Just to add some background to Floh’s wisdom, because I’m procrastinating:
When Media 100 first came out the far majority of people were working with analogue video, and even if they were working with digital video it was digitised through analogue connections (ie. component). The Media 100 cards were essentially analogue to digital converters, and the Media 100 system did it better than any of their competitors at the time.
Because Media 100 was converting the analogue video into a digital file, and the Media 100 was also converting that digital file back into analogue video, there was no compelling reason for Media 100 to follow anyone elses digital standards-eg D1 rectangular pixels. In fact as all computers work and display square pixels it made a lot of sense for Media 100 to digitise video as square pixels, so working with After Effects and Photoshop was much easier. As the ratio of a standard 4×3 TV is 1:1.333, and PAL has 576 vertical lines, you can see that 1.333 x 576 = 768 square pixels. That’s where the number comes from.
But as digital video technology became more accessible through DV, and digital betacam continued to become more widespread, Media 100 was upgraded to accept both SDI and DV digital video directly. Because there was no analogue to digital conversion, digitsing SDI and DV was just a matter of shuffling data around, and so the Media 100 system was adapted to work to the broadcast industry’s digital standard of 720 rectangular pixels. If you are digitising SDI or DV you have no choice but to capture at 720 pixels, 768 is simply not an option.
The squashing you refer to when working with widescreen footage is again normal and the industry standard. All widescreen video (standard def) still has 720 horizontal pixels, they’re just stretched out more than the ones in a 4×3 image. If you want to produce widescreen material with square pixels you need 1024 of them, not 768, and even then you have to squash those 1024 pixels down to 720 before you can output them via Media 100 because – just to repeat it – the industry standard for digital video has 720 horizontal pixels, for both PAl and NTSC, and 4×3 and 16×9 video.
So to sum up – 768 pixels is a legacy from the old days of working with analogue video. If you are working in a purely digital environment you can’t even select 768 pixels as a digitise option. The number of horizontal pixels – 720 or 768 – has no relevance to 4×3 or 16×9, as either way they have to be stretched and squeezed as you describe in After Effects. If you are working with analogue video then it’s best to use 720 pixels as you can then mix and match your footage with other digital sources if you need to.
Anyway I hope this historical background helps you understand where 768 fits in. i have to get back to work…
-Chris