352x240 resolution with pulldown

[iantri]

Hello,

I encoded some film source (23.976fps) video files at 352x240 MPEG 2 to be put on DVD.

I encoded them with CCE and then used the pulldown utility to set the pulldown flags -- just the one does for 720x480 resolution files for DVD. When I do it with the 352x240 files, and author them, however, the result does not work correctly.

On one of my DVD players, the DVD skips and stutters. On the other one, it plays but looks like it is not doing the telecine, and the playback is jerky.

If I re-author them at 720x480, they work.

Can 352x240 MPEG 2 files not be encoded with pulldown flags? The videohelp.com page for DVD (https://www.videohelp.com/dvd.php#tech) seems to indicate that this should work, as long as it is MPEG2...

[adam]

It is compliant and definitely works. I do it all the time. Maybe your encoding has some other compliancy problems. When using 352x240 you must use a DC precision of 8 and a 4:3 aspect ratio.

What are you using to author the DVD with?

[Gazorgan]

352x240 must be progressivley encoded, not interlaced. I use that resolution all the time, but only with 29.97 fps.

With no fields, what is there to telecine for the missing 6 fake frames??? It's the MPEG2 version of MPEG1, just with better compression and more Audio choices.

[adam]

It definitely works at 23.976fps. That is always how I encode when I use 352x240 resolution. 29.97fps lowers the quality too much and it requires that you deinterlace.

352x240 mpeg2 at 23.976fps is fully supported in the DVD standard, and the DVD player is required to perform the telecine. Don't ask me how it does it with a source who's resolution is too small to support interlacing, but it does. It also does this for VCDs which use mpeg1, which also doesn't support interlacing at any resolution.

If you are going to use 352x240 in a DVD its definitely a much better idea to use mpeg2 at 23.976fps then to use mpeg1. You of course get the more efficient framerate (mpeg1 can only use 29.97fps when on DVD) and you also aren't limited 1.856Mbits like you are with mpeg1. You also don't have to deinterlace, which kills quality.

[iantri]

Don't ask me how it does it with a source who's resolution is too small to support interlacing, but it does. It also does this for VCDs which use mpeg1, which also doesn't support interlacing at any resolution.

Well, that's because it has nothing to do with how it is stored. The DVD player reads the pulldown flags and outputs the 23.976 frames into 59.94 fields in a 3:2 pattern. I never really understood this (and how TV stations can resize interlaced video to put stock tickers or other information like they do at the end of every TechTV show, without damaging the interlacing) until I realized that really, with NTSC video, you are not dealing with 29.97fps of video with every other field on every other line, you are dealing with 59.94 field per second video.

Anyway, the files, when authored, need to be 23.976fps MPEG2s with pulldown flags set (making them 29.97fps internally), right? Unlike the MPEG1 files of this resolution, which you can author at 23.976fps and are automatically telecined by the DVD player, correct? (In the same manner as you author a VCD)

Either way, TMPGEnc DVD Author says they are valid files, when import them.

Maybe I just got stuck with two poor DVD players..

[DJ_Izumi]

I had this issue with an RCA DRC212N DVD player. It would become 'Confused' with Pulldown material at 240p. In my case it would playback the material at what seemed to be a field bob making it 'bounce' sorta during playback. This was a flaw within it's progressive scan process, an earlier RCA DVD player and a later model I played the same content on did NOT have this issue.

I troubleshoot the issue like this: I put a 352x240 24fps (With pulldown flags) onto an XSVCD forplayback, the result was the issue I just discribed. Then I put the -same- video stream on another XSVCD but WITHOUT pulldown flags in it, so it was a truely progressive 24fps stream. Playback held no issues. I also had the same playback issues with MiniDVD and DVDR playback, I concluded that the cause of the issues was 240p material with pulldown. A simple design flaw in the DVD player. I'm guessing you have the same issue, perhapps you should try the same test I did with XSVCDs?

[adam]

Well, that's because it has nothing to do with how it is stored. The DVD player reads the pulldown flags and outputs the 23.976 frames into 59.94 fields in a 3:2 pattern... you are not dealing with 29.97fps of video with every other field on every other line, you are dealing with 59.94 field per second video.

Yes I know this, and that's why it makes perfect sense whenever you use video with a resolution of 352x288 or higher. My point is that a resolution of 352x240 is too small to support interlacing, so you'd think it wouldn't be possible to break it into fields which is necessary for the telecine. There's just not enough lines of resolution to support interlacing. But of course you have the same problem with VCD (resolution and mpeg1) and it works so obviously the DVD player has special routines for this. I'm just saying I don't quite understand how those routines work because they are not necessarily the same as with a DVD encoded at 352x480 for instance.

I know that VCDs actually aren't telecined with a 3:2 pattern. The actual pattern you get depends on the player and it is always going to be a less efficient telecine then if you had made a full D1 DVD, for example. Perhaps this is also the case with 352x240 mpeg2 DVDs. If anyone is interested I'll find the source where I got this information.

Anyway, the files, when authored, need to be 23.976fps MPEG2s with pulldown flags set (making them 29.97fps internally), right? Unlike the MPEG1 files of this resolution, which you can author at 23.976fps and are automatically telecined by the DVD player, correct? (In the same manner as you author a VCD)

for any NTSC compliant DVD resolution, mpeg2 can either be stored natively at 29.97fps or internally at 23.976fps with the pulldown flags added. The latter is the preferred method of storing NTSC footage and yes that is what you should use.

Mpeg1 is only supported in the DVD standard at 29.97fps. It is not possible to add pulldown flags to mpeg1 video at all. When making a VCD, yes you can store it at 23.976fps and the DVD player is supposed to telecine it automatically. But this doesn't work with mpeg1 DVDs. You either have to re-encode it to 29.97fps mpeg1 (bad) or re-encode it to 23.976fps mpeg1 (better). TMPGenc DVD author accepts 23.976fps mpeg1 sources for DVD output? Strange, its definitely not compliant.

