WHAT IS THE BETTER CONVERSION? NTSC 29.97 OR NTSC 23.976?

[BCAESE]

1. Which conversion is better for standalone DVD players?
DVD NTSC @ 29.97fps or SVCD NTSC film @ 23.976fps

My original files are SVCD PAL @ 25fps

2. I use TMPGEnc to do my conversion, must I use DVDAVI to create a .d2v file or can I convert just using my .m2v.

If I do have to use the .d2v file, WHY

[yoda313]

Hello,
If your dvd player only plays dvd use the dvd 29.97 setting. If it can read svcds the svcd setting will work. I believe tmpgenc can use .m2v files but you need a plug in. Check the tools section in this site and you should find one.
P.S. Use svcd 29.97 - that is the ntsc tv setting (ie dvd players need it). There are guides on how to use it without changing but since your converting from pal go straight to 29.97.

[BCAESE]

Great thanks for the advice

For DVD NTSC @ 29.97, Do I keep the pulldown @ 3:2 or use the default which is interlace.

I have used my video which TMPGEnc demuxed but I need to know if using the .d2v file is better or why use it

Don't mean to be a pain but I was reading a Tutorial from this site and it said I must use it. is this step relavant

[yoda313]

Hello,
You can just use the .m2v (or .m1v) file. But if your using a guide do what it suggests. I'm not sure about the pulldown choice. I know that involves the 16x9 ratio stuff. For the pulldown I'd suggest looking up another guide on this website. There are whole sections dedicated to tmpgenc. I hope this helps.
Kevin

[adam]

I have to disagree with the above advice. When making an NTSC DVD or SVCD you always want to encode to 23.976fps is possible. Its much higher quality and actually less problematic. If you have a PAL source then its definitely a bad idea to go straight to 29.97fps. You have to perform a hard telecine as opposed to just slowing down the video to 23.976fps. This kills quality and can introduce all sorts of additional problems.

You simply need to slow down your video and audio to NTSCfilm and then apply pulldown flags. You can do this in TMPGenc by using its 3:2 pulldown while playback option on the video tab. For audio you should use a separate program like BeSweet which has a built in option for this. There are PAL-NTSC guides on this site that explain all of this.

As for whether you go with a DVD or SVCD, you'll get better compatibility with DVD but you'd need to decrease the resolution to 352x480. Converting your 480x576 source up to 720x480 or even 704x480 will look terrible. So in this case, making a SVCD would actually retain more of the source's quality, at the expense of compatibility. The added bitrate of DVD doesn't do you any good here since your SVCD source's quality is the limiting factor.

As for using DVD2AVI....do it! Frameserving via dvd2avi takes an additional min or so and it is much more stable. Loading mpeg files directly into TMPGenc uses direct show filters to decode it. These really aren't very stable and you may find that your encode crashes or that you get corrupt frames periodically.

[yoda313]

Hello,

adam-I have to disagree with the above advice. When making an NTSC DVD or SVCD you always want to encode to 23.976fps is possible

Thanks for pitching in. Nobody seemed to be responding so I wanted to give some advice. I'm by no means an expert at conversion so other viewpoints are always welcome.
Kevin

[BCAESE]

Wow, Adam you seem like you know your stuff

I do use BeSweet for my audio. I was just asking because I have this one (Toshiba SD1700) out of my 3 DVD players that plays my DVD-R funny. It shows a double picture, sound is fine video quality is too.

My DVD-R's are in SVCD NTSC film 480x480 @ 23.976fps format.

I thought if I encoded PAL files to DVD NTSC 720x480 @ 29.97 everything would be perfect. Seems like I would have more problems and I don't want lose quality! (is it video or sound quality that gets ruined?)

If there is another way to get rid of the double picture my DVD player is spitting out. Would converting to DVD NTSC 29.97 get rid of that problem

[adam]

Your DVD-rs are in 480x480? Well that's not a DVD compliant resolution and SVCD must be burnt on cdr. That's why you are getting decoding problems.

