Native, judder-free 1080p24: Joe Kane says it matters, eyewitness agrees

[Grubert]

I have the feeling that the manufacturers don't think this is an important question at all. They use 1080p as a marketing point, but no mention is being given of how that 1080p is being handled and displayed.

However, video guru Joe Kane is very concerned about that, as we can see from an interview he did with Microsof's Ben Waggoner. Some sections follow:

Originally Posted by Joe Kane

The most important thing to me is I can tell you that the most efficient way to encode film-based information is to encode it as 24 pictures per second. But that offers opportunities to me in playback, in trying to do the presentation. First of all, I think the idea of a player playing back 24p -playing it back or not playing it back- I'm actually in favor of talking to all manufacturers of players. I don't care if it's Internet download. Actually, in the computer domain right now, playing back 24p is fairly straightforward.

So as an example, you claim that the other camp- you discuss that the other camp is discussing how they're delivering 24p and of course we know they're not. At least in the first generation of players, the 1080p24 that's on the disc is being converted to 1080i60 and then being converted back to 1080p24. There is a potential of harm happening in this double conversion. Actually, the same double conversion could happen in a player that only put out 1080i because you would also do the same transformation back.

I am actually pushing really hard to have player manufacturers provide a true 24p output. I do not want the 24p that's on the disc to be processed. I want it to be handed out in the native format.

[...]

The most important point that I wanted to bring up with 24 is that in the US world, we normally convert that to 60 and add the 2-3 motion where there is judder. I'm actually going to all manufacturers that I'm working with and telling them they have to be able to accept this 24p as an input and then they have to be able to play it back as 24, 48 or 72, so there is no 2-3.

[...]

As much as I have been working with Microsoft, I'm also working outside of Microsoft so that what you produce in source video can look even better to the consumer than they have ever seen it before, and playing back in 24, 48 or 72 is a critical part of what I'm trying to make happen.

[...]

I honestly believe I'm in this for the consumer. I honestly believe that I'm trying to advocate for the consumer what the consumer should have. I don't think the consumers actually do know for themselves what they should have, and I'm actually trying to be an advocate for what they should have.

First-hand testimony of native 24 in action during the CEATEC fair in Japan:

Originally Posted by d3code

It also was the first time I saw a player which can do 24-frame output and also had a TV which could input that 24 frames. so for the very first time no pulldown stuff to 60i etc., just the real video on disc and output. And oh boy does it look good!!!!

[Rob Zuber]

...you discuss that the other camp is discussing how they're delivering 24p and of course we know they're not. At least in the first generation of players, the 1080p24 that's on the disc is being converted to 1080i60 and then being converted back to 1080p24.

Is it confirmed that all first generation players will do this? Or is Kane talking out his....?

[Mr. Hanky]

Isn't judder-free p24 an oxymoron? (Yes, I agree p24 support on LCD's would be a good thing, but not the end all fix to addressing poor frame rate.)

[darinp2]

Originally Posted by Joe Kane

At least in the first generation of players, the 1080p24 that's on the disc is being converted to 1080i60 and then being converted back to 1080p24.

I get the feeling that Joe might be confused and that what he is referring to is really:

Originally Posted by Joe Kane

At least in the first generation of players, the 1080p24 that's on the disc is being converted to 1080i60 and then being converted to 1080p60.

I haven't heard of any players supporting 1080p24 output that are going to 1080i60 first, but I do understand that the Samsung goes to 1080i60 before 1080p60.

Maybe Joe just mixed 2 concepts together. Or maybe one of the 1080p24 output players (like the Pioneer) is converting to 1080i60. I doubt it though.

--Darin

[Kosty]

My freind at Cedia talked to Joe Kane personally at the Samsung area, and he used the second phrasing during his conversation.

I just think he mispoke or was misquoted in the interview, he knew very well that the Sammy only outputed 1080p/60.

[wolfyncsu7]

Can anyone describe what judder-free 1080p24 video looks like? Is it more detailed or just a smoother picture during pans/ fast motion? (Yes... I used wikipedia, but what I found there didn't make too much sense).

[Mr. Hanky]

When converting 24p for presentation at 60 Hz, this will introduce a 6 Hz difference component (30 fps - 24 fps). This effect can be noticeable as a judder during pans.

By having genuine 24 Hz support in the display, this 6 Hz difference component can be avoided. It is essentially one step closer to seeing it as it was shown in a real projection theater.

This does not mean that all forms of judder during pans are solved, however. There will still be a jerky component to film as long as it remains as a 24p format.

