Split out EXIF parser from JPEG

[bitplane]

Just leaving a note here for discussion / future work. As discussed in #2262 it would be useful to have the EXIF parsert split out.

Formats that support EXIF:

JPEG, TIFF, HEIF, WebP, AVIF, CR2, CR3, NEF, ARW, DNG, JP2

I've had a look through the internet archive and a few other places, and collected up all the EXIF specs I could find. I've dropped them here:

• exif repo

[cjee21]

JPEG, TIFF, HEIF, WebP, AVIF, CR2, CR3, NEF, ARW, DNG, JP2

Add PNG to that: #2276

[cjee21]

The documentation and source code of ExifTool can be useful too.

https://exiftool.org/TagNames/EXIF.html
https://github.com/exiftool/exiftool

[cjee21]

I have split out the EXIF parser (for JPEG) and added for PNG too. #2276

[cjee21]

I tried to do some parsing with this commit just to understand the structure of EXIF data. When really implementing it, we may end up with something like File_Tiff.cpp.

Running MediaInfo --trace_level=1 <jpg file> using my commit results in this output showing the basic structure:

0014 FFE1 - APP1 - Application-specific marker 1 - Exif (386 bytes)
0014  Header (4 bytes)
0014   Marker:                               65505 (0xFFE1)
0016   Fl - Frame header length:             384 (0x0180)
0018  Name:                                  Exif
001E  File Header (8 bytes)
001E   Alignment:                            II*
0022   FirstIFDOffset:                       8 (0x00000008)
0026  IFD (102 bytes)
0026   Header (2 bytes)
0026    NrOfDirectories:                     8 (0x0008)
0028   Directory entry (12 bytes)
0028    TagID:                               270 (0x010E)
002A    Data type:                           2 (0x0002) - ascii strings
002C    Number of components:                18 (0x00000012)
0030    Data or offset to data:              110 (0x0000006E)
0034   Directory entry (12 bytes)
0034    TagID:                               282 (0x011A)
0036    Data type:                           5 (0x0005) - unsigned rational
0038    Number of components:                1 (0x00000001)
003C    Data or offset to data:              128 (0x00000080)
0040   Directory entry (12 bytes)
0040    TagID:                               283 (0x011B)
0042    Data type:                           5 (0x0005) - unsigned rational
0044    Number of components:                1 (0x00000001)
0048    Data or offset to data:              136 (0x00000088)
004C   Directory entry (12 bytes)
004C    TagID:                               296 (0x0128)
004E    Data type:                           3 (0x0003) - unsigned short
0050    Number of components:                1 (0x00000001)
0054    Data or offset to data:              2 (0x00000002)
0058   Directory entry (12 bytes)
0058    TagID:                               305 (0x0131)
005A    Data type:                           2 (0x0002) - ascii strings
005C    Number of components:                11 (0x0000000B)
0060    Data or offset to data:              144 (0x00000090)
0064   Directory entry (12 bytes)
0064    TagID:                               306 (0x0132)
0066    Data type:                           2 (0x0002) - ascii strings
0068    Number of components:                20 (0x00000014)
006C    Data or offset to data:              156 (0x0000009C)
0070   Directory entry (12 bytes)
0070    TagID:                               34665 (0x8769)
0072    Data type:                           4 (0x0004) - unsigned long
0074    Number of components:                1 (0x00000001)
0078    Data or offset to data:              196 (0x000000C4)
007C   Directory entry (12 bytes)
007C    TagID:                               36867 (0x9003)
007E    Data type:                           2 (0x0002) - ascii strings
0080    Number of components:                20 (0x00000014)
0084    Data or offset to data:              176 (0x000000B0)
0088   NextIFDOffset:                        0 (0x00000000)

[cjee21]

Another resource: https://www.media.mit.edu/pia/Research/deepview/exif.html

[cjee21]

Updated PR #2276 with some codes copied from File_Tiff. Initial parsing is considered done. After this would be work to refine parsing and also handle more sub IFDs and records as well as map them to MediaInfo fields for display.

