Rumbah
3rd December 2015, 18:58
Hi,
I thought of doing a public blind test with different codecs.
My idea is to take about 20 ten seconds scenes out of Tears of Steel 1080p that have a similar complexity as the whole movie.
Then encode them with different codecs/settings/bitrates so that you have about 15 variations of a scene.
Then view the variations in a random order without knowing which one you currently watch.
After every 10 seconds video you have <10 seconds to rate the video. Then the next variation plays.
Doing this three times for a scene takes about 15 minutes so I guess that's a reasonable time.
I'd like to hear your opinions about the idea.
As I don't know if there is a free tool available I thought about writing a little script that shows a slider on top of a MPC-HC portable after a scene. I know it's only for windows then but my programming skills are very limited.
Should the scene selection be completely random or should they be picked by hand? I thought that the scenes coded by the same crf in x264 should have about the same bitrate. That way you get roughly the same complexity as the whole movie as a whole movie is a typical encoding use case and not a collection of worst cases.
I thought of doing a public blind test with different codecs.
My idea is to take about 20 ten seconds scenes out of Tears of Steel 1080p that have a similar complexity as the whole movie.
Then encode them with different codecs/settings/bitrates so that you have about 15 variations of a scene.
Then view the variations in a random order without knowing which one you currently watch.
After every 10 seconds video you have <10 seconds to rate the video. Then the next variation plays.
Doing this three times for a scene takes about 15 minutes so I guess that's a reasonable time.
I'd like to hear your opinions about the idea.
As I don't know if there is a free tool available I thought about writing a little script that shows a slider on top of a MPC-HC portable after a scene. I know it's only for windows then but my programming skills are very limited.
Should the scene selection be completely random or should they be picked by hand? I thought that the scenes coded by the same crf in x264 should have about the same bitrate. That way you get roughly the same complexity as the whole movie as a whole movie is a typical encoding use case and not a collection of worst cases.