[wolfyncsu7]

Originally Posted by Mr. Hanky

When converting 24p for presentation at 60 Hz, this will introduce a 6 Hz difference component (30 fps - 24 fps). This effect can be noticeable as a judder during pans.

By having genuine 24 Hz support in the display, this 6 Hz difference component can be avoided. It is essentially one step closer to seeing it as it was shown in a real projection theater.

This does not mean that all forms of judder during pans are solved, however. There will still be a jerky component to film as long as it remains as a 24p format.

Ok... makes sense. And then supposedly if you take this signal and display it at 48, 72, or some other multiple of 24 Hz, it will be less jerky?

[Geographer]

Originally Posted by Rob Zuber

Is it confirmed that all first generation players will do this? Or is Kane talking out his....?

Yes, it is confirmed, none of the first generation players that are out now will output 24p. Goes for both camps. And before anyone points this out, I am talking about stand alones devices, not laptops.

-Mark

[audioNeil]

Originally Posted by Mr. Hanky

When converting 24p for presentation at 60 Hz, this will introduce a 6 Hz difference component (30 fps - 24 fps). This effect can be noticeable as a judder during pans.

By having genuine 24 Hz support in the display, this 6 Hz difference component can be avoided. It is essentially one step closer to seeing it as it was shown in a real projection theater.

This does not mean that all forms of judder during pans are solved, however. There will still be a jerky component to film as long as it remains as a 24p format.

Actually, it should be a 12 Hz judder (60 - 48). No-one displays at 30 fps (or shouldn't), they do 60 instead. If they did 30, you would have 6 Hz judder, but it wouldn't be 3:2 timing pattern. It would be nasty -- 6 frames per second would have to display double the time (a 2:1:1:1 pattern)! Now that would be terrible judder!

[Mr. Hanky]

Originally Posted by wolfyncsu7

Ok... makes sense. And then supposedly if you take this signal and display it at 48, 72, or some other multiple of 24 Hz, it will be less jerky?

At multiples, it will be exactly the same wrt motion smoothness. The multiples are more relevant for crt-based displays which need a certain refresh rate to prevent flicker issues. 24 Hz would be too low, but 48+ is usually sufficient for smooth lighting output. Some low latency CRT's may even need 72 Hz to become flicker-free.

[benwaggoner]

Originally Posted by wolfyncsu7

Ok... makes sense. And then supposedly if you take this signal and display it at 48, 72, or some other multiple of 24 Hz, it will be less jerky?

The key is making sure that source frames of equal duration turn into display times of equal duration.

Displaying 24p content at 60i/pi means that half the film frames are displayed for 2/60th of a second, and the other half are displayed for 3/60th of a second. So, a smooth pan comes out with a jerky "judder." 60i/p doesn't matter at all for judder. What we want is to have each film frame to be displayed for the same amount of time. 24, 48, 72, etecetera doesn't matter for most display types (obviously refresh rate is a big factor for CRT).

If anyone gets a chance to check out our HD DVD truck tour, you'll see HD DVD being played out as 48 Hz through a great Marantz projector. No judder, and without the flicker of a 2-blade projection system. Very, very nice stuff.

Note that the distinction between "native" 24p and inverse telecined 24p is inconsequential as far as the final image goes, since 3:2 removal is an entirely reversible process (basically you insert 12 repeat fields on playback, and then take them out with image processing). Additionally, a properly made HD disc today should be fully compatible with 24p playback in 24p native players (like the aformentioned laptop players).

There's been speculation about whether or not "24p" is going to matter in the format war, but there isn't any practical barrier to doing good 24p with either format.

[Mr. Hanky]

Originally Posted by audioNeil

Actually, it should be a 12 Hz judder (60 - 48). No-one displays at 30 fps (or shouldn't), they do 60 instead. If they did 30, you would have 6 Hz judder, but it wouldn't be 3:2 timing pattern. It would be nasty -- 6 frames per second would have to display double the time (a 2:1:1:1 pattern)! Now that would be terrible judder!

I don't know if 60 vs. 48 applies technically, since 48 contains no more motion information than the original 24. "60" implies interlaced output, which complicates things further wrt how often a full frame occurs. So that is why I maintain that there are 24 fps of material and there are 30 full frames per second under a 60 Hz refresh configuration. Hence, the difference component should correlate to 30-24=6. If you want to go with 12 Hz for the juddered frame, it is still just the same frame repeated twice at 6 Hz.

