How to Work With Image Metadata

Metadata is news these days: new data retention laws make a stab at defining what it is in the context of tracking phone and message traffic. The non-veracity of image or video data can sometimes be detected by inconsistencies in the metadata. Whilst digital manipulation can create images or videos of non-existent scenes for nefarious purposes, image and video metadata is so complex that efforts to hide the manipulation are frequently unsuccessful if they are attempted at all.

A concise definition of metadata as ‘data about data’ captures it pretty well. Metadata exists as a group of fields, each of which stores some aspect of the data. All digital images on any device exist as files – regions of binary storage that contain bytes describing a scene. Files can only be easily accessed via a filesystem, which provides a framework for storing and accessing files. Metadata can occupy any number of bytes – decoding it relies on applications distinguishing the bytes used for metadata from those used for describing the image scene.

Digital images contain a large number of metadata fields describing its attributes as a file. The most conspicuous field is the file name. For digital cameras, this field is defined automatically to be unique for each photograph, usually by including an ascending number in the file name. When files are transferred between devices (such as from a camera or mobile phone to a computer), metadata fields common to the source and destination filesystems are transferred. However, the fields describing the same data attribute may have different names in the source and destination fields, so even at this most basic level, the file metadata appearing on a different device from the one on which the image was taken will be different. Date metadata is particularly problematic – should the original Created date of a file be preserved when copying to another device or should it be set to the date of copying? There is no right answer here – the result is a matter of convention, and conventions differ between filesystems.

Names of metadata fields are not included in the data. Metadata format definitions typically specify names, indexes and data types for fields, but applications may use different names for fields which they display. Software components for accessing metadata used by applications may use different names from those specified in the standard and by the application. The variety of names for the same metadata field in different applications adds to the confusion.

For images, data about the image and the settings of the device used to record it form another domain of metadata. There is a standard (EXchangeable Image Format, or EXIF) for storing image and audio file metadata as follows:

• Date and time information. Digital cameras will record the current date and time and save this in the metadata. Mobile phone cameras usually include Global Positioning System (GPS) data from a built-in receiver including latitude and longitude in decimal degrees.
• Camera settings. This includes static information such as the camera model and make, and information that varies with each image such as orientation (rotation), aperture, shutter speed, focal length, metering mode, and ISO speed information.
• A thumbnail for previewing the picture on the camera's LCD screen, in file managers, or in photo manipulation software.
• Descriptions
• Copyright information

Nearly all phone and digital camera manufacturers support this standard for JPEG files which they create.

There are many other groups of metadata fields as well as EXIF which may be found in JPEG files.

The Wikipedia page on EXIF provides a lot more detail on this standard, its implementation, and its limitations. Many image processing programs (such as Adobe Photoshop) store a record of their use as a separate group in metadata when the file is saved in a format supporting EXIF metadata. However, image editing software may not always preserve all existing metadata when saving a file and it may add some metadata of its own.

Most of the time, you needn’t do so, unless you’re a professional photographer. Then you might be interested in adding a wide range of metadata in a robust way that’s always attached to the image, whatever platform it is stored on.

You may have an interest in setting date and location metadata on scanned images of old slides and photos so that they can be integrated into presentations including images from digital cameras. Tablets and phones work excellently at displaying images which they have recorded, using data recorded at image capture, but if you upload external images to them you may find that they appear out of time order, with incorrect times, or with incorrect or missing location data. One tablet will only display images in filename order using the native image viewer. If this happens, you’ll have to delve into setting image metadata, which involves a few technical hurdles.

There is no one answer to the question “What is the image date/time? ” in the metadata. The listing of metadata fields for a mobile phone image copied to a desktop contains 8 date/time fields that might be used as shown below:

5 different dates are present in these metadata fields. The GPS and EXIF dates differ only from the use of GMT rather than local time, but the range of filesystem dates reflect dates of copying and modification of the file.

