I don't understand the hype for 1080p/24 output

[Bohdy]

You mean as shown in this image?...
That's not exactly the same as 24 fps though.

Sorry, I changed my post because I think I misunderstood your question. As for that image, it just shows how 3:2 pulldown is done by the player to convert 24p into 60i. A TV with Inverse Telecine capability can reverse it losslessly. But not all of them can sync the frames at a perfect 24hz to prevent judder, apparently.

In before someone starts talking about compression nonsense.

 

[Shifty Geezer]

Sorry, I changed my post because I think I misunderstood your question.

Okay. So 1080p24 will be a better quality thing, and not just a spec-sheet tick-box, when it comes into effect.

Going one step further, how do the sets synchronize? Could you get an LCD type display that can sync to 60 fps and any lower figure, even 3 fps? Perhaps a signal tells them to refresh? Sets that could do that would be great for gaming as you could have variable framerate (okay, better for gaming, seeing as great for gaming is fixed, high framerates!) without tearing.

 

[Bohdy]

Okay. So 1080p24 will be a better quality thing, and not just a spec-sheet tick-box, when it comes into effect.

Well as LB said, the point of 1080p24 input is to be able to display 24fps film without judder. But the point I was making is that it is possible to have the same effect with same movie using 1080i60 input as well, depending on the set. Though it might be a bit of a moot point, as the sets that support one seem to support the other. Though it does show that it really depends on the set whether you have smooth 24fps, and not so much on the player having native 24p output.

 

[Gubbi]

You mean as shown in this image?...

That's not exactly the same as 24 fps though.

He means like this:

You get timeskewing of ½ 60Hz frame (8ms).

Cheers

 

[Ventresca]

More information about the Pioneer set:
http://www.ultimateavmag.com/flatpaneldisplays/1106piofhd1/index2.html

The result can be very nice, but a 24p source on a 24p display would always be the best solution...but i guess here we agree to disagree.

I still don't suppose Ventresca will admit that he/she was wrong, though.

Well this is is what happnes when i deal with mpeg2hd, i wish it was just an idea i had.
The same poblem explained in my link for SD Dv 24p , is here for HDV mpeg2 24p , thats why hdv cameras have the 24p advanced pulldown options too.
An hdv camera record a 1080 mpeg2 24p singal and store it on a normal 60i dv tape inserting the standard 3:2 pulldown, or, if you choise to, an advanced pulldwon.
When i have need to extract the original 24p frames , the only way to avoid any degradation is by the 2:3:3:2 advanced pulldown where the C frame is not splitted and can be recovered as first generation frame.
But the the advanced pulldown produce a non standard video not displayable on a normal Tv...it a pain in the ass to deal with.
I wish this was not the case, no loss, no degradation, no problems at all, i would go all the way with standard pulldown and it would make all the hdv post easier

Btw, LB: The point you are missing is that there are already sets around that give you identical quality and lack of judder as those that will accept 1080p/24p.

I guess all the people on the avs forum waithing for a 24p capable display have a different opinion.

 

[Bohdy]

I am not debating there is no loss in quality if the 3:2 pulldown removal is made with an uncompressed singal, i am debating this give the same quality result as having a real 24p stream directly from the source.
The television woking with an uncompressed 60i stream dont have the same recovering quality as a device receiving the original mpeg2stream with flags

Can you explain why it wouldn't be possible? As you can see from 3:2 Pulldown diagrams, all of the original fields are still in the stream after pulldown and arranged in a standard, fixed pattern. It's not a difficult prospect for the TV to be able to get back the original 24p frames if there has been no deflicker filtering applied.

I simply can't because this is is what happnes in the real life when i work with mpeg2hd, i really wish it was just an idea i had...but it is not.
The same poblem that explained in my link for SD Dv 24p , is here for HDV mpeg2 24p , thats why hdv cameras have the 24p advanced pulldown options too.

