50p->50i travails

The EBU (European Broadcasting Union) declared in 2009 that 1080p50 is the future-proof production standard which offers easy conversion to other formats in the short term. True it reduces very cleanly to 720/50p by scaling, very cleanly to 1080/25p by discarding every other frame, and seemingly very cleanly to 1080/50i by discarding half the lines from each frame.

Certain problems arise when an edited 1080/50p video is reduced to 1080/50i. Excepting with CRTs, it's uncertain how a 1080/50i will be displayed. Non-CRT displays and projectors in effect deinterlace the 50i. Some convert the 50i to 25p; some convert the 50i to 50p. Some use very crude deinterlace methods (even discarding half the 50i data); some use sophisticated methods. QuickTime Player offers a small taste of this variety where you can deselect both 'single field' and 'deinterlace' or else select one or the other. (Selecting both is nonsense and simply ignores the 'single field' selection.)

With both 'single field' and 'deinterlace' deselected QuickTime adds two consecutive fields to make a frame. This is how QuickTime was set when I watched the 50i made from an edited 50p and I was horrified. Half of the cuts appeared as 1 frame dissolves. The edited 50p incorporated some 25p material which FCP rendered as doubled frames. Half of this material now appeared as moving double images. The moral is that 50p camera material can be cleanly reduced to 50i, but once you edit the 50p material there's equal chance a cut will occur at an even frame as at an odd frame of the film. And for the incorporated 25p material, there's equal chance its doubled frames will be even and odd frames, as odd and even frames.

I returned to the 50p edit and tweaked it to have all its cuts at even frames and all rendered 25p material start with two like frames. The resulting 50i should have some immunity to cuckoo display.

Dennis Couzin
Berlin, Germany

Yea, you are quite right on this one. Wish there was a setting in 50p (even 720p) that can force the edit decisions to only occur on even frames. It's nuts to cut a feature in 50p and realize that you didn't observe this rule and now have to spend a good weekend running through the cut to fix this issue, or worse still, be actively conscious of this fact whole you are trying to craft the story.

Did you try making an SD DVD or hd DVD and playing that on the Apple DVD Player app? I believe it actually displays 50 Hz on a computer monitor. At least I think it was 50 Hz that I saw, as I was using it to identify inverted fields once when I didnt have a broadcast monitor or tv.



www.strypesinpost.com

strypes: I confirm your recollection about Apple DVD Player. It does display 50 Hz on a computer monitor.

I made a 50p test .mov of a hand sweeping around a dial. It can be downloaded here. From this a 50i test .mov was made with Compressor. It can be downloaded here. Playing the 50i test .mov with either QT or FCP doesn't put 50 Hz on the monitor and you don't see the hand sweeping correctly. From the 50i test .mov an .m2v was made with Compressor. Playing the .m2v with either QT or MPEG Streamclip doesn't put 50 Hz on the monitor and you don't see the hand sweeping correctly. Finally I burned an SD DVD from the .m2v and just as you said Apple DVD Player puts a good 50 Hz display onto the monitor. The DVD looks very similar to the 50p original. I'm glad I overcame my cheapness and burned the DVD to see this.

I tried the same experiment with a black, white, black, white, ... flickering 50p test clip. Again the 50i .mov files did not display correctly with FCP or QT. They showed no flicker at all. But the .m2v didn't capture the interfield flicker. That's too much to expect from any interfield compression system. It yielded a slightly nervous grey.

Once we leave the realm of CRT display, it is not always clear what the correct display of a 50i is. For the sweeping hand, yes. For the flicker, no. We're in a pickle today because 50i has outlasted the CRT.

Dennis Couzin
Berlin, Germany

Apple DVD Player may be the strangest DVD player in the world. This is so at least for version 5.0.3 working in my system of OSX 10.5.8 with a nVidia 8800GT card, and I guess was so for strypes' version in his system. Same version 5.0.3 working with same OSX 10.5.8 with a Radeon 9600 card skips the strangeness and behaves much like other software DVD players.

My Apple DVD Player effectively deinterlaces 50i DVDs to 50p. (It probably deinterlaces 60i DVDs to 60p too.) With the right arrow key you step through the 50 frames produced from the 50 fields. (The left arrow key doesn't work.) Apple DVD Player cares whether the field dominance of the source material matches the field dominance set during the .m2v compression, and jitters plenty if it doesn't. No other software DVD (or .m2v) player I've tried does this. The other players show something like

at 25 per second while Apple DVD Player shows

at 50 per second.

The downside is that for complex 50i subjects Apple DVD Player can serve up astonishing rubbish. From my test DVD, where other players show something like

Apple DVD Player shows a sequence of




etc. Where did these blobs come from? Part of them might be in the 50i .m2v and simply cancelled in the combination frames shown in the other players. But it looks like the Apple DVD Player is at least partly responsible.

The moral: DVDs shouldn't be appraised solely from viewing on Apple DVD Player. It can make a DVD look better or much worse than it looks on the majority of players.

Dennis Couzin
Berlin, Germany