Subjective quality of internet video codecs article on EBU website

[Arachnotron]

Maybe the following is of interest to someone:
In order to evaluate the performance of video codecs for the internet, EBU Project Group B/VIM has developed a new subjective evaluation methodology called SAMVIQ (Subjective Assessment Methodology for Video Quality). This new methodology was used recently during B/VIM’s Phase 2 subjective evaluations of four codecs designed for internet use: Envivio MPEG-4, QuickTime 6, RealNetworks 9 and Windows Media 9.

This article gives a short description of SAMVIQ and summarizes the main findings of the Phase 2 subjective evaluations.

Get it here (http://www.ebu.ch/trev_301-samviq.pdf).

[Nic]

Interesting read...

An example from the text:
What is the quality improvement between one generation and the next of the same codec standard?

The new generation of codecs (e.g. WM9) is not always better than the previous generation ones (WM8). Sometimes, the reverse is true: WM8 performs better at 500 and 700 kbit/s than WM9. Similarly, RealNetworks 9 is slightly worse than RealNetworks 8 at 250 kbit/s but similar at higher bitrates. The QuickTime codec has the same difficulty: the grades of the new version of QT6 are consistently lower at all bitrates than those of the older version (at 1.4 Mbit/s this difference amounts to 15
points).

Controversial, ;)
-Nic

[Wilbert]

Besides, CIF (let alone QCIF) sized video is much too small for streaming video nowadays.

[Arachnotron]

Besides, CIF (let alone QCIF) sized video is much too small for streaming video nowadays.Maybe for streaming to mobile devices?

In general, I can imagine that as broadband internet continues to become available and bandwidth increases, compression quality at lower bitrates becomes less important.