Storing video at 29.97fps is inefficient. You have 20% more frames each second, which eat up bitrate, and your video has to be interlaced which requires alot more bitrate as well. If you store your video at 23.976fps progressive it gets telecined at playback by the dvd player, so the end result is the same but you physically have to store alot less data on the disk, so the encode is much better. NTSC vs NTSCfilm only affects the video. The audio stream for both is identical because the difference in the video's framerate is due to new frames created from duplicated fields. So there are just more frames played each second, but there are still the same number of seconds...thus the audio syncs with video regardless of which way it is stored.

My first choice for someone in your situation would be to just get a cheap Apex player that supports PAL. If this is not an option I would personally just make an NTSC SVCD since it will require less of a change in resolution. But you've got to burn on cdr media to be guaranteed proper playback. If you choose to use DVD then resize to 352x480.

[lenti_75]

convert it to 23.9 fps...the quality is a lot better, and it is compatible.
and if the input is interlaced, kepp it interlaced.

i use Canopus porcoder to convert, i'm not an expert in tmpenc

anyway, keep a low resolution 352x480, and high bitrate.

[lordsmurf]

convert it to 23.9 fps...the quality is a lot better, and it is compatible.
and if the input is interlaced, kepp it interlaced.

i use Canopus porcoder to convert, i'm not an expert in tmpenc

anyway, keep a low resolution 352x480, and high bitrate.

With Procoder, how do you make it so the 29.97 playback is set for the 23.976 souce? It has no option like TMPGENC.

[dphirschler]

Make sure your footage is film. If it is video (25i), then you should probably do a 29.97i conversion instead of a 23.976p.


Darryl

[BCAESE]

OK Adam I understand thanks

The reason why I don't burn on CD-R because I can't bother with 2 discs.
I have had pretty decent success with encoding at 23.976 but I was going to try 29.97 for compatiblity because I know DVD player playback everything at 29.97fps. I'll stick with 23.976 (play it safe)

If I encoded @ 23.976 352x480 would this get rid of the double picture
Using TMPGEnc, if you choose NTSCfilm 23.976 can you adjust the default size of 480x480 to 352x480 I don't know how

Is 352x480 a compliant DVD format

Your help is great

[adam]

I'm assuming that your double picture problem was due to using a non-compliant DVD resolution (480x480.) If so, then yes using a DVD compliant resolution will solve that problem. The NTSC DVD compliant resolutions are 352x240, 352x480, 704x480 and 720x480. You generally don't ever want to have to upscale your resolution by too much, so a 480x480 source should really only be transferred to DVD at 352x480.

In TMPGenc, after loading any template you then have to load the unlock template to make certain settings configurable. I would do this, load the NTSC DVD template then load the unlock template. On the video tab set the resolution to 352x480, framerate to 23.976fps (internally 29.97fps) and enable the 3:2 pulldown while playback option. On the advanced tab enable the "do not framerate conversion" tab. With this enabled TMPGEnc will slow your video down by 4.03% to 23.976fps. With this option unchecked TMPGenc will just delete about 1 frame every second...which will lead to jumpy video.

As someone else mentioned, a PAL interlaced source cannot just be slowed down to NTSCFilm. You'd most likely have to deinterlace first, so for DVD output you'd be better off converting to 29.97fps interlaced. But for SVCDs I would argue that deinterlacing is the better option. 29.97fps interlaced even at 2.6mbits is simply no good for most sources. I'd much rather have the added compressibility of a progessive source, even after a field deinterlace. Of course this is entirely subjective.

Luckily, most PAL sources are progressive, or so I've been told.

[BCAESE]

Excellect you answered everything I asked

You said most PAL files are progressive, Therefore I must ignore you last paragraph If I have a PAL file that is interlaced then I must apply you last paragraph

Would I select top field A or B

[FOO]

I can't seem to remember the reason that interlaced PAL
needs to be treated differently than Progressive.
Whats different about applying 3:2 pulldown to interlaced
source whatever the original framerate ? I realize the
fields are separated in time , but I don't know why that
hurts anything.

