Other Tests

Back Story
Other Tests

I came across one test which attempted to demonstrate this problem. I shall dub it the Halo 2 Test.

The Halo 2 Test consisted of a pair of Xboxes running Halo 2. These Xboxes were connected via System Link and connected to two separate screens, on an HDTV and one a CRT monitor. The players were positioned in roughly the same location and were both facing a part of a map where a computer controlled train flies by. As the train flew by, the tester took a picture of both screens side-by-side which clearly showed that the train was not in the same location on both screens. The tester proceeded to mark the difference with black lines in order to show just how far out of sync the two screens were.

This particular test has several issues with it. Basically, the two displays are not displaying the same image nor should anyone be disillusioned into thinking that the images are remotely similair.

The displays are being fed from separate Xboxes connected via a network in a multiplayer game. If you're unfamiliar with possible issues this raises, I'll point them out for you here:

  • Network latency
  • Client prediction
  • Server processing lag

There's no way to gauge exactly how much any of these would have affected the actual test.

In addition to these potential variables, the final result is hardly quantifiable. I suppose you could measure the distance in the picture between the two train locations on the screens, but this is more or less akin to catching a fish this big.

A second test, which actually was the first version of the test I ended up settling on, involved measuring the audio latency between the direct output from the source and the output from the television.

This basically involved setting up a laptop with a microphone in front of the speakers of the television and the receiver. The volumes were levelled to be roughly the same and then a DVD was played and the audio recorded on a laptop.

Once the audio was recorded, it was opened up into an audio editor what would show the actual waveform of the sound.

Playing the sound directly would produce a blatantly audible echo, isolating a single sound as a waveform wasn't difficult (drum hits work nicely) and then getting a rough measurement from the beginning of the actual sound and the echo sound was simply a matter of measuring the time between them in the audio editing application.

This particular test also has a problem with it. It makes the assumption that the television is re-syncing the audio to the delayed video. In my particular case, this is exactly what the television was doing and this test corresponds with the results from the actual test I settled on.

I can not however guarantee that all HDTVs work this way. It is important to remember that the audio is a symptom of the video problem. Therefore it may or may not be completely accurate.

Next : My Test