LCEVC, Baseline Video Codec & Bandwidth (1)
One of the surprising facts I learns was that YouTube are now doing up to 2Mbps H.264 Video for 1080P content. And the quality is quite impressive for Internet Video. And the world doesn't seems to want to save bandwidth from Audio Codec, vast majority of it are using 256Kbps AAC / Opus rather than 128Kbps.
As bandwidth gets cheaper. At what point do we arrive demonising return of newer better codec? Let say providing a new codec on Youtube, would actually add 1 Exabyte of storage as they convert original video to new codec, while only using 30% of the H.264 Bandwidth. Because of the long tail of adoption curve, For the first 5 years less than 40% of user will use this codec. So their saving would be 28% of their bandwidth.
Does 28% saving of bandwidth / per year outweighs the cost of extra 1 Exabyte storage and increasing? The roadmap for both storage and networking is very clear. The cost of storage unit isn't dropping much, while we continue to see double bandwidth of Ethernet and Fibre Optics every 4 years. 400Gbps Ethernet now, 800Gbps Standardise, 1.6Tbps standard is coming in 2025 and on market by the end of the decade. It is not so clear if the price /GB of HardDrive will drop all that much. Even in the current market where SSD ( commodities ) are dropping fast, Hard Drive RSP hasn't moved a bit.
So if we were to optimise for Storage and not Bandwidth, may be the best idea if not to introduce a completely new codec, but to build on top of H.264? LCEVC [1] MPEG-5 is basically an enhancement layer on top of existing codec that brings better quality. You get the best of both world.