NTSC video with a film-like/PAL to NTSC conversion type of look that shouldn't

[Bix]

For the last 2 months on WWE's Classics on Demand VOD service, much of the content has looked like a PAL conversion to varying degrees of severity. It's not on my end as it's been discussed in various places online by customers of many different cable companies across the US. It's especially frustrating given how much of the content is rare and largely from the era before DVD recorders and high res capturing.

My questions are:

- How could this have happened?
- Is there any way to fix it on my end in software after capturing?

I can upload a sample clip if needed.

Thanks!

[manono]

I can upload a sample clip if needed.

We shouldn't even have to ask for one. No one wants to guess how to treat this mystery video.

[Bix]

No need to be like that, I thought that it was pretty self-explanatory given that said "look" is pretty universal among PAL to NTSC conversions and video treated to look like film. It's not like I said "THERE ARE THINGIES IN THE PICTURE!!!!!!!!!!"

I'll upload a clip tomorrow.

[manono]

No need to be like that, I thought that it was pretty self-explanatory given that said "look" is pretty universal among PAL to NTSC conversions and video treated to look like film.

It wasn't self-explanatory at all. If you knew what it was and what to do about it you wouldn't be asking for help. Since you don't know the problem, why should we rely on your description of what's wrong with it? Your description' doesn't help at all.

And I hope you know some AviSynth. Otherwise you may not be able to carry out any fixes proposed.

[Bix]

No need to be like that, I thought that it was pretty self-explanatory given that said "look" is pretty universal among PAL to NTSC conversions and video treated to look like film.

It wasn't self-explanatory at all. If you knew what it was and what to do about it you wouldn't be asking for help. Since you don't know the problem, why should we rely on your description of what's wrong with it? Your description' doesn't help at all.

There's a difference between knowing what a problem looks like and knowing the cause, plus you're being a jerk about it for no apparent reason. You could've just said "Yes, please post a clip" instead of "GRRRRRRRRRRRRR! STUPIDITY! HOW COULD I KNOW WHAT HE'S TALKING ABOUT? IT'S NOT LIKE PAL TO NTSC CONVERSIONS ALL LOOK LIKE...wait..." Plus look at the time stamp, it was late and I didn't have time to post a clip, but I wanted to get the post up while I still had it in my head in case anyone knew what was up.

Are you honestly telling me that you frequent this forum and have no idea what I'm talking about when I say something looks like a PAL conversion? It looks like the frames are blended together and for whatever reason it's been happening to almost all of the programming on a single SVOD channel for weeks.

Anyway: http://www.mediafire.com/?30nln0jnd2k

It may be more subtle than the some of the other shows up now but this was the easiest for me to most now and it's definitely there.

[manono]

Is this how it was broadcast? I don't think so. I don't cap, but as far as I know no channel broadcasts 640x480 XviD AVIs with crappy MP3 audio. When someone asks for a sample they mean a source sample, not something you or someone else has already reencoded..

It's not a conversion from PAL. It's just deinterlaced NTSC video from some kind of analog source, maybe VHS tapes, but since I don't have the source broadcast I can't be much more specific.

Are you honestly telling me that you frequent this forum and have no idea what I'm talking about when I say something looks like a PAL conversion?

Hardly. I'm honestly telling you that I won't believe you and won't offer advice based on your statement without a sample to prove it. Or, in this case, to disprove it. Although it's not 100% true, in general, programs originating in the US aren't rebroadcast via PAL2NTSC field-blended conversions.