I've tried to explain that these problems are specific to recording cameras which perform pulldown PRIOR to encoding. A few people have tried to tell you now, but when just playing say a DVD source there is no recompression nor any of these other issues because the frames have been compressed in linear matching order, with the pulldown being performed AFTER encoding.

I guess all the prople on the avs forum waithing for a 24p capable display have a different opinion.

Many Avsforum people tend to believe in hype and voodoo without really understanding how everything works on a technical level.

 

[Ventresca]

Can you explain why it wouldn't be possible? As you can see from 3:2 Pulldown diagrams, all of the original fields are still in the stream after pulldown and arranged in a standard, fixed pattern. It's not a difficult prospect for the TV to be able to get back the original 24p frames if there has been no deflicker filtering applied.

Not all the movies are made in the same way, you can't make a good IVTC of a source with no proper and consistant patterns to work with and there are movies with broken 3:2 pattern everywhere.

I've tried to explain that these problems are specific to recording cameras which perform pulldown PRIOR to encoding. A few people have tried to tell you now, but when just playing say a DVD source there is no recompression nor any of these other issues because the frames have been compressed in linear matching order, with the pulldown being performed AFTER encoding.

We alredy debated this point, and i alredy said that i agree in the case of IVTC on an un compressed stream, this give no loss of quality, the problem of working with an uncompressed stream with no flags is in case of a non standard 3:2 pulldown

Many Avsforum people tend to believe in hype and voodoo without really understanding how everything works on a technical level.

The people i am refering at in this case are pretty knowledgable people

 

[Bohdy]

That's right, basically.

Bust most (all?) Holywood movie releases use a standard pattern. Some TV shows, improper region conversions, obscure stuff, etc. can be a different story.

 

[Mintmaster]

Bohdy, I think most of us here are agreeing with you (Colourless, Gubbi, myself, etc). 1080i is a perfectly suitable signal standard to transmit the information of 1080p/24.

Ventresca, Bohdy is saying DVDs are encoded at 24fps progressive, and the players are generating the 60i output. If there are any artifacts, it's due to filtering by the DVD player. Simply disabling it on the DVD player, if such an option was available, would allow perfect, lossless reconstruction of 24p by the TV if it chose to do so. All that would be necessary is for the TV to recognize the telecine pattern, reassemble the frames, and change it's own display refresh rate to a multiple of 24.

Like you can see, in orderer to recover the C frame you have need to decompress the green and the magenta frames and recompress them into a new 24p frame, this process introduce a loss of quality.

The 'C' frame in your diagrams has all the information even though it is split across two 60i frames. Magenta and green simply refer to how a 60i TV treats the information, not how it would look on a 24p display device. Notice that the green frames have a B and a C. That green frame is the one that would look funny on a 60i TV. The green/magenta frame on the last line would have no artifacts whatsoever.

There is no decompressing of frames as you say. Information is coming one field at a time. "Compressing" the C frame is exactly the same as "compressing" the A frame from it's two fields.

Imagine a 5 page book with 8 different half-page paragraphs and 2 duplications. You want to pair these paragraphs together into 4 larger paragraphs in a new 4 page book. Sometimes both paragraphs in the pair are on the same page. Other times, you have to turn the page to see the two paragraphs. Obviously that doesn't make it any harder to pair them, right?

Judder is due to a TV not being able to change it's refresh rate. 60i formats are not at fault.

(BTW, ShiftyGeezer, just because a TV is reading a 60i signal doesn't necessarily mean its display must be 60Hz)

 

[DemoCoder]

Most highend TVs today have internal IVTC/scalers, like DCDi Faraouja chips. The fallacy of the non-benefits of direct 24p->24p mapping is the following:

1) the assumption that discs are encoded properly. Turns out that many aren't (DVD). That's why the HQV tests exist, because many DVDs today lack correct flagging, and many of the IVTC logic on players guesses wrong, mostly on non-standard cadences. If everything was flagged properly, and all IVTC's operated correctly, there would be no issues and no need to buy third party IVTC video processors or special DVD players. A $26 CyberHome DVD player won't do IVTC properly on all discs.