[adam]

Like I said, its a subject thing. An interlaced source is best left interlaced, no doubt. And if you were converting to DVD I would keep it interlaced. To do this I would actually use a separate program like Avisynth. I think if you tried to do this in TMPGenc you would get very poor results.

But interlaced sources require more bitrate, and so do 29.97fps sources compared to 23.976fps sources. For DVD its not a problem, but SVCD has such a limited max bitrate. I personally think you'd get better results deinterlacing and treating it like a regular progressive clip. Just my opinion, and since it seems you are going the DVD route, you can just ignore all of this.

If the clip is progressive it doesn't matter what field order you select since there are no fields. If it is interlaced then double click on the deinterlace filter and set it to odd/even field (field). Scroll through the movie. If it looks ok, your field order is correct. If its jumpy your field order is wrong. Change it and then make sure you disable the deinterlace filter.

[adam]

I can't seem to remember the reason that interlaced PAL
needs to be treated differently than Progressive.
Whats different about applying 3:2 pulldown to interlaced
source whatever the original framerate ? I realize the
fields are separated in time , but I don't know why that
hurts anything.

I agree, and I've done exactly that, slow 25fps interlaced down to 23.976fps interlaced and played it back via pulldown...worked great. It should too, because what's the first thing that a decoder does before telecining the NTSCfilm clip to NTSC??? It interlaces it of course, or more specifically simply separates the fields...technically the stream is always interlaced.

But for whatever reason, the DVD standard requires 23.976fps material to be encoded using frame pictures. Its mentioned in DVD Demystified and Scenarist spits the sources out when I try 23.976fps interlaced w/pulldown flags. I think the same is true for SVCD too, but I don't know.

[BCAESE]

OK since I am going the DVD route selecting the top field A or B doesn't matter! Knowledge

I will what you said and I'm sure to have success

Thanks

[FOO]

In that case, interlacing can consist of just merging the fields
instead of merging and filtering to remove mice teeth , etc
because as you say , they will just take it apart again.
I think most people think of de-interlacing as making
each resulting frame look good by itself , which would
actually be undesirable in PAL to NTSC and take longer

[dphirschler]

FOO, your method is interesting. However, the problem I have with it is the slowdown. Slowing down from 25 fps to 23.976 would be only a slight slowdown, but still very noticeable esepcially on a music concert (many of which are shot on video).

Another possible problem is after it is encoded 23.976, will it not play back at 29.97 fps and telecine it? That would introduce some slight stuttering, but even more noticeable if you leave the frames interlaced.


Darryl

[adam]

dphirschler: Remember that most PAL sources were filmed at 24fps to begin with. When you convert to 23.976fps most of the time you are just undoing the previous conversion to PAL, and literally nothing is lost by way of this speed change. Also the difference between FILM and PAL is so slight, that it almost never causes problems, and when they do occur its pretty much confined to the audio due to the pitch change. (check out the R2 release of Pink Floyd's the Wall, for example.) But this is a problem inherant in the PAL standard and only acceptable because its SOOOO rare. If your source is such that it causes problems due to this speed change, then 3/4ths of the rest of the world is experiencing it too, so at least you're in good company.

If done properly, the telecine will be no worse then a normal one. All mpeg sequences are interlaced, in that they are made up of fields. When we think of a progressive clip that just means it uses frame pictures, which means both fields are displayed at the same time. When we think of an interlaced clip that just means the two fields are played in sequence. For NTSCfilm material, during playback the fields are split, some are duplicated, and then they are all played in sequence. The interlaced clip would basically just be one step ahead in the process. Regardless of whether the source is stored as progressive or interlaced, it undergoes the same process. I'm sure there are very practical reasons why our standards require frame pictures for telecining, I just don't know what they are.

