An interlacing pattern you may not have seen before?

[aramkolt]

I spent several hours tonight trying to figure it out an interlacing mystery because I thought my hardware had to be malfunctioning since it looked like I was getting a progressive video despite recording from a an analog composite source which should always be interlaced.

The source is a U-Matic tape that would have been used by TV stations like MTV to play music videos.

Of note, the file actually is interlaced, it just looks like it is progressive. Also, every 4th or 5th frame is a duplicate.


What I believe is going on here is that it is PsF or progressive segmented frame. While I had seen those letters put together, but I've never seen it in the wild.


Basically every pair of fields is from the same moment in time and there's only motion introduced every 2 fields.

Captured file segment attached if you're curious what it looks like. The file is actually interlaced and came right from the AJA KiPro this way without any modification other than to trim the clip to a few seconds. The AJA KiPro can do hardware deinterlacing at the time capture, but it was not set that way.

Maybe this will help someone else not think their hardware is acting up and doing deinterlacing without their permission!

[vwestlife]

Could it have been 50 Hz PAL video converted to 60 Hz NTSC?

And then possibly even converted to 24 fps to give it that "film look", and then converted back to 60i for broadcast...

[RayNotes]

vwestlife is right. It's 24P converted to 30P by duplicating every 4th frame, then interlaced to 60i. The attached is your video, first deinterlaced to 59.94P then decimated to 24P while removing duplicate frames.

De-Interlaced 24fps.mp4

[aramkolt]

Well, it really shouldn't be a PAL source, the footage is recorded in NTSC for sure and this is an American band (EVE6).

Full YouTube version here that starts at the timecode of the sample:
https://www.youtube.com/watch?v=cjta4AKnHP0&t=46s

Here's the wikipedia page about PsF:
https://en.wikipedia.org/wiki/Progre...egmented_frame

From the page, they describe 24PsF as: "the first universal video standard which transcends continental boundaries, an area previously reserved for film"

Agree, they would have had to duplicate some frames to get it to play on 29.97FPS players for the NTSC version of the recording, so that part doesn't surprise me. Technically the source file posted *IS STILL INTERLACED*, just every pair of fields is unchanged, and that seems to be the point of PsF altogether - more or less storing a progressive stream as interlaced that can be played back on interlaced hardware and over standards like composite that were interlaced only.

So I would say they started with a 24P source, made it 24PsF, then duplicated frames to get it to 29.97PsF which plays on the same hardware as 29.97i and technically is regular NTSC/29.97i if I'm understanding correctly.

[RayNotes]

Correct - we're on the same page. I've never heard of 24PsF, I was only analyzing the video you uploaded. The version I replied with is the 24P that I processed out of your interlaced video - by deinterlacing it to 59.94P, then removing all duplicate frames back down to 24 FPS.

[aramkolt]

Cool, what was the actual script or method used to do it? I've never really heard what the appropriate deinterlacing method would be in terms of treating an interlaced frame as progressive and then removing the duplicates? Would be interested in whatever the most specific way I could reproduce it as I'm not great with scripts just yet.

The one thing I also noticed is that it seems like one of the fields is always slightly darker than the other, so you can kind of see what almost look like scan lines. I take it there probably isn't a good way to get the two fields that make up a progressive frame to match a little better?