2) Not only do some players have poor implementations, but some progressive TVs have shitty IVTCs (and scalers) which is why IVTC/upscaling DVD players are needed. If a standard DVD player had perfect IVTC and TVs had quality scalers, it would be a moot point.

3) Audio sync issues. Many IVTC/scalers introduce delays which are not compensated for correctly, leading to either fixed audio delay, or an accumulated error over time.

It's not an issue of what's technically possible to do. Sure, perfect loss-free IVTC is possible. The reality is, many players, TVs, set top boxes, etc use cheap commodity chips, or proprietary techniques which introduce the potential for problems, which together,can harms the industry as a whole when consumers get a negative impression from buggy implementations.

I think systems should be engineered to KISS principles: "Keep it Simple Stupid!" This not only reduces the probability of error/bugs in implementation, but makes implementations cheapers. Since we are saddled with legacy, unfortunately, we can't eliminate the need to support player conversions. However, at the very least, we can provide a path to the future, just like we allow DVI/HDMI to carry progressive signals together. In reality, any progressive stream can be transmitted "losslessly" as an interlaced stream, but it's a pointless conversion only needed for older sets, so the display interconnect standards naturally support non-interlaced modes of transport.

In that regard, IMHO, raw 24p output should be a supported output mode for those with TVs that can support it. Sure, you could transmit it at 30p and let the 24p signal be recovered perfectly by a perfect IVTC HDTV, but why? If the set is capable of 24hz, 48hz, or other refresh rates, why not let the set avoid the overhead. It would reduce latency, as well as eliminate bandwidth required, which would lower bit-errors on the cable, as well as reducing bandwidth by 20% which might be important for future home video distribution products.

From a "purist" point of view, I think it's desirable, just like "pure" audio codecs and output. There are actually people who are sensitive to judder and latency believe it or not.

 

[Ventresca]

Ventresca, Bohdy is saying DVDs are encoded at 24fps progressive, and the players are generating the 60i output.If there are any artifacts, it's due to filtering by the DVD player. Simply disabling it on the DVD player, if such an option was available, would allow perfect, lossless reconstruction of 24p by the TV if it chose to do so. All that would be necessary is for the TV to recognize the telecine pattern, reassemble the frames, and change it's own display refresh rate to a multiple of 24.

I am not debating this point, but i dont know of any dvd release encoded as 24pfs progressive, and i tend to agree with what Fruitfrenzy said :

Despite what was stated earlier DVDs can not store 24fps video. They can however store NTSC format 30fps video the is encoded as progressive frames with the flags "repeat_first_field" and "top_field_first". The MPEG2 decode process includes the processing of these flags to generate a 30fps sequence from 24fps input. The sequence frame rate is defined as the rate at the output of the decoder which will be 30fps for a compliant decoder.

If a DVD player knows it is decoding to a progressive display it can choose to skip the final stage of the decode and leave the video at 24fps but this lies outside of the MPEG2 specification.

Although maybe some players can be capable of doing so, and i have heard of 24p dvd more than one time , from what i know all the dvd are encoded as interlaced.

There is no decompressing of frames as you say. Information is coming one field at a time. "Compressing" the C frame is exactly the same as "compressing" the A frame from it's two fields.

As stated in the article , in this particular case you must have need of doing decompression and recompression :

"The A, B, and D frames can be recovered by using two fields from the same 60i frame. The C frame cannot be; it is split across field 2 of the third (green) 60i frame and field 1 of the fourth (magenta) 60i frame. In an intraframe-compressed format like DV, that's an important distinction: the two fields of each 60i frame are compressed together, as a frame. If the two fields are very similar, as happens in the red, yellow, and blue frames, the compression uses a comparatively efficient "8x8 DCT" mode. If the fields are very different, as happens in the green and magenta frames when there's a lot of motion between the B, C, and D frames, a less efficient 2x4x8 DCT is used, possibly leading to more image degradation in those frames compared to their red, yellow, and blue companions.

