Why does QT play FC dv media with sq. pixels?

[Dan Riley]

I believe it works like this;
If your display is a computer monitor
then you have a choice to make as far as display goes.
If you are using FCP there is a checkbox in the viewer and canvas
for square pixels so it’s displayed correctly assuming it’s DV or
601 video. There isn’t a checkbox in the quicktime player per say,
but like you said, you can change the display.

If you play that same DV quicktime out to an NTSC monitor, like I can do with
Aurora Pipe Studio and I believe AJA cards too, it will be displayed with
square pixels. And if you exported that DV quicktime to a 640 by 480 size,
it would display correctly on the computer screen and the TV.

Dan

[David42]

I hear you, that “normal” video monitoring ought to be thru an NTSC signal chain; say your AJA, or at least the FW deck to a monitor. That said, straight ntsc editing/output for tape playback is a declining part of my business. VGA, U-tube is as “real” as videotape and TV. Conference big screens are often fed from laptops.

What I’m trying to understand is why 16:9 widescreen DV 720×480 exports and plays wide in Quicktime (display at 853 pixels), while a 4:3 versions of the movie, letterboxed to a standard 4:3 dv 720 x 480 plays square pixel, instead of automatically setting to display at 640. When I made an AVI 720×480 of the letterbox, quicktime opened and displayed at 640, the way it should.

There is virtually no circumstance when I want a 720 x 480 frame to display square pixels, although I tolerate it for Photoshop graphics work. A 720-480 Quicktime ought to batch-default to CCIR601 4:3, the same way it corrects to 16:9.

Understanding my signal path, various playback ap characteristics isn’t rhetorical, or limited to 720 vs 640 displays. It comes up in review mp4 and h264’s, and in final output forms, is potentially more problematic in HD variations, which are frought with non-square pixels.

I should be able to send full quality movies off to clients without having to open and modify the pb controls on each clip to play as shot.

grumble.

[David42]

OK, I have peace, if not satisfaction, on the question of why some FCP 720×480 exports play in QT without distorting, and other’s defaulted to square pixel. To keep it simple, let me talk export from FCP, not Compressor. Your results may vary, but this is what I got.

If I mark a clip (not a sequence), export a movie using the File/Export dropdown. there are two choices, Quicktime, and Using Quicktime Conversion. My habit was to default to the first choice, because it was preset to “Current Settings”. That produced DV movies at native resolution, a no brainer, no options. However, in a standard 4:3 DV project, this export path leaves off the 4:3 “display at 640” playback instructions on the resulting movie. There are other options behind “current settings”, but they still didn’t give me 4:3 playback results, and had less control than the next option.

Alternately, I mark a clip, take the File/Export/Quicktime Conversion path. I get a more complex dialogue that allows you to select options. It also defaults to “Most Recent Settings”, which is handy, once you’ve defined export settings. By selecting options, I can choose video settings for DV/DVCPRO, and it reports back a 4:3 aspect, and (640×480) display from a 720×480 file size. There are a world of options here, besides the obvious dv 16:9.

By exploring the options in the Windows/Export Que toolset, I got similar options of less control sq.pxl vs more control 4:3 results . I’m going to guess that Compressor options also give results of sq.pxl vs good 4:3 display results.

I never used to worry about this, since it all came out in the wash if I re-imported to FCP, printed to tape, or output to DVD. It wasn’t until I got tangled in non-NTSC movies for vga display that the missing 4:3 pb setting became important enough to lose hair over. It doesn’t explain why the capture tool creates QT files that default to sq pxl playback in quicktime, but i guess I can live with it or deal with it, now that I have a hook to hang my ignorance on.

Hope this helps someone else keep their hair.

[Bret Williams]

I think the answer to your last question is very simple for DV at least. It doesn’t alter anything in any way. It copies the pixels from the DV tape and puts them in a qt shell. Perhaps to add the sq pixel marker it would have to do something that might alter or disrupt capture. Probably not, but it’s a theory.

However, I would suggest you render out your 720×480 files to 640×480 anyway since they would be a much sharper image than an interpolated on the fly 640×480 QT. The sharpest display is always pixel for pixel actual size. Not .9 pixels per pixel.

[David42]

“The sharpest display is always pixel for pixel actual size. Not .9 pixels per pixel.”

Good point, one I had not grok’d, and possibly explains why FC encourages us to monitor video on a video screen, and dodges the “display as 640” instruction. Getting the sharpest QT window is one of the reasons I go to “open in editor” from FC, eliminating “not really real” artifacting. ‘Actual pixel” is sharper on my digi-vga screens.

640 certainly makes sense for VGA release/playback of progressive scan movies. Apple has now set the mark at 640, with their Itunes entry into digital movie distribution.

Unfortunately digital video cameras and DVD took a different path, and have a lot of tech momentum and inertia. In terms of mastering/archiving, I want to maintain a 1:1 relationship with the aquisition tools and tape format, where 720/CCIR601 interlaced is the best SD performance, less processing, and nominally no trancode/generation loss from tape to desktop to tape.

[Bret Williams]

Pre HD, TVs were 4:3. A little math tells you that 720×480 is not 4:3. But Televisions only have “pixels” in one dimension. Vertical scan lines. There essentially is no horizontal resolution on the horizontal. So cameras and editing gear can only increase their rez in one direction. Somewhere along the way 720 became the norm. Probably with the first digital gear since an actual pixel number had to be used. You used to hear a lot more claims or “500 lines of resolution” or “600 lines” etc in the analog days. And they meant that you could discern 500 separate lines. To me that meant that you had to have the equivalent of 1000 horizontal resolution because there has to be space inbetween also, right? Seemed like hype.

Did you know that in QT you can turn on “high quality” and you’ll see the DV stream just as clear as in FCP. Sharper actually. You can turn that on and save the file so that it then plays with it on.