Usually several test sets are created periodically to handle any type of situation in a variety of resolutions and depending on which part of the codec people a

Usually several test sets are created periodically to handle any type of situation in a variety of resolutions and depending on which part of the codec people are working on, this process is very complicated and take a lot of time to ensure that all the kind of content is representative. The metric is not optimal for video content, but gives an idea if they are going in the right direction. Ultimately, blind tests with people (experts and non experts) are done on the content to have subjective measures added to the objective ones