[Bix]

1. I've seen plenty of people here post compressed MPEG4 samples from their copies (plus those who asked for help without samples and got it happily) have them accepted with open arms, so now you're just going out of your way to be unpleasant.
2. I know it's not a conversion from PAL. I said that. I said that there was some sort of problem on the source end. I said that that regardless of what the problem is, that's what it looks similar to when it shouldn't.
3. How exactly am I supposed to post the "original source footage" program from a cable video on demand service? I don't know if you're in North America or not, but this sort of material, while being streamed digitally from the content provider to the cable company to my cable box, is not offered in a way where the stream can be ripped directly. Anything I post would be re-encoded since it's only available at the cable box's standard outputs. As far as I know, only regular channels can be captured over Firewire, not video on demand. Plus, even for what DOES work, it varies by the cable provider, and my box doesn't have a firewire port in the first place. So if you're ranting because I didn't post an original broadcast capture, that's because to the best of my knowledge, it's not possible. If you mean a few seconds of interlaced DVD or lossless AVI, just say it.

[poisondeathray]

Maybe you can point out specifically what you are getting at, because I didn't see any clear blends in the processed xvid sample (as in ghosting or blended fields). There was some haloing , noise, deinterlacing artifacts, and overall low quality - were any of these what you were getting at?

An unprocessed sample might help (i.e whatever you captured)

[Bix]

As I said, this is a more subtle case, but here's a clip from the DVD I made on my Pioneer recorder (DVD Shrink re-author mode "No Compression"): http://www.mediafire.com/?mywyazymnow

I can also provide a sample from another program in a little while.

[manono]

1. I've seen plenty of people here post compressed MPEG4 samples from their copies (plus those who asked for help without samples and got it happily) have them accepted with open arms, so now you're just going out of your way to be unpleasant.

Not true at all, although you certainly make it inviting. One can't offer informed advice without a decent sample. It's standard practice.

2. I know it's not a conversion from PAL.

Oh? Then why mention PAL2NTSC conversions in each of your posts so far? It doesn't even 'look' like one. It's not the least bit blurry or jerky playing. It just has that 'analog' look.

3. How exactly am I supposed to post the "original source footage" program from a cable video on demand service?

Beats me. I said I don't cap. However, I'm sure you could have capped something much closer to the source than a poorly deinterlaced XviD AVI. Even DV AVI would be better - much better. Or the MPEG, if it was broadcast as MPEG. What you showed us is far removed from the source.

[Bix]

Taking some more looks at it now, and it should be burned and watched on a TV as for whatever reason it's not noticeable on a PC on the original DVD sample.

Anyway:

- After taking a few peaks on the first page, I saw 2 posts where people asked for help without posting a sample and got it. I'm sure I could find a lot more. As far as MPEG-4 samples being posted here without the OP being yelled at, here's one. It's a bad idea to declare something like that when searching the board is so easy.
- I said it looks like a PAL to NTSC conversion, not that it is a PAL to NTSC conversion. By itself, I can see why that could read confusingly, except for how the subject says that it's an NTSC video that has that type of look but shouldn't. As far as whether it does or doesn't, see above Re: PC vs TV.

[edDV]

Bix, how did you capture this and do you have the capture file? The file you used to make the xvid.

That is what is needed to answer your questions. If a cable VOD capture, it should be 720x480i and not deinterlaced.

[Bix]

That's what I uploaded the second time, the 720x480i DVD capture. That's the best I can do.

[poisondeathray]

That's what I uploaded the second time, the 720x480i DVD capture. That's the best I can do.

Was this from a DVD recorder, that you used on a VOD stream? The 2nd file you uploaded is not interlaced; it's progressive content . Maybe your recorder applied some post processing?

- After taking a few peaks on the first page, I saw 2 posts where people asked for help without posting a sample and got it. I'm sure I could find a lot more. As far as MPEG-4 samples being posted here without the OP being yelled at, here's one.

The reason why it's important to examine the original (or as "original as possible") source, is that most deblend filters work on individual fields to get rid of blends/ghosting. Your title suggested it had something to do with that, hence the insistence on an unprocessed sample. Also, whatever process you might have done may have caused the issue in the first place. For other types of issues, sometimes a process xvid sample is sufficient. As manono mentioned, blends and jerks are common signs of the poorly done PAL<=>NTSC conversions, but I still don't see any blends... So I'll ask again: can you clarify what you're getting at or trying to do?

