Codec tests

Tests consisted of playing a previously recorded Hi Band Umatic SP tape through the CODECs at 128kbps, 384kbps and finally at 2Mbps. The still images on the tape consisted of text, handwriting, line drawings, well defined colour graphics (maps and charts) and poorly defined graphics (leaves and earth samples). Moving images on the tape consisted of surgery and radiological clips, talking heads’, classroom shots and sports.

The output from the CODECs were viewed in parallel on four specially matched monitors and recorded individually on a separate Umatic recorder. To prevent bias the outputs of the four CODECs were sent to monitors which labelled A, B, C and D rather than being labelled with the manufacturers name. An audio tape was then played individually through each CODEC at 384kbps as a test of their audio performance.

Five test sessions were held with between five and ten people the same group of users or potential users attending each session. Panel members were classified as real end users’ if they had used or wanted to use LIVE-NET and interested parties’ otherwise.

The panel members preferred the CLI CODEC to the BT CODEC by a 2:1 ratio. Nobody expressed a preference for the PictureTel or V-TEL CODECs.

The CLI CODEC seemed to be preferred by those (who formed a majority of the panel) who cited picture brightness and a good representation of the edges, details and accurate colour reproduction of still images as reasons for their choice. These users tended to say that their LIVE-NET material was at least equally visual and verbal. Some of the qualitative comments in favour of the CLI CODEC were :-

• “better sharpness of image and good colour”
• “better definition at 384”
• “sharper and clearer for still subjects”
• “better for detailed still pictures at low speeds [128kbps]”
• “slightly better for stills”
• “better for X-rays”
• “allows background to be still while foreground moves”
• “better for movement”
• “better on sound”
• “best on sound”

There were also comments against the CLI CODEC, however for those who preferred this device these were not too important :-

• “image persistence problems”
• “takes the longest time to get into synchronisation upon changes of scene”
• “considerable time lag, particularly at lower rates”
• “jerky movement”
• “less good at showing lecturers”

BT seemed to be preferred by those for whom personal presentation is important. For these people it is important that the system produces a smooth moving image with good reproduction of facial expressions and synchronisation between sound and lip movements. Some of the qualitative comments in favour of the BT CODEC were :-

• “Better graphics”
• “motion is better”
• “better movement at 384”
• “better for movement and synchrony”
• “overall at the highest bit rate, BT was the best in all respects”
• “clearer definition of people lecturing”
• “better for movement in close-up and mid/long shots and in general for images of people”

The rating of sessions on a verbal/visual balance by the various groups are given here. The first group preferred the BT CODEC whilst the others chose the CLI.

More verbal than visual :-

Ancient Greek philosophy, ancient warfare and administration

Equally verbal and visual :-

Geology, microscopy, global environmental change, medicine, computer science, book preservation and library user education

More visual than verbal :-

Mathematical physics

Strongly visual :-

Radiology and surgery

Most of the differences between the two groups were relatively small, different people would often cite the same reasons for preferring different CODECs. Some qualitative comments indicating the reasons for the preferences given :-

• “Couldn't tell the difference between monitors”
• “Difference between A and D is marginal”
• “At the highest bit-rates there was little difference”
• “Close between A and D”
• “Very difficult to decide”
• “It is hard to choose between A and D”

Any differences would be unlikely to greatly effect teaching :- “All in all the differences in performance were inconsequential in most teaching settings

The H261 standard for data compression ensures interoperability between equipment from different manufacturers. It is however a lowest common denominator since it has been frozen in time to allow manufacturers to conform to it and therefore does not allow the use of the most up to date algorithms or equipment. Proprietary algorithms such as used by PictureTel can increase the quality of the picture but at the loss of inter-operability.

---

Contents

Graphics     Multimedia      Virtual Environments      Visualisation      Contents