Better to encode interlaced?

[8-BaLL]

Hey guys,

I wanted to encode my HDTV cap, but when I deinterlace the file (w Yadif), it doesnt have the same smooth motion as when its interlaced. Megui scanned the file and says its interlaced normally, top field first and suggested Yadif.

When I encode interlaced though, the smooth motion is still there as opposed to deinterlaced.

Here is an example file. Put all 3 to MPC (because it deinterlaces automatically when needed) playlist and pay attention when the camera goes up to the right wall.

The source and the interlaced encode are really smooth. The deinterlaced one is less smooth.

15 MB:

http://netload.in/datei8sL0tkNZOA.htm


So heres my question: is it really better to encode interlaced or am I just using wrong deinterlacing method?

[Ma-Xell]

By deinterlacing you discard half of the frames, which means you'll have half the frame(25 or 30p) rate of the original 50 or 60fps (depending on where you live).
If you want to deinterlace and keep the smooth motion (the 50 or 60fps) you have to tell the deinterlacer to keep these frames and not to discard them.
You can do some reading on http://100fps.com/

[8-BaLL]

Thanks for the info, but why does Media Info report the video then as 25 frames per second? (source, interlaced HDTV video)

Also doesnt an interlaced video basically contain of 25 progressive frames, where every frame got split in 2 halves, making this way a 50i video?
Or does a 50i vieo contain 50 different and unique frames per second, which got interlaced?

So judging by that I will simply encode the file interlaced as MBAFF. The file size is almost the same, but the motion fluidity stays as opposed to deinterlacing (unless BOB was used, which doubles the size I guess).

[Ma-Xell]

Thanks for the info, but why does Media Info report the video then as 25 frames per second? (source, interlaced HDTV video)
Because technically the video is 25 frames (which are interlaced) and this can later be deinterlaced to produce a 50 fps progressive video.(This is done automatically by TVs)

Also doesnt an interlaced video basically contain of 25 progressive frames, where every frame got split in 2 halves, making this way a 50i video?
Or does a 50i vieo contain 50 different and unique frames per second, which got interlaced?
(I think) A 50i video would have 50 technical frames(but interlaced) and it could be deinterlaced to produced 100fps.
And yes, interlacing means: There are two frames stored in a single frame, and this makes that single frame interlaced.

For your video, your goal is the smoothness, so you can
-either keep it interlaced (when you watch it on your tv it will look smooth and when you watch it on your pc you need to tick: deinterlace and! double fps or something along those lines in ffdshow or whatever decoder you use.)
-deinterlace and keep all 50 frames per second and THEN encode into 50fps progressive video (this case your pc will play it back smooth instantly and so should your tv)

[TheSkiller]

(I think) A 50i video would have 50 technical frames(but interlaced) and it could be deinterlaced to produced 100fps.Basically yes, I would call it that way too, but many use the term "50i" when they talk about a video with a framerate of 25 with interlaced frames (yielding 50 fields per second).
I personally think it is inconsistent to mention the number of frames in the case of progressive video (like "25p") and the number of fields per second (like "50i") in case of interlaced video instead of talking about frames the whole time (like "25i") – in the end it's always the frames that matter, fields just mean the playback rate is doubled and that's not so hard to forget, is it.


And yes, interlacing means: There are two frames stored in a single frame, and this makes that single frame interlaced.Yes, but such a "frame" within a frame is called "field" because it's not a full frame but half a frame. :)

[Ma-Xell]

Yes, but such a "frame" within a frame is called "field" because it's not a full frame but half a frame. :)

Touché :stupid:

[Asmodian]

I personally think it is inconsistent to mention the number of frames in the case of progressive video (like "25p") and the number of fields per second (like "50i") in case of interlaced video instead of talking about frames the whole time (like "25i") – in the end it's always the frames that matter, fields just mean the playback rate is doubled and that's not so hard to forget, is it.

Actually I think 50i and 25p are more correct, 50i contains video information from 50 points in time every second while 25p only contains information from 25 points in time every second. A progressive source converted to interlaced is a different story of course but conversions can get complicated fast.

[Didée]

framerate'codingtype /temporalresolution

25p - PAL FILM
25i/25 - PAL FILM, encoded as interlaced (2:2 pulldown)
25i/50 - PAL VIDEO
24p - NTSC FILM
30p - NTSC progressive
30i/24 - NTSC FILM (3:2 pulldown, encoded interlaced)
30i/60 - NTSC VIDEO


There exists a convention like this, just that too few are actually using it.

But hey, where would be the fun in using distinct nomenclature. 20% of the forum's database consists of conversation trying to figure what each other is actually talking about.

[8-BaLL]

speaking of interlaced stuff guys, is it possible that a PAL VOB file (official untouched music video) isnt interlaced at all? I have both NTSC and PAL version, the PAL one doesnt seem to be interlaced at all (750x576). Wonder if its possible.

[TheSkiller]

Sure, that's possible.
If it was shot and produced in 25p the PAL version would not have any interlacing. In such case the NTSC version of the same footage would most likely be a fieldblended normconversion of the progressive PAL video. It would therefore show interlacing.

There are also other possibilities (that more or less butcher the video) but without looking at the video, this seems to be the most plausible answer.

[dukey]

Originally I would have said, always deinterlace. But if you use EVR and have say an nvidia graphics card, the motion adaptive deinterlacing it uses is pretty damn amazing. In these cases it's best left interlaced imo.

[drbuzz0]

Just a quick question: What do you intend to do with the encoded video? Is this for casual viewing for pleasure or for editing into something or for use in a web video etc?

I usually prefer to deinterlace at the time of encoding because it takes the burden off the system when playing back. You don't have to deinterlace in real time, you can take as long as you want to deinterlace the video and just leave the system to run overnight if you want.

As for loss of smoothness, yes, that can be an issue, because you are effectively dropping the field rate down (not the frame rate but the field rate)

But it depends on *how* you deinterlace the video. You will never get it to be quite as good as truely progressive video, but you can get very close. YADF is a good deinterlace filter but it's not necessarily the best around if you want to absolutely preserve quality and get the best results. There are some pretty good full motion compensation deinterlace methods out there. You could use a framerate doubler to deinterlace and keep the motion very smooth.

[8-BaLL]

@drbuzz0- its a live recorded TV show "beat the star". While I will not watch it very often, Im a big fan of fluid motion and I planned to encode it interlaced, since deinterlacing with yadif to 25p took away some motion fluidity.

I dont see the purpose for using a frame doubler when we can use MBAFF option in x264 and keep the file originally interlaced and reduce bitrate. I think the doubled frames (BOB) make the file size bigger and defeats the purpose of encoding.

Since my main player (Dune HD TV-301) has great "Motion Adaptive Deinterlacing" I dont mind interlaced files at all. If it keeps the motion more fluid, Id rather not deinterlace a file.