In general, the 2-pass process achieves a higher rate of data compression (smaller file sizes) and better image quality. But that's at the cost of speed during encoding. Normal video editing software like Pinnacle Studio enables the 2-pass process, but it is cumbersome. Both passes must be started manually, one by one. Applications that are optimized using DivX, for example Flask XMPEG 4.2a, can automatically trigger the two processes without anyone having to lay a finger on it. This is considerably more convenient.

In the test, the northbridge of the KT333 chipset caused problems on the Gigabyte GA-7VRXP (Athlon XP). It is only with an additional cooling unit for the northbridge that the sporadic crashes could be prevented in 2-pass encoding. All of the other three test platforms, however, ran flawlessly.

General Parameters: Now We're Getting "psychovisual"

Before we get to this feature, we would first like to explain the bases of "psychovisual" modeling. Human perception occurs in such a way that we can very quickly distinguish the important things in an image from the unimportant ones. That is the idea towards which psychovisual modeling is oriented. An algorithm scans a scene and is supposed to separate the important objects from the unimportant ones. A less precise compression process is allocated to unimportant objects; important ones are compressed more precisely. In our first tests, we experimented with this feature on a handful of video sequences. We have not determined any significant differences in quality up to now. We will inspect this function more closely at a later date. Even from DivX.com we hear: "This area is full of possibilities we have only just started on and will continue to explore."