Have we reached the limit of lossy codecs?
Reply #5 – 2008-04-20 00:09:57
I don't know if we will ever reach full efficiency with lossy encoders. There is no such thing as "full efficiency" aka "perfection heaven". The gains just get smaller and smaller but never hit a brickwall. A technology isn't at its end of life when nothing can be improved anymore, but when the cost of improvement vastly outweigths the gains. Without any disruptive innovation - a wonder if you want to call it like that - stereo lossy encoding has already reached this point - the current improvements only provide very minor improvements or fix exotic use-cases. If you disagree, then i propose to get your referencepoints right: Compare the current improvements with the improvements made 6 years ago, and you get a better picture of where we are now. So, without a wonder, stereo lossy encoding has already reached its optimum - significant progress may only happen outside of that scenario (i.e. surround encoding). HOWEVER, if we change the criteria from soundquality/filesize to something else - like for example platform support and ways to use lossy compression, then there definatelly is a lot of room left for improvement. - Lyx P.S.: With this, i do not mean that every developer of lossy codecs should drop the project. What i propose however is this: How many lossy compression codecs do we need? One? Two? Three? Okay, lets asume three. How many devs do we need per codec, to care about maintaining deadend stereo lossy codecs? Probably no more than 5. Thats 15 decs worldwide for developing compression-efficiency of stereo lossy codecs. Not much. Now add more devs who look into improving usability, features and support of codecs. Thats what i would call an efficient distribution of dev resources. The primary target should no longer be improving conventional stereo lossy codec compression - this does not mean abandonning improvements, but making it second or even third priority.