[Bix]

Yeah, I figured out the issue with the Xvid in this case last night, thanks for kindly explaining that in detail. Anyway...

That second link, the ZIP of a VIDEO_TS folder, was from the recorder that I used to record the VOD from my box via composite. MediaInfo reports it as interlaced (top field first) so I'm not sure why you think it's progressive.

Here's a sample from another show (Same recorder): http://www.mediafire.com/?r3tmqtz3fzt

When I was watching it with a friend who doesn't know much at all about this stuff, he immediately remarked that it looked like film, as if someone post-processed it to look more like film than video. Being a studio wrestling show, he knew it had to be shot on videotape, but when he walked in the room and looked at the TV, he asked me what was up with it looking like film. MediaInfo sees VTS_01_1.VOB as interlaced, just like the first sample.

Like the first clip, the differences only seem to be noticeable on an interlaced TV display, so I suggest burning them to an RW. In a few minutes, I'll post a good/normal/proper clip from a few months ago on the same service for comparison.

[poisondeathray]

That second link, the ZIP of a VIDEO_TS folder, was from the recorder that I used to record the VOD from my box via composite. MediaInfo reports it as interlaced (top field first) so I'm not sure why you think it's progressive.

It's progressive because the content is progressive. There are no separate fields, and the temporal resolution is 30 samples/sec. An interlaced 30i stream would be 60 sample/sec. Mediainfo just reports the stream header and may or may not be correct. (it's incorrect in this case).

I don't have much experience with recording VOD, or what process you used, but that might have something to do with it

When I was watching it with a friend who doesn't know much at all about this stuff, he immediately remarked that it looked like film, as if someone post-processed it to look more like film than video. Being a studio wrestling show, he knew it had to be shot on videotape, but when he walked in the room and looked at the TV, he asked me what was up with it looking like film. MediaInfo sees VTS_01_1.VOB as interlaced, just like the first sample.

the term "looks like film" can suggest many many different things, from speed (judder) at 24p , to grain /noise, to gamma response, to depth of field , etc...

[Bix]

That second link, the ZIP of a VIDEO_TS folder, was from the recorder that I used to record the VOD from my box via composite. MediaInfo reports it as interlaced (top field first) so I'm not sure why you think it's progressive.

It's progressive because the content is progressive. There are no separate fields, and the temporal resolution is 30 samples/sec. An interlaced 30i stream would be 60 sample/sec. Mediainfo just reports the stream header and may or may not be correct. (it's incorrect in this case)

Gotcha. What tools show the sample rate?

I'm usually good at eyeballing interlacing or lack thereof in a program that doesn't post-process it, so I opened it in VirtualdubMod (I watch my DVDs in Media Player Classic, which automatically deinterlaces), and yup, the clips I posted so far definitely looks progressive while the "normal" sample (linked below) definitely looks interlaced. So I guess my recorder detects the source and encodes accordingly but always writes the header as interlaced? So I guess the footage was encoded by WWE as progressive. Could this be (at least part of) the problem?

When I was watching it with a friend who doesn't know much at all about this stuff, he immediately remarked that it looked like film, as if someone post-processed it to look more like film than video. Being a studio wrestling show, he knew it had to be shot on videotape, but when he walked in the room and looked at the TV, he asked me what was up with it looking like film. MediaInfo sees VTS_01_1.VOB as interlaced, just like the first sample.

the term "looks like film" can suggest many many different things, from speed (judder) at 24p , to grain /noise, to gamma response, to depth of field , etc...

Speed issue or at least the appearance thereof. Hence the PAL->NTSC comparison.

Here's a clip that was shown on the service in October (same DVD recorder): http://www.mediafire.com/?ccmou0iniho

This looks as it should on my TV and as I said, looks to be interlaced.

Thanks for your help.

[poisondeathray]

No automated tool can tell you. They all make mistakes.

