v1.2.0 - ?

[Python] Added option "--config-dir" to specify path to display_models.json and other config files
[Python] When passed "--display ?", a list of available display models will be printed.
[Python] Added PQ EOTF, fvvdp_display_photo_gog replaced with fvvdp_display_photo_eotf.
[Python] sRGB EOTF used by default (instead of gamma). The predictions may differ slightly.
[Python] Added --full-screen-resize to enlarge videos to the full screen resolution.
[Python] Now GPU 0 is used by default (if available).
[Python] Added --metrics parameter, with an option to compute PU-PSNR.
[Python] Added --temp-padding parameter.
[Python] Added --feature for dumping per-channel features
[Python] --heatmap-dir argument renamed to --output-dir. Now used for both heatmaps and features
[Python] Minor performance improvements.
[Python] --verbose will report much more information now.
[Python] It can create much larger heatmaps now (stores them in the CPU memory as float16)

v1.1.3 - 18/10/2022

[Python] Added "raw" heatmap type to the command line
[Python] Changed the way pyfvvdp classes are imported to avoid clash between the file and class names (see updated pytortch_examples)
[Python] Installation of PyEXR is now optional (caused problems on some operating systems)

[Python] Updated Python dependencies - now works with earlier versions of PyTorch, Numpy and SciPy

[Python] Thoroughly redesided the interface so that the code can be easily run from python code.
[Python] All code examples are now both in python and Matlab.
[Python] Optimized code for the Laplacian pyramid decomposition - faster predictions. There is a small (<0.02) difference in the predicted JODs.
[Python] Improved compatibility with pytorch 1.12 + better performance
[Python] PIP package
[Matlab] Improved computation of eccentricity for wide-field-of-view displays
[Matlab] Added new option 'ignore_boundary_pixels'
[Matlab] Fixed the cortical magnificatioon bug in the CSF

Updated examples in README.md
[Python] Fixed temporal convolution issue, which could cause problems on some versions of PyTorch. The code should be also a bit faster.

Version 1.0 is explained in the SIGGRAPH Asia 2021 paper (see README.md).
[Matlab] Updated calibration: added parameter on cortical magnification and optimized for the weight of the transient channel
[Matlab] Added support for reading videos from files (as in the python version)
[Matlab] Metric now reverts to CPU of no CUDA CPU available
[Matlab] added "foveated" parameter to 'fvvdp' (instead of an option)
[Matlab] fvvdp is now more verbose about its photometric and geometric settings
[Matlab] added an option to list available displays
[Matlab] extended the documentation and fixed a few typos
[Python] added the same "image" mode as Matlab version for much faster processing
[Python] added "foveated" parameter. Non-foveated is the default (used to be the opposite)
[Python] improved error reporting
[Python] added ambient light reflection to the display model (to sync with Matlab)
[All] raised ambient light levels for most displays to 250 lux (recommended office illumination)