Plugins to allow QImage to support
extra file formats.
This framework provides additional image format plugins for QtGui. As such it is not required for the compilation of any other software, but may be a runtime requirement for Qt-based software to support certain image formats.
The following image formats have read-only support:
- Animated Windows cursors (ani)
- Camera RAW images (arw, cr2, cr3, dcs, dng, ...)
- Gimp (xcf)
- Krita (kra)
- OpenRaster (ora)
- Pixar raster (pxr)
- Portable FloatMap/HalfMap (pfm, phm)
- Photoshop documents (psd, psb, pdd, psdt)
- Radiance HDR (hdr)
- Scitex CT (sct)
- Sun Raster (im1, im8, im24, im32, ras, sun)
The following image formats have read and write support:
- AV1 Image File Format (avif)
- DirectDraw Surface (dds)
- Encapsulated PostScript (eps)
- High Efficiency Image File Format (heif)
- JPEG 2000 (jp2, j2k, jpf)
- JPEG XL (jxl)
- JPEG XR (jxr)
- OpenEXR (exr)
- Personal Computer Exchange (pcx)
- Quite OK Image format (qoi)
- SGI images (rgb, rgba, sgi, bw)
- Softimage PIC (pic)
- Targa (tga): supports more formats than Qt's version
See the QImageIOPlugin
documentation for information on how to write a new plugin.
The main difference between this framework and the image formats of Qt is the license. As such, if you write an image format plugin and you are willing to sign the Qt Project contributor agreement, it may be better to submit the plugin directly to the Qt Project.
To be accepted, contributions must:
- Contain the test images needed to verify that the changes work correctly
- Pass the tests successfully
The TGA plugin supports more formats than Qt's own TGA plugin; specifically, the one provided here supports indexed, greyscale and RLE images (types 1-3 and 9-11), while Qt's plugin only supports type 2 (RGB) files.
The code for this cannot be contributed upstream directly because of licensing. If anyone were willing to write fresh code to improve Qt's TGA plugin, it would allow the TGA plugin in this framework to be removed.
The DDS plugin is a fork from Qt 5.6 with bug fixes and improvements.
The plugin was forked because Qt Project no longer supports its DDS plugin.
The JP2 plugin is based on the popular and wide used OpenJPEG library.
The Qt project has a no longer supported JPEG 2000 plugin based on Jasper.
This framework is licensed under the LGPLv2.1.
The CMake code in this framework is licensed under the BSD license.
The current implementation of a plugin may not be complete or may have limitations of various kinds. Typically the limitations are on maximum size and color depth.
The various plugins are also limited by the formats natively supported by Qt. For example, native support for CMYK images is only available since Qt 6.8.
HDR images are supported via floating point image formats from DDS, EXR, HDR, JXL, JXR, PFM and PSD plugins. It is important to note that in the past these plugins stripped away HDR information, returning SDR images.
HDR images return R, G and B values outside the range 0.0 - 1.0. While Qt painters handles HDR data correctly, some older programs may display strange artifacts if they do not use a tone mapping operator (or at least a clamp). This is not a plugin issue.
Metadata support is available in formats that include it via
QImage::setText() and QImage::text(). To ensure consistent metadata
functionality, the following keys have been adopted.
About the image:
Altitude: Floating-point number indicating the GPS altitude in meters above sea level (e.g. 35.4).Author: Person who created the image.Comment: Additional image information in human-readable form, for example a verbal description of the image.Copyright: Copyright notice of the person or organization that claims the copyright to the image.CreationDate: Creation date and time in ISO 8601 format without milliseconds (e.g. 2024-03-23T15:30:43).Description: A string that describes the subject of the image.DocumentName: The name of the document from which this image was scanned.HostComputer: The computer and/or operating system in use at the time of image creation.Latitude: Floating-point number indicating the latitude in degrees north of the equator (e.g. 27.717).Longitude: Floating-point number indicating the longitude in degrees east of Greenwich (e.g. 85.317).Owner: Name of the owner of the image.Software: Name and version number of the software package(s) used to create the image.Title: The title of the image.
About the camera:
Manufacturer: The manufacturer of the recording equipment.Model: The model name or model number of the recording equipment.SerialNumber: The serial number of the recording equipment.