[Mr. Hanky]

Originally Posted by benwaggoner

If anyone gets a chance to check out our HD DVD truck tour, you'll see HD DVD being played out as 48 Hz through a great Marantz projector. No judder, and without the flicker of a 2-blade projection system. Very, very nice stuff.

You guys aren't setting up at any Fry's stores, are you?

[Wendell R. Breland]

Originally Posted by benwaggoner

but there isn't any practical barrier to doing good 24p with either format.

Ron (dr1394) states otherwise for HD-DVD, native 24p output is not in the specs. Has this not been covered in detail?

[lymzy]

Originally Posted by Wendell R. Breland

Ron (dr1394) states otherwise for HD-DVD, native 24p output is not in the specs. Has this not been covered in detail?

24p will be an additional feature for HD DVD players. It has nothing to do with HD DVD spec. Keith agrees with this. I hope this could end the no 24p for HD DVD FUD.

Also, Ron said the decoder couldn't ignore the flag per spec. He didn't say HD DVD player couldn't IVTC 1080i60 to 1080p24.

[Mr. Hanky]

So what exactly is on the disc on an hdvd?

[lymzy]

Originally Posted by Mr. Hanky

So what exactly is on the disc on an hdvd?

1080p24 stream with 3:2 sequence header or so called flag for movie source.

[TheLion]

Originally Posted by benwaggoner

Note that the distinction between "native" 24p and inverse telecined 24p is inconsequential as far as the final image goes, since 3:2 removal is an entirely reversible process (basically you insert 12 repeat fields on playback, and then take them out with image processing).

Ben, and yet Joe Kane says: "At least in the first generation of players, the 1080p24 that's on the disc is being converted to 1080i60 and then being converted back to 1080p60. There is a potential of harm happening in this double conversion. Actually, the same double conversion could happen in a player that only put out 1080i because you would also do the same transformation back.

I am actually pushing really hard to have player manufacturers provide a true 24p output. I do not want the 24p that's on the disc to be processed. I want it to be handed out in the native format."


Who is right then? And why are people (including me) seeing different levels of PQ for different IVTC implementations (eg. HQV/Gennum/VP50 Scaler versus Sony Ruby's internal IVTC processing (with no other filter enabled ) ) in combination with 1080i/60 output from HD-DVD if the process "is entirely reversible"? How can we be a 100% sure that nowhere during this "double conversion" (decoding into a 1080i/60 stream, IVTC back to 1080p/24) ANY form of filtering, smoothing, averaging,... "happens". Thats the potential harm that Joe is talking about - and you better believe Joe Kane

[benwaggoner]

Originally Posted by Wendell R. Breland

Ron (dr1394) states otherwise for HD-DVD, native 24p output is not in the specs. Has this not been covered in detail?

The exact details for how to decode HD DVD @ 24p aren't specified, but it's not hard to do. It's easier than with DVD, and 24p playback has been working well with DVD for years and years.

[Mr. Hanky]

Originally Posted by TheLion

Who is right then? And why are people (including me) seeing different levels of PQ for different IVTC implementations (eg. HQV/Gennum/VP50 Scaler versus Sony Ruby's internal IVTC processing (with no other filter enabled ) ) in combination with 1080i/60 output from HD-DVD if the process "is entirely reversible"? How can we be a 100% sure that nowhere during this "double conversion" (decoding into a 1080i/60 stream, IVTC back to 1080p/24) ANY form of filtering, smoothing, averaging,... "happens". Thats the potential harm that Joe is talking about - and you better believe Joe Kane

To add, it's one more opportunity for chroma error (ICE and ICP) to creep into the process. If that whole scene can be circumvented with certainty, that is a bonus, imo.

I guess the principle in effect here is a variation on the "keep your laws off my uterus" abortion chant. Ours would be "keep your damn, dirty hands off my bitstream!"

[benwaggoner]

Originally Posted by TheLion

Ben, and yet Joe Kane says: "At least in the first generation of players, the 1080p24 that's on the disc is being converted to 1080i60 and then being converted back to 1080p60. There is a potential of harm happening in this double conversion. Actually, the same double conversion could happen in a player that only put out 1080i because you would also do the same transformation back.

I am actually pushing really hard to have player manufacturers provide a true 24p output. I do not want the 24p that's on the disc to be processed. I want it to be handed out in the native format."