The only reliable way is to manually determine it. Basically you separate the fields, and if consecutive fields are taken from different moments in time (i.e they are different), it's interlaced. If consecutive filelds are from the same moment in time, it's progressive

Here is a mini-guide:
http://neuron2.net/faq.html#analysis

Your latest sample is interlaced, so it will show 60 moments in time per second vs. 30. So it will look a lot smoother when played in a regular DVD player, or you use a bob-deinterlacing player on a PC. A 30p video will look a lot "jerkier" because in 1 second only 30 moments in time per second are represented. So I can see why your friend made those comments. However, the "30p look" is still quite different from the "24p cinematic film" look, the latter is noticably more "jerkier"

You cannot make those 30p videos into 30i , because you only have 30 moments in time per second represented. You need a 60p source to do this.

So your 1st 3 videos had been single rate deinterlaced to 30p , not bob-deinterlaced to 60p . There might be something to do with the way you recorded this, not sure. It might have been broadcast like that (30p content in 30i wrapper, or 2:2 pulldown) , but I'm not sure

[Bix]

No automated tool can tell you. They all make mistakes.

The only reliable way is to manually determine it. Basically you separate the fields, and if consecutive fields are taken from different moments in time (i.e they are different), it's interlaced. If consecutive filelds are the same moment in time, it's progressive

Here is a mini-guide:
http://neuron2.net/faq.html#analysis

Your latest sample is interlaced, so it will show 60 moments in time per second vs. 30. So it will look a lot smoother when played in a regular DVD player, or you use a bob-deinterlacing player on a PC. A 30p video will look a lot "jerkier" because in 1 second only 30 moments in time are represented. So I can see why your friend made those comments. However, the "30p look" is still very different from the 24p cinematic film look

Right, that makes sense then. I've noticed with content shot on video like wrestling (and especially with the constant movement in wrestling) that if you take a progressive video and encode it to DVD, it looks a lot like this, so now it all adds up.

You cannot make the 30p videos into 30i , because you only have 30 moments in time per second represented.

This I know, but at least now I know what the problem was in the first place. There's no reason that they'd be doing this intentionally, is there (especially since some of the videos still look normal, which I now know means interlaced as the source is)?

Thanks!

[poisondeathray]

I'm not sure why they would be doing that. It might be that they fubared the source or made some mistakes. I don't know too much about broadcast, edDV would know more. I *think* the 30p in 30i signal is compliant with broadcast (2:2 pulldown), but not sure. So it might be possible that those videos were broadcast like that

Did you use the exact same configuration and setup for all captures? (it's possible that the method in which you captured or recorded processed it in some fashion with the 1st 3 videos)

[Bix]

Yup, all the same, and this is happening to all subscribers, so WWE encoding content progressively instead of interlaced explains it.

[poisondeathray]

Well then a Flying Elbow or DDT is in order for those WWE content programmers!

[manono]

Yup, all the same, and this is happening to all subscribers, so WWE encoding content progressively instead of interlaced explains it.

You have to understand that it was encoded as interlaced, the second sample was. That's why MediaInfo told you it was interlaced. And if that's how it was broadcast, which by the evidence of the third sample being interlaced, it really is how it was broadcast, then the difference is that the 2nd sample has deinterlaced and progressive content while the third one has interlaced content. There's nothing really wrong with that - encoding progressive content as interlaced. Perhaps 99% of PAL movies on DVD, progressive 25fps - are encoded as interlaced. It's not really such a good idea to deinterlace something shot on video because you lose half of the motion fluidity and get deinterlacing artifacts. Maybe that's what made you think it was film-like.

Do you really think anyone could have gotten to the bottom this without a decent sample to study? Do you think anyone at all could have diagnosed the problem (if there really is a problem) based on your first post alone? Especially after you threw us the red herring about PAL to NTSC? Good work, pdr.

[Bix]

Yup, all the same, and this is happening to all subscribers, so WWE encoding content progressively instead of interlaced explains it.