About the lens:
LensManufacturer: The manufacturer of the interchangeable lens that was used.LensModel: The model name or model number of the lens that was used.LensSerialNumber: The serial number of the interchangeable lens that was used.
Complex metadata (requires a parser):
XML:org.gimp.xml: XML metadata generated by GIMP and present only in XCF files.XML:com.adobe.xmp: Extensible Metadata Platform (XMP) is the metadata standard used by Adobe applications and is supported by all common image formats. Note that XMP metadata is read and written by plugins as is. Since it may contain information present in other metadata (e.g.Description), it is the user's responsibility to ensure consistency between all metadata and XMP metadata when writing an image.
Supported metadata may vary from one plugin to another. Please note that only the most common metadata are supported and some plugins may return keys not listed here.
EXIF (Exchangeable Image File Format) metadata is a standard for embedding information within the image file itself.
Unlike the metadata described above, EXIF metadata is used internally by some plugins to standardize image handling. For example, the JXL plugin uses them to set/get the image resolution and metadata. They are also needed to make the image properties appear in the file details of some file managers (e.g. Dolphin).
When reading, EXIF metadata is converted into simple metadata (e.g.
Description) and inserted into the image if and only if it is not already
present.
On writing, the image metadata is converted to EXIF and saved appropriately. Note that, if not present in the image to be saved, the following metadata are created automatically:
Software: Created usingapplicationNameandapplicationVersionmethods ofQCoreApplication.CreationDate: Set to current time and date.
ICC profile support is implemented in all formats that handle them using
QColorSpace. When saving, some
plugins convert the image using color profiles according to format
specifications. In particular, HDR formats almost always convert to linear
RGB.
Where possible, plugins support large images. By convention, many of the
large image plugins are limited to a maximum of 300,000 x 300,000 pixels.
Anyway, all plugins are also limited by the
QImageIOReader::allocationLimit(). Below are the maximum sizes for each
plugin ('n/a' means no limit, i.e. the limit depends on the format encoding).
- ANI: n/a
- AVIF: 32,768 x 32,768 pixels, in any case no larger than 256 megapixels
- DDS: n/a
- EXR: 300,000 x 300,000 pixels
- EPS: n/a
- HDR: n/a (large image)
- HEIF: n/a
- JP2: 300,000 x 300,000 pixels
- JXL: 262,144 x 262,144 pixels, in any case no larger than 256 megapixels
- JXR: n/a, in any case no larger than 4 GB
- KRA: same size as Qt's PNG plugin
- ORA: same size as Qt's PNG plugin
- PCX: 65,535 x 65,535 pixels
- PFM: n/a (large image)
- PIC: 65,535 x 65,535 pixels
- PSD: 300,000 x 300,000 pixels
- PXR: 65,535 x 65,535 pixels
- QOI: 300,000 x 300,000 pixels
- RAS: n/a (large image)
- RAW: n/a (depends on the RAW format loaded)
- RGB: 65,535 x 65,535 pixels
- SCT: 300,000 x 300,000 pixels
- TGA: 65,535 x 65,535 pixels
- XCF: 300,000 x 300,000 pixels
All plugins work fine on random access devices while only some work on sequential access devices. Some plugins, such as PSD, allow reading RGB images on sequential access devices, but cannot do the same for Lab files.
Important: some plugins use QIODevice transactions and/or
QIODevice::ungetChar(). Therefore, the device used to read the image must not
have any active transactions.
Qt has added many image formats over time. In older plugins, to support new
formats, QImage conversion functions have been used, causing memory
consumption proportional to the size of the image to be saved.
Normally this is not a source of problems because the affected plugins
are limited to maximum images of 2GiB or less.
On plugins for formats that support large images, progressive conversion has been used or the maximum size of the image that can be saved has been limited.
PSD plugin loads CMYK, Lab and Multichannel images and converts them to RGB without using the ICC profile.
JP2, JXR, PSD and SCT plugins natively support 4-channel CMYK images when compiled with Qt 6.8+.
This plugin can be disabled by setting KIMAGEFORMATS_DDS to OFF
in your cmake options.
The following defines can be defined in cmake to modify the behavior of the plugin:
DDS_DISABLE_STRIDE_ALIGNMENT: disable the stride aligment based on DDS pitch: it is known that some writers do not set it correctly.
