that IFDs can point to other IFDs, creating a tree structure.

IFD Layout

2 bytesEntry count (number of tags)

12 × N bytesTag entries (12 bytes each)

4 bytesNext IFD offset (0 if none)

Tag Entry Structure (12 bytes)

Tag IDTypeCountValue/Offset

2 bytes2 bytes4 bytes4 bytes

If data fits in 4 bytes, it's stored directly in the value field. Otherwise, value field contains an offset to the data.

Tag Types

TIFF is strongly typed. Each tag specifies not just what it contains, but how to interpret the bytes. This is why you need to understand the type system to parse TIFF correctly.

ID	Name	Size	Description
1	BYTE	1	Unsigned 8-bit integer
2	ASCII	1	Null-terminated string (8-bit)
3	SHORT	2	Unsigned 16-bit integer
4	LONG	4	Unsigned 32-bit integer
5	RATIONAL	8	Two LONGs: numerator/denominator
6	SBYTE	1	Signed 8-bit integer
7	UNDEFINED	1	Raw bytes (binary data)
8	SSHORT	2	Signed 16-bit integer
9	SLONG	4	Signed 32-bit integer
10	SRATIONAL	8	Two SLONGs: signed num/denom
11	FLOAT	4	IEEE 32-bit float
12	DOUBLE	8	IEEE 64-bit double
13	IFD	4	Pointer to nested IFD

Value Storage Rule

If size × count ≤ 4 bytes, the value is stored directly in the 4-byte value field. Otherwise, the field contains an offset pointing to where the data is stored in the file.

IFD Offset Calculations

The hardest part of parsing TIFF is following the offset chain. Everything is relative to the start of the file, and you'll be jumping around a lot. Here's how the math works:

// IFD0 location
IFD0_offset = bytes[4:8] // From TIFF header

// Next IFD location (e.g., IFD1 for thumbnail)
next_IFD = IFD0_offset + 2 + (entry_count × 12) + 4

// SubIFD location (EXIF, GPS)
subIFD_offset = tag_value // From pointer tag (0x8769, 0x8825)

Important Tags

Image Data Tags (Preserved)

0x0100:ImageWidth

0x0101:ImageLength

0x0102:BitsPerSample

0x0103:Compression

0x0111:StripOffsets

0x0117:StripByteCounts

Metadata Tags (Removed)

0x010F:Make (camera manufacturer)

0x0110:Model (camera model)

0x0132:DateTime

0x013B:Artist

0x8298:Copyright

0x9003:DateTimeOriginal

Pointer Tags (Sub-IFDs)

0x8769:ExifIFDPointer

0x8825:GPSInfoIFDPointer

0xA005:InteroperabilityIFDPointer

SubIFDs

The main IFD (IFD0) often contains "pointer tags" that reference sub-directories. This is how EXIF, GPS, and other metadata gets organized into logical groups while keeping the main IFD clean.

EXIF SubIFD

Contains camera-specific metadata: exposure, aperture, ISO, focal length, flash, white balance, and more.

GPS SubIFD

Contains geolocation data: latitude, longitude, altitude, timestamp, satellites, speed, direction. Major privacy concern.

Maker Notes

Proprietary camera data. Format varies by manufacturer. May contain serial numbers, firmware version, custom settings.

How PicScrub Processes TIFF

Parse Header

Determine byte order (II/MM) and locate first IFD

Build Tag Whitelist

Identify essential tags for image rendering

Filter IFD Entries

Remove metadata tags, zero out SubIFD pointers

Rewrite File

Reconstruct TIFF with clean IFDs and copied image data

Offset Recalculation

TIFF files use file offsets extensively. When metadata is removed, PicScrub must recalculate all offsets to maintain file integrity.