You have to understand that it was encoded as interlaced, the second sample was. That's why MediaInfo told you it was interlaced. And if that's how it was broadcast, which by the evidence of the third sample being interlaced, it really is how it was broadcast, then the difference is that the 2nd sample has deinterlaced and progressive content while the third one has interlaced content. There's nothing really wrong with that - encoding progressive content as interlaced. Perhaps 99% of PAL movies on DVD, progressive 25fps - are encoded as interlaced. It's not really such a good idea to deinterlace something shot on video because you lose half of the motion fluidity and get deinterlacing artifacts. Maybe that's what made you think it was film-like.

Well, as I said, I can't post the original "broadcast" VOD service encode sent to my cable box because it's impossible to grab it. The best I can post is my capture, the DVD recording, which is what I posted after the Xvid sample. When I'm talking about progressive scan, I'm talking about the file used for the VOD service. The point, which poisondeathray seems to agree on, is that WWE is taking interlaced material (their DV encoded captures of various interlaced broadcast analog tape formats) and is encoding it as progressive scan instead of interlaced when creating the files that are sent to the cable company. That's what I was looking to find out, and clearly it's not something that can be fixed on my end, so kudos to poisondeathray for figuring it out.

Do you really think anyone could have gotten to the bottom this without a decent sample to study? Do you think anyone at all could have diagnosed the problem (if there really is a problem) based on your first post alone? Especially after you threw us the red herring about PAL to NTSC? Good work, pdr.

It's not a red herring because that's what it looks closest to in my eyes and in the other forums where I post about something like this, people would've understood it, because they all said the same thing when they first saw it. In understand why you didn't, but it's not my fault that you didn't understand what it meant. Plenty of people refer to it that way because that's their best frame of reference for jerky video, even if they don't know about PAL and NTSC (people saying that something looks like it's from a European TV show is another one that I've heard plenty of times, including a TV show host snarking that a jerky Mariah Carey video looked like "cheap Italian porn"). I've heard it so much that I was honestly shocked that someone here wouldn't know what I meant.

Plus, I don't know why you're still hung up on the idea that I said it actually WAS PAL->NTSC when I didn't.

When plenty of collectors and average consumers that don't know much about the technical end other than that the stuff they have from Europe looks similar all seem to know what each other are talking about, why would I expect people at an experts' forum to NOT know what I was talking about?

As far as the sample, I made it clear that I would post one if needed and just wasn't able to at the moment. Why not just be civil and say something like "I'm not sure what you mean so please post a sample" and then "For motion problems the original capture is best since a re-encode could introduce interlacing issues"? It's not hard. Poisondeathray was able to be perfectly polite. You tried to play internet tough guy, and you failed spectacularly. Also, you get negative bonus points for insisting that there's a good chance that there's no problem ([sarcasm]Yes, I'm sure that a VOD provider would intentionally and inconsistently send 30p video that's clearly problematic after sending 30i video for its entire run, including some of the programs available right now, including 2-parters where Part 1 is one way and Part 2 is the other.[/sarcasm]) and then congratulating poisondeathray for finding the problem.

VideoHelp is generally a wonderful, civil forum. Don't ruin it.

[manono]

Well, as I said, I can't post the original "broadcast" VOD service encode sent to my cable box because it's impossible to grab it. The best I can post is my capture, the DVD recording, which is what I posted after the Xvid sample. When I'm talking about progressive scan, I'm talking about the file used for the VOD service. The point, which poisondeathray seems to agree on, is that WWE is taking interlaced material (their DV encoded captures of various interlaced broadcast analog tape formats) and is encoding it as progressive scan instead of interlaced when creating the files that are sent to the cable company.

No, just the opposite. They're taking a progressive source (the deinterlaced video from the second sample) and encoding it as interlaced. I'm reasonably convinced after you posted 2 videos, one deinterlaced and progressive and the other interlaced, which you say were capped using the exact same settings, that both caps are pretty true to the source. Any app with which you wish to check it, MediaInfo, GSpot, ReStream, DGIndex, etc., will confirm that it was encoded as interlaced. In fact, almost all progressive 29.97fps sources, whether on DVD or broadcast, are encoded as interlaced.

