2-pass mode - How does it work?

[Gew]

Hi!

I'm just curious. I was sitting in the coach when the question struck me. I quickly came up with a theory, that sounded really logical to me, but it has happened before that my "logic" has been way of the radar, so therefore I must ask for confirmation.

So, here goes my theory. Simply told, during 1st pass, eg. Xvid simply measures the compressability of each frame in the clip. From this calculations (stats-files) it knows an exact floating point value that shall be used for Quantization factor throughout the film, to achieve the desired average bitrate. Period.

Now, is this a good (yet perhaps a bit simplified) explanation on the process, or am I all wrong?

Ty in adv~
Regards~

[jagabo]

During the first pass Xvid compresses the video in target quantizer mode with a quantizer of 2. It performs a full analysis at the current settings. It then knows how much relative bitrate each frame will require. During the second pass it uses this information to allocate bitrate to each frame/scene. But it's not as simple as using a fixed quantizer. It is sensitive to what can be seen by the human eye at normal playback speeds. Take for example a shot where a single frame is brighter because of a still camera's flash. A fixed quantizer approach would allocate a lot of bitrate to the brighter frame because virtually every pixel changes brightness. But Xvid knows that you won't notice if that single bright frame is encoded at a lower quality -- it will flash by too fast for you to perceive it.

Also note that fixed quantizer encoding does not give linear results and may not be comparable between different videos. The relationship between quantizer and bitrate will vary depending on the video. So just because a particular video ends up at 1000 kbps at Q=3 and 2000 kbps at Q=2 doesn't mean that the same ratio will be true for another video.