[iantri]

DJ_Izumi says:

I concluded that the cause of the issues was 240p material with pulldown. A simple design flaw in the DVD player.

Yes, I'm sure you're right. 352x240 commercial DVDs are extraordinarily uncommon (sometimes the videotaped special features are in this format, but I've never seen anything at 24fps, only 29.97fps NTSC.)

My one DVD player (Sony SA-HT80, it's crap, doesn't work with CD-R or DVD+-RW) plays it as if it were a progressive stream -- it does not telecine it and has a fit in general when I try to play it.

The other DVD player (Malata DVP-393) jerks for a few seconds then starts playing it properly.

adam says:

Yes I know this, and that's why it makes perfect sense whenever you use video with a resolution of 352x288 or higher. My point is that a resolution of 352x240 is too small to support interlacing, so you'd think it wouldn't be possible to break it into fields which is necessary for the telecine. There's just not enough lines of resolution to support interlacing.

It's the opposite -- double 240 is 480 (full height NTSC). The DVD player does the telecine in real-time -- it outputs the video into 59.94 fields (each of which is 240 lines tall) in an alternating pattern of 3:2.

I know that VCDs actually aren't telecined with a 3:2 pattern. The actual pattern you get depends on the player and it is always going to be a less efficient telecine then if you had made a full D1 DVD, for example. Perhaps this is also the case with 352x240 mpeg2 DVDs. If anyone is interested I'll find the source where I got this information.

I'd be quite interested to know where that came from -- my DVD player's output (from VCD and from film DVDs) is definately proper 3:2 -- it is a trivial matter to do this in real time.

Thanks for your help.

[adam]

DJ_Izumi maybe you were just rounding up, but if you actually used 24fps then that is not supported so it will definitely cause problems on players.

iantri I don't understand what you mean. It is common knowledge that 352x240 is too low of a resolution to support interlacing. I am fully aware of how a soft telecine works. All I am saying is that a telecine cannot be performed until the frames are split into fields, and 352x240 does not support fields. Now I can only assume that this limitation is in regards to how it is stored, not played...since obviously it works and after all, tv's are field based scan displays. I'm just saying that the process is most likely different since, once again, 352x240 can NOT support interlacing. Basically, you had to lose half of your horizontal fields to get to 352x240 in the first place. When you play this back interlaced you are just line doubling. Yes it will play, but this is not the same as taking, say a 720x480, source and outputting each frame as two fields, and then performing the telecine. The frames making up the 352x240 source do not have enough lines of resolution to be split into fields.


iantri how are you determining that your DVD player is outputting your fields in a 3:2 pattern? That's impossible to tell by looking at it. Also its certainly not trivial if you don't have RFF/TFF flags which a VCD doesn't. I'm pretty sure the document I mentioned is on mpeg.org. I'm looking for it now.

[DJ_Izumi]

Yes, I did mean 23.967 and 29.97fps, I was just rounding up for simplicity.

A simple work around would be to resize to 720x480, apply a hardcoded 3:2 pulldown, then resize back down to 352x240 and encode. Resizing from D1 to CIF is a fairly 'cheap' way of de-interlacing, but it'll technicaly work.

You could alternativly switch to a resolution of 352x480, something I use all the time, since it has a vertical resolution of 480 the player shouldn't have much problem with the pulldown there, No promises however, it worked in my case, that's all.

[iantri]

iantri how are you determining that your DVD player is outputting your fields in a 3:2 pattern? That's impossible to tell by looking at it. Also its certainly not trivial if you don't have RFF/TFF flags which a VCD doesn't. I'm pretty sure the document I mentioned is on mpeg.org. I'm looking for it now.

Perhaps it is just me -- the difference between telecined 23.976fps video, 25fps progressive video, 25fps interlaced video, 29.97fps progressive video and 29.97fps interlaced video is obvious, to me, at least.

Any pattern other than 3:2 would be fairly obvious as the motion would be less smooth than usual.

Anyway, this is beside the point. You can connect the DVD player to a capture card and examine the output.

[DJ_Izumi]

Regardless, the issue is as simple as fallows: Certian DVD players have no issues 240p material with pulldown flags, other DVD players have issues.

There's not much you can do about it to make it work, you can switch to a 480p resoluion, buy a new DVD player, or resize the video to 720x480, apply a hard coded pulldown, resize back down to 352x240 and then encode that.

[adam]

iantri with all due respect, that is ridiculous. There are hundreds of different telecining patterns that are used. 3:2 is just the most common one. When you watch broadcast tv you will surely get a different pattern every time you change the channel. There is literally no way to see any difference with the naked eye. Indeed there is no way you could see the difference between 23.976fps telecined to 29.97fps and video stored at 29.97fps interlaced because if the same telecining pattern was used they are literally identical.

Yes you could capture the signal and analyze each frame. If this is how you determined that your VCDs were playing back in a 3:2 pattern then why didn't you just say so? That was exactly the question I was asking. I am still looking for the iso document that mentioned how dvd players cannot maintain the typical 3:2 pattern for VCDs. It was a long time ago since I last saw it. It really shouldn't come as a suprise though. DVD players cannot even maintain a perfect 3:2 pattern for commercial DVDs. It has to change it during scene changes. And it has the benefit of RFF/TFF to instruct the player, which VCDs lack.

In my opinion, the jerkyness that people are experiencing with 1/4 resolution NTSCfilm mpeg2 DVDs is a result of the less then stellar line doubling capabilities of the DVD player.