Who is right then? And why are people (including me) seeing different levels of PQ for different IVTC implementations (eg. HQV/Gennum/VP50 Scaler versus Sony Ruby's internal IVTC processing (with no other filter enabled ) ) in combination with 1080i/60 output from HD-DVD if the process "is entirely reversible"? How can we be a 100% sure that nowhere during this "double conversion" (decoding into a 1080i/60 stream, IVTC back to 1080p/24) ANY form of filtering, smoothing, averaging,... "happens". Thats the potential harm that Joe is talking about - and you better believe Joe Kane

As he said, there's a potential for harm, not a certainty. A native 24p output would be simpler once displays support it, and native 120 Hz remains my dream goal. But for those who want a great 24p experience today, we know we can deliver it. It'll be cheaper and easier as time goes on, of course.

There's two different angles here - the player perspective, and the format perspective. And a content guy, I'm mainly concerned with what the format can do, not particular CE devices. We're making discs today that'll work fine in future 24p playback systems, and also work fine with IVTC solutions today.

[RobertR1]

Am I getting this right?

Mutiples of 24 are ideal but 24 itself is too slow. Using a 3:2 pulldown is what's causing the "judder" during scenes that pan.

I'm now more sensitive to this and other IQ features. This is ofcourse good from a quality stand point but bad from a wallet standpoint

[dialog_gvf]

Originally Posted by benwaggoner

A native 24p output would be simpler once displays support it, and native 120 Hz remains my dream goal.

There are displays that support it now, like Pioneer plasmas.

Let's not let the player makers off the hook by saying it's OK for them to wait until there are many more displays available that accept 24p. Because the display makers may take the attitude that they won't bother with the cost of adding 24p inputs if the source devices aren't available.

Gary

[benwaggoner]

Originally Posted by RobertR1

Am I getting this right?

Mutiples of 24 are ideal but 24 itself is too slow. Using a 3:2 pulldown is what's causing the "judder" during scenes that pan.

Yep, that's it!

We want each frame's displayed duration to be identical. 3:2 pulldown (60i or 60p) doesn't do this, hence judder. Ironically for all the "BD is progressive" hype, it's actually better to use 60i playback than 60p for more systems these days, since they can do inverse telecine with 60i but not with 60p...

[rlsmith]

Yes, judder matters, and I can see it myself.

However, I have to note that a lot of people do not seem to be able to notice it very well, at least consciously.

I once had my wife at a demonstration where judder elimination was the point (converting to 72P with pull-down removal etc.) We must have watched a scene from Titanic 20 times (the digital pan of the ship flying overhead), and she simply could not see it. She generally can see artifacts when they are pointed out to her (but doesn't care).

Almost no one I know mentions judder as an issue.

What I do think is true is that higher production frame rates would make an incredible difference to almost everyone. Film at 60FPS is incredible! I would like to see 4320p96 as a production standard for Hollywood films. But 24FPS is so poor that the judder is a minor artifact at worst IMHO.

[Rob Tomlin]

Originally Posted by rlsmith

What I do think is true is that higher production frame rates would make an incredible difference to almost everyone. Film at 60FPS is incredible! I would like to see 4320p96 as a production standard for Hollywood films. But 24FPS is so poor that the judder is a minor artifact at worst IMHO.

I would love to see 60fps used too. But, from what I recall from another thread, it will never happen.

[Mr. Hanky]

I'd love to see them pull a measily 30 fps for film or digital capture. It would also sidestep this whole judder issue, as well, w/o breaking the bank for filmstock or digital storage.

[rlsmith]

Originally Posted by Rob Tomlin

I would love to see 60fps used too. But, from what I recall from another thread, it will never happen.

I am not so sure.

The studios clearly need to do something to bring theatrical exhibition back into play. Theatres still make a lot of money and also provide the launch pad and validation for other forms of release. 3D, more resolution, and higher frame rates are obvious choices for the direction of technology.

When you see the effect of higher frame rates, you really know you are looking at something.

I cannot imagine there is a technical problem. Moreover, the cost of digital storage is already a joke with 250GB drives selling for $50, and holographic storage just around the corner.

[benwaggoner]

Originally Posted by Rob Tomlin

I would love to see 60fps used too. But, from what I recall from another thread, it will never happen.

The DCI spec has a mode for 48 Hz (although mainly meant for 2x24 Hz 3D projection). If DCI digital projeciton catches on, I could see someone trying to do a sports or action movie at 48 Hz, and it'd be pretty cool.

And it's semi-easy to make a 24p 35mm print for a 48 Hz DI if needed. 60 to 24 gives us judder the other direction . Although the 48 fps master probably would use 96 Hz shutter... Maybe shoot a digital 72 Hz shutter to split the difference?