I'll ignore the rest of your comments.

[jagabo]

As I said, this is a more subtle case, but here's a clip from the DVD I made on my Pioneer recorder (DVD Shrink re-author mode "No Compression"): http://www.mediafire.com/?mywyazymnow

That was originally 30i video (60 fields per second). It was deinterlaced before transmission (maybe because they also make this footage available as digital files for computer viewing?) to 30p (30 frames per second). DVD recorders always record analog inputs in interlaced mode because that's what comes over the wire (composite and s-video are always transmitted one field at a time). It doesn't matter (to the DVD recorder) that pairs of fields from the analog input were from the same point in time.

[2Bdecided]

It would be strange for 30p (e.g. web content!) to be used for 60i broadcast in place the original 60i content. Maybe it's a mistake, or maybe it's intentional - lots of broadcasters in the UK try to make video look like film as a "special effect" by dropping every other field (or doing something slightly better).

Cheers,
David.

[Bix]

As I said, this is a more subtle case, but here's a clip from the DVD I made on my Pioneer recorder (DVD Shrink re-author mode "No Compression"): http://www.mediafire.com/?mywyazymnow

That was originally 30i video (60 fields per second). It was deinterlaced before transmission (maybe because they also make this footage available as digital files for computer viewing?) to 30p (30 frames per second). DVD recorders always record analog inputs in interlaced mode because that's what comes over the wire (composite and s-video are always transmitted one field at a time). It doesn't matter (to the DVD recorder) that pairs of fields from the analog input were from the same point in time.

That's what I thought.

It would be strange for 30p (e.g. web content!) to be used for 60i broadcast in place the original 60i content. Maybe it's a mistake, or maybe it's intentional - lots of broadcasters in the UK try to make video look like film as a "special effect" by dropping every other field (or doing something slightly better).

Since it's an American service and being applied inconsistently, I figure that it has to be a mistake.

As for manono: You seem to have trouble grasping what I'm saying. I know that the 80,000+ hours of DV encoded footage (from formats like 1", 2", 3/4", and Betacam SP) sitting on WWE's servers in Connecticut are interlaced. I know that my DVD recorder that captured the samples always records interlaced. I had been trying to figure out what made some of the programming on WWE's subscription video on demand service look jerky, and it was determined that the video was deinterlaced in the production process, possibly when they encoded the MPEG-2 files that are then sent to cable companies across the US to be stored on their Video on Demand servers. I then recorded the video from my cable box outputs to my DVD recorder inputs via analog composite video and analog RCA stereo audio. The samples being posted are from my DVD recorder, not a broadcast capture, since it is impossible to do a direct stream capture of this type of programming. For the rest, see jagabo's post.

[manono]

I don't see where anything jagabo said contradicts anything I said just before. In fact, it seems to confirm it, as near as I can tell. The only difference is that your cables by definition always cap as interlaced and I'm also saying that it was really encoded and broadcast as interlaced.

[Bix]

I didn't say jagabo was contradicting you, just that his post explained the rest of what I was trying to say about the entire chain from WWE to consumer.

You haven't explained why you think there wasn't any deinterlacing in the entire chain of events, which is, at the very least WWE loading the DV video into an editing suite, adding the new intro/outro/bug graphic, outputting that to DV, and then encoding to MPEG-2, but could very well include more steps. Maybe in the editing suite before the final WWE encode for delivery to the cable company, maybe in the final encoding itself, but it happened somewhere before it got to my (and every other subscriber's) cable box. Everyone else seems to agree that it was the problem, which was seen by every WWE Classics on Demand subscriber and was not introduced by my DVD recording. It fits the problem perfectly and is noticeable if you examine the video in VirtualDubMod and other programs, while the "normal" sample clearly maintained its interlacing every step of the way. What explains that discrepancy if it wasn't deinterlaced at some point in the chain?