[FOO]

I know one reason. According to the MPEG2 Spec, the
pulldown flags are not recognized unless the Progressive
flag is set. Believe it or not , I think that has nothing to do with
whether the video is stored as 1 frame or 2 fields. That
flag refers to how it is to be displayed and changes meaning
if the TFF RFF flags are set

[dphirschler]

dphirschler: Remember that most PAL sources were filmed at 24fps to begin with.

I am talking about the ones that don't fit into that category... PAL 25i, or in other words PAL video. I know with progrssive PAL (ie a film) PAL just speeds it up 4% to 25 fps. NTSC telecines it like so:

Frames ABCD

ABBCD
abcdd

or: AaBbBcCdDd

Notice the C and D frames come out of sequence. It looks relatively smooth on an NTSC set.

But if you have two fields in each "frame" (instead of a truely progressive frame) you will bring forth a problem. Because each field represents a different moment in time, you will get a severe stuttering effect because you will be showing moments in time out of sequence. This is not a problem if it is progressive, but a big problem if it is interlaced. The best solution to this is to deinterlace the "frames" so that they are truely progressive.

But then you introduce a 4% slowdown when it gets telecined. How best to handle this? You could delete one out of every five frames, but that would introduce a slight skip.

I suggest instead to use the AVIsynth script to convert PAL video to NTSC video if it is 25i video. If it is 25p film then telecine it.


Darryl

[adam]

Well that's a very narrow category indeed. You are not just talking about interlaced PAL clips but pure interlaced PAL clips, like DV footage for example. No I wouldn't recommend trying to convert this to NTSCfilm under any circumstances.

[dphirschler]

I am not sure how rare it is since I live in the US. But it stands to reason that a lot of TV shows, sports broadcasts, live music events, and home video footage in Europe would fit into this category.

For instance, my experience with this is a European live broadcast of a U2 concert.


Darryl

[BCAESE]

Well Adam did what you said as far as encoding to DVD NTSC and making the adjustments to the the fps and the resolution (23.976, 352x480 & 3:2 pulldown)

I put in my very picky DVD player and everything is fine! No double picture and did not lose much quality!

However I noticed by the later part of the movie there was an audio sync problem I think this is due to the fact that I merged my mpegs with TMPGEnc before encoding it. The only reason why I did it is because when authorizing with DVDLab, sometimes the movies would not link properly and the movie would pause and not go to the next mpeg.
Also merging is alot easier saves you from repeating the process.

Is there a better merging tool? I have merged my mpegs before everything was fine. I guess it depends on the mpeg??

What do you guys think

[adam]

dphirschler anything filmed that way would be problematic. You're right, this would apply to various types of broadcasts of live events and such, but a typical movie broadcast on tv generally wouldn't be pure interlaced. Anyway, regional format conversions are always going to be somewhat problematic with any kind of interlaced source.

BCASE, merging in TMPGenc, or any other mpeg joiner for that matter, often cause sync problems. You may try other mpeg joiners but I've never found one that works perfectly. Mpeg joing is just messy business. Since you have to frameserve your sources anyway, in the future it would be best to just frameserve everything together, then later joining isn't necessary. Just load each of your mpg clips into dvd2avi in succession and then create the d2v project.

[BCAESE]

Are you saying you can load both video sources together in DVD2AVI and create a merged .d2v file

If that is the case what do I do with my audio? I BeSweet the audio, there will be 2 separate files. How do I join them with my new merged .d2v file and create a perfect sync

[adam]

Yes you can load as many sources as you want into DVD2avi and it will frameserve them all in succession. You could also take two separate d2v sources and frameserve them in succession into your encoder via Avisynth.

Have DVD2AVI decode the audio. It will output one long wav file which you can then encode in BeSweet. Or if for some reason you wanted to decode your audio to wav with something else, you could easily merge the two wav tracks using any wav editor. Wav can merge flawlessly, unlike mpeg.

[BCAESE]

Adam

So DVD2AVI will merge my original 2 audio files into 1 wav file.
Then I can use BeSweet to to encode the wav file to 23.976 mp2?
Then use TMPGEnc to merge my new .d2v and my new mp2?

Will this give a nicely synced NTSC mpeg