Current MediaTrace output for EXIF part in JPG:

MediaTrace

0014 FFE1 - APP1 - Application-specific marker 1 - Exif (386 bytes)
0014  Header (4 bytes)
0014   Marker:                               65505 (0xFFE1)
0016   Fl - Frame header length:             384 (0x0180)
0018  Name:                                  Exif
001E  File Header (8 bytes)
001E   Alignment:                            II*
0022   FirstIFDOffset:                       8 (0x00000008)
0026  IFD (102 bytes)
0026   Header (2 bytes)
0026    NrOfDirectories:                     8 (0x0008)
0028   Image Description (12 bytes)
0028    TagID:                               270 (0x010E) - Image Description
002A    Data type:                           2 (0x0002) - ASCII strings
002C    Number of components:                18 (0x00000012)
0030    IFDOffset:                           110 (0x0000006E)
0034   X Resolution (12 bytes)
0034    TagID:                               282 (0x011A) - X Resolution
0036    Data type:                           5 (0x0005) - unsigned rational
0038    Number of components:                1 (0x00000001)
003C    IFDOffset:                           128 (0x00000080)
0040   Y Resolution (12 bytes)
0040    TagID:                               283 (0x011B) - Y Resolution
0042    Data type:                           5 (0x0005) - unsigned rational
0044    Number of components:                1 (0x00000001)
0048    IFDOffset:                           136 (0x00000088)
004C   Resolution Unit - 2 (12 bytes)
004C    TagID:                               296 (0x0128) - Resolution Unit
004E    Data type:                           3 (0x0003) - unsigned short
0050    Number of components:                1 (0x00000001)
0054    Data:                                2 (0x0002)
0056    Padding:                             (2 bytes)
0058   Software (12 bytes)
0058    TagID:                               305 (0x0131) - Software
005A    Data type:                           2 (0x0002) - ASCII strings
005C    Number of components:                11 (0x0000000B)
0060    IFDOffset:                           144 (0x00000090)
0064   Modify Date (12 bytes)
0064    TagID:                               306 (0x0132) - Modify Date
0066    Data type:                           2 (0x0002) - ASCII strings
0068    Number of components:                20 (0x00000014)
006C    IFDOffset:                           156 (0x0000009C)
0070   Offset to Exif SubIFD (12 bytes)
0070    TagID:                               34665 (0x8769) - Offset to Exif SubIFD
0072    Data type:                           4 (0x0004) - unsigned long
0074    Number of components:                1 (0x00000001)
0078    ExifSubIFDOffset:                    196 (0x000000C4)
007C   DateTimeOriginal (12 bytes)
007C    TagID:                               36867 (0x9003) - DateTimeOriginal
007E    Data type:                           2 (0x0002) - ASCII strings
0080    Number of components:                20 (0x00000014)
0084    IFDOffset:                           176 (0x000000B0)
0088   IFDOffset:                            0 (0x00000000)
008C  Image Description - Created with GIMP (18 bytes)
008C   Data:                                 Created with GIMP
009E  X Resolution - 300.000 (8 bytes)
009E   Numerator:                            300 (0x0000012C)
00A2   Denominator:                          1 (0x00000001)
00A6  Y Resolution - 300.000 (8 bytes)
00A6   Numerator:                            300 (0x0000012C)
00AA   Denominator:                          1 (0x00000001)
00AE  Software - GIMP 3.0.2 (11 bytes)
00AE   Data:                                 GIMP 3.0.2
00BA  Modify Date - 2025:04:24 23:16:28 (20 bytes)
00BA   Data:                                 2025:04:24 23:16:28
00CE  DateTimeOriginal - 2025:04:24 23:16:13 (20 bytes)
00CE   Data:                                 2025:04:24 23:16:13
00E2  IFD (90 bytes)
00E2   Header (2 bytes)
00E2    NrOfDirectories:                     7 (0x0007)
00E4   DateTimeOriginal (12 bytes)
00E4    TagID:                               36867 (0x9003) - DateTimeOriginal
00E6    Data type:                           2 (0x0002) - ASCII strings
00E8    Number of components:                20 (0x00000014)
00EC    IFDOffset:                           286 (0x0000011E)
00F0   CreateDate (12 bytes)
00F0    TagID:                               36868 (0x9004) - CreateDate
00F2    Data type:                           2 (0x0002) - ASCII strings
00F4    Number of components:                20 (0x00000014)
00F8    IFDOffset:                           306 (0x00000132)
00FC   OffsetTime (12 bytes)
00FC    TagID:                               36880 (0x9010) - OffsetTime
00FE    Data type:                           2 (0x0002) - ASCII strings
0100    Number of components:                7 (0x00000007)
0104    IFDOffset:                           326 (0x00000146)
0108   OffsetTimeOriginal (12 bytes)
0108    TagID:                               36881 (0x9011) - OffsetTimeOriginal
010A    Data type:                           2 (0x0002) - ASCII strings
010C    Number of components:                7 (0x00000007)
0110    IFDOffset:                           334 (0x0000014E)
0114   OffsetTimeDigitized (12 bytes)
0114    TagID:                               36882 (0x9012) - OffsetTimeDigitized
0116    Data type:                           2 (0x0002) - ASCII strings
0118    Number of components:                7 (0x00000007)
011C    IFDOffset:                           342 (0x00000156)
0120   User Comment (12 bytes)
0120    TagID:                               37510 (0x9286) - User Comment
0122    Data type:                           7 (0x0007) - undefined
0124    Number of components:                25 (0x00000019)
0128    IFDOffset:                           350 (0x0000015E)
012C   Color Space - 1 (12 bytes)
012C    TagID:                               40961 (0xA001) - Color Space
012E    Data type:                           3 (0x0003) - unsigned short
0130    Number of components:                1 (0x00000001)
0134    Data:                                1 (0x0001) - sRGB
0136    Padding:                             (2 bytes)
0138   IFDOffset:                            0 (0x00000000)
013C  DateTimeOriginal - 2025:04:24 23:16:13 (20 bytes)
013C   Data:                                 2025:04:24 23:16:13
0150  CreateDate - 2025:04:24 23:16:13 (20 bytes)
0150   Data:                                 2025:04:24 23:16:13
0164  OffsetTime - +08:00 (7 bytes)
0164   Data:                                 +08:00
016C  OffsetTimeOriginal - +08:00 (7 bytes)
016C   Data:                                 +08:00
0174  OffsetTimeDigitized - +08:00 (7 bytes)
0174   Data:                                 +08:00
017C  User Comment (25 bytes)
017C   Data:                                 (25 bytes)