This plugin is disabled by default. It can be enabled by settings
KIMAGEFORMATS_HEIF to ON in your cmake options.
The plugin is disabled due to issues with the heif library on certain distributions. In particular, it is necessary that the HEIF library has support for HEVC codec. If HEVC codec is not available the plugin will compile but will fail the tests.
The following defines can be defined in cmake to modify the behavior of the plugin:
EXR_CONVERT_TO_SRGB: the linear data is converted to sRGB on read to accommodate programs that do not support color profiles.EXR_DISABLE_XMP_ATTRIBUTE: disables the stores XMP values in a non-standard attribute named "xmp". Note that Gimp reads the "xmp" attribute and Darktable writes it as well.
The plugin uses Ghostscript to convert the raster image. When reading it
converts the EPS to PPM and uses the Qt PPM plugin to read the image.
When writing it uses QPrinter to
create a temporary PDF file which is then converted to EPS. Therefore, if
Ghostscript is not installed, the plugin will not work.
The following defines can be defined in cmake to modify the behavior of the plugin:
HDR_HALF_QUALITY: on read, a 16-bit float image is returned instead of a 32-bit float one.
This plugin can be disabled by setting KIMAGEFORMATS_JP2 to OFF
in your cmake options.
JP2 plugin has the following limitations due to the lack of support by OpenJPEG:
- Metadata are not supported.
- Image resolution is not supported.
The current version of the plugin limits the image size to 256 megapixels according to feature level 5 of the JXL stream encoding.
The following defines can be defined in cmake to modify the behavior of the plugin:
JXL_HDR_PRESERVATION_DISABLED: disable floating point images (both read and write) by converting them to UINT16 images. Any HDR data is lost. Note that FP images are always disabled when compiling with libJXL less than v0.9.JXL_DECODE_BOXES_DISABLED: disable reading of metadata (e.g. XMP).
This plugin is disabled by default. It can be enabled by settings
KIMAGEFORMATS_JXR to ON in your cmake options.
The following defines can be defined in cmake to modify the behavior of the plugin:
JXR_DENY_FLOAT_IMAGE: disables the use of float images and consequently any HDR data will be lost.JXR_DISABLE_DEPTH_CONVERSION: remove the neeeds of additional memory by disabling the conversion between different color depths (e.g. RGBA64bpp to RGBA32bpp) at the cost of reduced compatibility.JXR_DISABLE_BGRA_HACK: Windows displays and opens JXR files correctly out of the box. Unfortunately it doesn't seem to open (P)RGBA @32bpp files as it only wants (P)BGRA32bpp files (a format not supported by Qt). Only for this format an hack is activated to guarantee total compatibility of the plugin with Windows.JXR_ENABLE_ADVANCED_METADATA: enable metadata support (e.g. XMP). Some distributions use an incomplete JXR library that does not allow reading metadata, causing compilation errors.
The KRA format is a ZIP archive containing image data. In particular, the rendered image in PNG format is saved in the root: the plugin reads this image.
The ORA format is a ZIP archive containing image data. In particular, the rendered image in PNG format is saved in the root: the plugin reads this image.
PSD support has the following limitations:
- Only images saved by Photoshop using compatibility mode enabled (Photoshop default) can be decoded.
- Multichannel images are treated as CMYK if they have 2 or more channels.
- Multichannel images are treated as Grayscale if they have 1 channel.
- Duotone images are treated as grayscale images.
- Extra channels other than alpha are discarded.
The following defines can be defined in cmake to modify the behavior of the plugin:
PSD_FAST_LAB_CONVERSION: the LAB image is converted to linear sRGB instead of sRGB which significantly increases performance.PSD_NATIVE_CMYK_SUPPORT_DISABLED: disable native support for CMYK images when compiled with Qt 6.8+
Loading RAW images always requires a conversion. To allow the user to choose how to convert the image, it was chosen to use the quality parameter to act on the converter. The quality parameter can be used with values from 0 to 100 (0 = fast, 100 = maximum quality) or by setting flags to selectively change the conversion (see also raw_p.h).
The default setting tries to balance quality and conversion speed.
XCF support has the following limitations:
- XCF format up to version 12 (no support for GIMP 3).
- The returned image is always 8-bit.
- Cannot read zlib compressed files.
- The rendered image may be slightly different (colors/transparencies) than in GIMP.