Images from digital cameras (not phones) will seldom have GPS data included and scanned JPEG images will have not have EXIF or GPS data.

The choice of which field to use for the image date is up to the application designer. Filesystem date fields are always present but may be modified by processing. The image date may sometimes be conditional on the presence of particular fields or use the field has the earliest date. Most commonly, one of the EXIF fields is used. In many cases, the actual time is less important than the ordering of the images, but if you want to avoid showing photos of yourself in full daylight with a time of 3 am you may need to do some adjustment.

The location information presented as text in phone and tablet photo display apps is derived from the GPS latitude and longitude using a reverse geolocation web service, which is why the location text takes some time to appear after the photo is taken. After reverse geolocation, the location structure is stored in metadata for rapid access.

The free Windows desktop application GeoSetter can set location(Lat/Long and text fields) and date information, but the interface is somewhat daunting. It needs careful configuration as described here in order to reproduce the functionality of Photo Gallery. Locations are set by clicking on a map. Other geotagging applications (GPicSync, GeoTag Photos Pro) are specifically designed for geotagging born-digital photos by correlating dates with locations. It’s not clear if these would work efficiently with scanned images. The Google free geolocation service of providing a latitude/longitude from a location name no longer seems to be available. There are a number of web applications offering geotagging of images (eg GeoTagMyPhoto, GeoImgr) but not all of them appear to work.

When images are viewed on phones and tablets, only the image pixels are visible using native viewing/editing applications. The only metadata which can be displayed or used for ordering is date and location. The timeline (date ordered) view is particularly popular. Images can be grouped into albums, and text can be written on top of the image pixels. Information stored in other metadata (including the file name) that desktop users might expect to see, is not shown. If you want to provide text information (such as the names of people) in the image without writing over existing pixels, a review article here looks at some software you might use for this. Some software will automatically map text from existing metadata fields onto image pixels and provide facilities for setting dates. Ken Watson’s excellent website Digital Photos gives an overview of many aspects of dealing with digital images for archival purposes, where metadata is crucial.

Most image viewing software shows some metadata, most commonly the pixel dimensions of the image and a date, but no image viewing applications show all of it, probably because there are such a large number of different fields which may contain data. (See Exiftool section for an example of the 141 metadata fields stored in an iPhone image.) Applications may also choose their own names for metadata fields when displaying them. In Windows, some metadata fields can be read (and set) via the Details tab of file properties, accessed via a right-click on the file. However, the names shown do not correspond to metadata field names. Metadata may be processed before display – the Location data shown by iPhone Photos app is a text string derived from the latitude and longitude stored by the built-in GPS receiver. Many applications use metadata for their own purposes.

Microsoft Photo Gallery

Microsoft’s now-unsupported Photo Gallery, which was included in Windows Essentials, offered editing and storage of metadata for People, a Geotag, a Caption, and Descriptive Tags within image file metadata as listed below. It included face recognition to identify named faces in loaded photos in the same way as Google Photos. Although no longer supported as part of Windows Essentials (and not available for download from Microsoft), downloads are available from other sources and the code will run under Windows 10. Geo-tagging and facet recognition (to identify named people in other photos) are no longer available. Many fields accessible from Photo Gallery are stored in more than one metadata field as shown below:

For users who have processed images with Photo Gallery, there is a blog post on how to move on. Windows desktop application Caption Pro also provides inbuilt functionality for transferring Photo Gallery metadata to image pixels, as well as IPTC and XMP metadata added by professional photographic software.

ExifTool

Metadata handling is very much the domain of the enthusiast and anyone with an interest in the area owes a debt to Canadian physicist-photographer-programmer Phil Harvey, creator of the multi-platform ExifTool freeware, which can be used for viewing and editing image metadata. Under development for 15 years, ExifTool is a command-line utility that can list and edit a vast range of image metadata. An example of 141 metadata fields extracted from a mobile phone camera image by ExifTool is shown below. If you don’t know what most of them describe, you’re not alone.