[cjee21]

I made some more progress on the parsing. Updated the previous comment above.

[cjee21]

The PR for initial Exif parsing is finally merged.

Potential next work:

• Implement Exif parsing for other formats (other than JPG, PNG and MP4-based) using the new parser
• Improve Exif data consistency (for example date/GPS formatting or better handle case of different but similar data available in other metadata)
• Map or display more Exif tags

It will be a good idea if we share what we are working on here to prevent doing duplicate work.

[JeromeMartinez]

Implement Exif parsing for other formats (other than JPG, PNG and MP4-based) using the new parser

At which format do you think? I see TIFF/DNG but it needs more than Exif parsing, I think I'll work on that soon.

Improve Exif data consistency (for example date/GPS formatting or better handle case of different but similar data available in other metadata)

I am on the GPS formatting right now.
Date is the next topic, I think I'll do it quickly.
@cjee21 Maybe you could check other fields?

Map or display more Exif tags

There is a lot to do there, @cjee21 motivated?

[cjee21]

At which format do you think?

The first post at the top here lists them.

There is a lot to do there, @cjee21 motivated?

I was thinking maybe @bitplane wants to help?

[cjee21]

@JeromeMartinez I have NEF files and they show FileExtension_Invalid : tiff tif. JPEGs from Nikon show Writing hardware : NIKON CORPORATION NIKON D6

[JeromeMartinez]

I have NEF files and they show

TIFF based files, also e.g. DNG, are not well supported for the moment, on my todo-list.

[JeromeMartinez]

NIKON CORPORATION NIKON D6

This is the kind of string I would like to "standardize", removing any "Corporation", company name uppercases etc, so "Nikon D6".
(dream...)

[cjee21]

This is the kind of string I would like to "standardize", removing any "Corporation", company name uppercases etc, so "Nikon D6". (dream...)

Randomly checked through photos I have and it seems Nikon consistently starts with NIKON CORPORATION NIKON D for various models.