Furthermore, the original A, B, and D frames can be copied in their compressed form from the 60i video data into a new 24p data file, but recovering the C frame requires decompressing the green and magenta frames and recompressing them into a new, 24p DV frame. That puts the C frame a generation down compared to A, B, and D."

 

[Bohdy]

Democoder: Latency is not as big of a problem with video as with gaming, as long as the audio is delayed by the same factor.

You make a good point about keeping it simple and not relying on the quirks of the TV, though. 24p output does have merit in that context, which I suppose resolves my question at the making of this topic--as long as it is clear that it's not "compulsory" for judder-free 24p display.

 

[Ventresca]

That's right, basically.

Bust most (all?) Holywood movie releases use a standard pattern. Some TV shows, improper region conversions, obscure stuff, etc. can be a different story.

Lots of hollywood movies are released without standard patterns.

 

[Ventresca]

From a "purist" point of view, I think it's desirable, just like "pure" audio codecs and output. There are actually people who are sensitive to judder and latency believe it or not.

Exactly, thats the whole point here, it would be the best solution...also, if a 24p international standard will be ever made, the europeans will stop to get their movies speeded up to 25fps.

 

[Mintmaster]

As stated in the article , you must have need of doing recompressing and recompressing :

As stated several times by Brody, we're not talking about DV, and hence we're not talking about intraframe compression. That applies only to Firewire and DV camcorders with portable realtime compression processors. It doesn't apply to DVD/HD-DVD, Bluray, or TV broadcast. Even in your specific case, it makes more sense to change the DV tape format, which wasn't meant for progressive recording anyway. I hope they made modifications in HDV.

but recovering the C frame requires decompressing the green and magenta frames and recompressing them into a new, 24p DV frame. That puts the C frame a generation down compared to A, B, and D."

Again, only applicable to a DV stream. The 1080i and 480i signals used by HDMI and component are simply one uncompressed field followed by another (with blanking and field signals between).

You sure you're a video pro? wink: just kidding)

 

[Mintmaster]

Democoder: What you say makes sense. It would definately spur manufacturers to make proper 24p support with actual refresh rate changes.

 

[DemoCoder]

Democoder: Latency is not as big of a problem with video as with gaming, as long as the audio is delayed by the same factor.

Yes, but there is no standard mandating this, which has caused scores of audio-sync problems. When ATSC OTA broadcasts first hit, many set top boxes had really bad audio sync problems. The more devices you chain together, the more likely the problem. So if you've got a DVD player going into an A/V Receiver going into a TV, you've multiple latencies being added, and most dumb-and-cheap systems don't strive to reprocess audio if they don't have to, which means many systems simply don't care about accumulated latency.

You make a good point about keeping it simple and not relying on the quirks of the TV, though. 24p output does have merit in that context, which I suppose resolves my question at the making of this topic--as long as it is clear that it's not "compulsory" for judder-free 24p display.

No, but then again, progressive output isn't compulsory for "interlace" free display, nor is 480p output compulstory for 480p display. The issue is that display devices exist amongst an ecosystem of A/V devices, and the more points of failure where things can go wrong raises the chance the things will go wrong.

 

[Ventresca]

As stated several times by Brody, we're not talking about DV, and hence we're not talking about intraframe compression.

As stated as much as times by me, i am talking about the case where IVTC is applyed by the player on the mpeg2 60i stream with flags stored on the disc, this would be the only case where you have 100% of the times a disc with a standard pattern to work with and this is done before the stream gets uncompressed.

That applies only to Firewire and DV camcorders with portable realtime compression processors. It doesn't apply to DVD/HD-DVD, Bluray, or TV broadcast. Even in your specific case, it makes more sense to change the DV tape format, which wasn't meant for progressive recording anyway. I hope they made modifications in HDV.

