Perceptual Image Diff is an image comparison software that utilizes a computerized version of the human visual system.
Perceptual Image Diff is an incredible image comparison tool that leverages a computational model of the human visual system to analyze and compare two images. This software is specifically designed for carrying out automated regression testing on images that are perceptually identical but have minor pixel noise from different random number generators or small changes to algorithms.
To get started with this software, users need to use the Cross Platform Make Utility and the libtiff library. Additionally, you will also need Subversion to obtain the source. You can follow these steps to compile pdiff:
1. Edit CMakeLists.txt as necessary to point to libtiff
2. Type cmake .
3. Type make .
4. pdiff should be compiled for your target system
Once you've compiled pdiff, you can start using it to analyze your images. The usage of this software is very straightforward. All you need to do is use the pdiff command followed by the two images you want to compare, along with optional parameters. Some of the parameters you can use include:
-verbose: Turns on verbose mode.
-fov deg: Can be used to set the field of view in degrees. Usually between 10.0 to 85.0.
-threshold p: Sets the number of pixels to reject. For example, if p is 100, then the test fails if 100 or more pixels are perceptibly different.
-gamma g: Sets the gamma to use to convert to RGB linear space. The default value is 2.2
-luminance l: Sets the luminance of the display the observer is seeing. The default value is 100 candela per meter squared.
In the latest release of this software, the FreeImage IO code has been patched in along with a -luminanceonly flag. Additionally, PNG, TIFF, and JPG formats are now supported for input and output, along with all the other formats supported by FreeImage.
Overall, if you're on the lookout for a powerful image comparison utility that is intuitive to use and backed by advanced technology, Perceptual Image Diff is an excellent option that won't disappoint.
To get started with this software, users need to use the Cross Platform Make Utility and the libtiff library. Additionally, you will also need Subversion to obtain the source. You can follow these steps to compile pdiff:
1. Edit CMakeLists.txt as necessary to point to libtiff
2. Type cmake .
3. Type make .
4. pdiff should be compiled for your target system
Once you've compiled pdiff, you can start using it to analyze your images. The usage of this software is very straightforward. All you need to do is use the pdiff command followed by the two images you want to compare, along with optional parameters. Some of the parameters you can use include:
-verbose: Turns on verbose mode.
-fov deg: Can be used to set the field of view in degrees. Usually between 10.0 to 85.0.
-threshold p: Sets the number of pixels to reject. For example, if p is 100, then the test fails if 100 or more pixels are perceptibly different.
-gamma g: Sets the gamma to use to convert to RGB linear space. The default value is 2.2
-luminance l: Sets the luminance of the display the observer is seeing. The default value is 100 candela per meter squared.
In the latest release of this software, the FreeImage IO code has been patched in along with a -luminanceonly flag. Additionally, PNG, TIFF, and JPG formats are now supported for input and output, along with all the other formats supported by FreeImage.
Overall, if you're on the lookout for a powerful image comparison utility that is intuitive to use and backed by advanced technology, Perceptual Image Diff is an excellent option that won't disappoint.
What's New
Version 1.0.2: N/A