Is Vegas reliable in importing 23.976fps footage?

[Richard Bartlett]

Chris – Apologies for any confusion. Video is an art and a science but more and more the science goes into the wings and the art is king. I’d like the 787 qualified by Dr Dropout or someone else fully authoritative. As I’ve said earlier I’m not the owner of the horses mouth. Also your milage might vary.

What the 655×480 said was that it was the native 1:1 pixel AR format for importing computer generated graphics into Vegas (assuming a D1/DV project with its Standard Def AR set). If you supply a 720×480 still frame with the project set for the DV/ITU601 aspect ratio – this will be adapted to the native resolution depending on your choice of setting for “simulate target AR” in preview.

Sonic Foundry (now Sony Pictures Digital – MediaSoftware division) have a history of calculating these things exactly. The industry has a tendency of creating standards based on choices made for how to represent the viewable area of 525 and 625 line TV systems.

If 768×576 works for you and draws circles round with a pixel AR of 1:1 then either Sony have written an exception for that size, your monitor or playout/render device may have a calibration issue, or you may have determined by trial that SoFo mis-informed us originally. 768 is a power of 2 number, or otherwise known as a computer or digital number. So it can be represented readily by digitisation circuits like TV cards etc. It is difficult for a signal digitiser that supports 787 pixels to not also support 1024 pixels and to do this without any further hardware changes. So if 1024 was available, the marketing departments would insist that it was used. Inside software, if 787 is the best representation for square pixels where 720 rectangular pixels were their equivalent, we just have the programmer do the math for us as we progress.

If I were a programmer, I’d draw 720×576 pixels and have the operator adjust the horizontal stretch of their monitor to to fit a graticule that the operator measured with a ruler. For importing square pixels, I’d offer crop or scale in the preferences with a right mouse option on the timeline element to override the default. This method would work OK for CRT, but would be quite bad for folks with LCD/plasma panels of almost any resolution or ratio. The traditional NLE assumed you’ve got the monitor set to show square pixels as that is what is expected from all the well known resolutions from the main graphics card vendors.

Irrespective of your resolution, if the still format you choose specifies the correct AR, Vegas should honor this. However this may end up with scaling artefacts. You may scrub and be more native with 787, 786 or 768, but when you render to DV, 720 discrete luma samples will be in the bits packed into each YUV 4:1:1 line and 360 U+V samples. If you render out to HiDef (upsampled) or for computer playback or perhaps a film transfer, then you’ll not compromise your target from the CG/stills point of view, but you may have an impact on your conversion from any DV footage you have when it goes out as 787 pixels per line scaled from the original 720.

If the science becomes painful, stick to the art. The science usually has some give in it.

[Chris Young]

Richard ~

You have thrown some light on the subject. It made me do some checking up with my first employer, the BBC. Of course back in the stone age when I first entered this crazy business there was no such thing as ‘digital video’. Using 768 x 576 has been an error I now discover. A small extract from their, massive on this subject, current technical training info (re PAL) follows:

Most computer painting systems use square pixels and therefore to
use such a package to make a still for export to Rec. 601 would
suggest that the “square pixel” image should start at 1024 by 576,
being resampled to 702 by 576, then placed centrally within a 720
by 576 black background. However the instant rise time transition
of blanking could cause severe ringing when an analogue output
signal is created. The better width to use for the “square pixel”
source is (1024 * 720 /702) = 1050 samples. After resampling to
720 by 576 the image can be output from an ITU-R BT.601 system.
Any subsequent DAC process is responsible for applying standard
“analogue system” blanking to the signal, giving a 52 microsecond
H.A.D. line. A 4 by 3 image should start life as a 787 or 788
sample per line “square pixel” source (ideally 787.69 samples!!).

Andy Woodhouse
Lecturer
BBC Training & Development.

So I now know that a horizontal 787 or 788 is the optimum to use. I shall be doing some serious experimentation with this. You live and learn don’t you! Thanks for pointers.

Chris Young
CYV Productions
Sydney

[Edward Troxel]

Here’s a long discussion on the topic:

https://www.sonymediasoftware.com/forums/ShowMessage.asp?ForumID=4&MessageID=81874

Edward Troxel