It apply to hd-dvd if the disc is encoded as 1080 60i and the IVTC is done by the player before the video is uncompressed, clearly it does not if the IVTC is done by the TV on the alredy uncompressed HDMI stream.

Again, only applicable to a DV stream.

It happens the same way for HDV with its mpeg2hd stream too...simply put, when you deal with compressed material and make a reverse telecine, this does happnes every time.

The 1080i and 480i signals used by HDMI and component are simply one uncompressed field followed by another

I am not talking about ITVC done with uncompressed stream, a real ITVC can't be done on the HDMI stream of movies with broken 3:2 pulldown patterns and this happens in a lot of movies that i know.

Bye,
Ventresca.

 

[Mintmaster]

Well that case is what Bohdy is talking about. Look at the title. 1080p/24 output. What you're talking about is the need for 24p recording.

This whole thread is talking about output. We mentioned DVDs and TVs many times (even you did!), so obviously we're talking about the signal coming through HDMI or component. Here is a perfect example of your misunderstanding:

As I explained, video which is encoded progressively with pulldown applied at playback (which you yourself admitted can be done) doesn't have this problem, and is reversible without quality loss, compression, recompression, whatever else.

It does not change anything, when the player apply the 3:2 pulldown and output a 60i stream , the C frame in the 60i stream is splitted and if it is displayed at 60i you are not reversing the video so you are fine, but if the Tv deinterlace the 60i stream to 24p again , it do a reverse telecine.

You're wrong here. The original info on the DVD was 24p MPEG2. The 60i output of the player is uncompressed. It's simply one field after another. It doesn't matter that C is split across two frames. Combining those fields is no different than combining the two fields that comprise A. 1080p/24 makes no difference to the information sent to the display.

 

[Ventresca]

Well that case is what Bohdy is talking about. Look at the title. 1080p/24 output. What you're talking about is the need for 24p recording.

The discussion covered also the question of how a native 24p ouput is better than a recovered 1080p ouput...i am not off topic.

A 1080p 24 output originated from an hd-dvd player playing a 1080i encoded disc wil still be 1080 24p output , it goes through the process i have exposed and its signal is worst in respect of another player with a movie encoded at 24p and outputting the 24p signal...

The other problem with a non 24p output is with the Tv that will not always be able to recover a perfect 24p signal from the source , with no flags to work with.

This whole thread is talking about output. We mentioned DVDs and TVs many times (even you did!), so obviously we're talking about the signal coming through HDMI or component.

does mentioning DVD means talking about the output in the first place...?

Here is a perfect example of your misunderstanding:
You're wrong here. The original info on the DVD was 24p MPEG2. The 60i output of the player is uncompressed. It's simply one field after another. It doesn't matter that C is split across two frames. Combining those fields is no different than combining the two fields that comprise A. 1080p/24 makes no difference to the information sent to the display.

Even if the original info was 24p dvd, with no flags there is the risk of image degradation like i said before :

"It is also possible, but more difficult, to perform reverse telecine without prior knowledge of where each field of video lies in the 2-3 pulldown pattern. This is the task faced by most consumer equipment such as line doublers and personal video recorders. Ideally, only a single field needs to be identified, the rest following the pattern in lock-step. However, the 2-3 pulldown pattern does not necessarily remain consistent throughout an entire program. Edits performed on film material after it undergoes 2-3 pulldown can introduce "jumps" in the pattern if care is not taken to preserve the original frame sequence (this often happens during the editing of television shows and commercials in NTSC format). Most reverse telecine algorithms attempt to follow the 2-3 pattern using image analysis techniques, e.g. by searching for repeated fields."
http://en.wikipedia.org/wiki/Telecine#Reverse_telecine_.28a.k.a._IVTC.2Finverse_telecine.29

It's a process with a certain percentage of risk , errors , artifacts, in my opinion, 24p native will always be the better solution