Panasonics (European models) 24p, telecine inversion and bandwith: Current status

[Jeroen1000]

A lot about this has been said already, so I was a bit surprised that there isn't a 100% definitive answer (unless I really suck at googling). Let's clear it out so I can finally decide to buy now or wait for the next generation (G12) panels. Do read carefully, I've noticed that in videoland every detail matters:-). My apologies if everyone has figured this out already:-)

1. 24p playback and Real Cinema (24p blue-ray source!):

• In the case of the BR player outputing 24p the Panasonic does a 4:4 pull down, effectively using a 96hz refresh rate = No judder = IFC (intelligent frame creation = disabled). This is called native 24p playback.
• Same scenario as just above but with IFC enabled. What does this exactely do? Add 4 frames so the refresh rate is 100hz? This is called Real Cinema and does some nasty things? I'm asking this because no one mentions the refresh rate when using Real Cinema.

------------------------------------------------------------------
2. Things that should be treated as 24p but are not detected as such, causing improper deinterlacing of film content, not video content!

The main thing leading to my confusion is people talking that the deinterlacer is bad. Whilst it correctely deinterlaces video mode material. So IF I get it correctely it should somehow extract 24p from 1080i broadcasts only and only if the broadcast is film. The technique used to do this is called reverse 3:2 pulldown?

• What does this mean for 1080/50i as I think 3:2 reverse pull down is only for 1080/60i.
• As I understand, it uses video deinterlaceing mode instead of film deinterlacing mode beacause it cannot recognize the difference between 24p film and normal video. What the heck is normal video then? And what is broadcasted in normal video?
• Couldn't we just force a setting that sets the panasonic to film deinterlace?

---------------------------------------------------------------------
3. a third and (luckily) final issuse is concerned with the bandwith

This site states that motion resolution goes up to 1080p as long as IFC is turned off.

Panasonic TH42PZ85B Plasma HDTV Calibration & Benchmark

But if you check out this PDF, no panasonic gets the full 1080 lines:

http://hdguru.com/wp-content/uploads...-125-hdtvs.pdf

I know the panasonic in the first link (TH-42PZ85B) is not listed in the PDF of the second link. But this one is:

Panasonic TH42PZ80 / TH42PZ80B Review: 42" Viera Plasma HDTV

Why does hdguru.com get other results then hdtvtest.co.uk? Is there such a big difference between US models and UK models?
And just for the record could somone please explain the issue with motion resolution vs static resolution. I get it is an issue, but haven't a clue why.

I hope the answers for these questions are useful for many of us noobs:-)

Cheers,

Jeroen

[choddo2006]

1. ok so you don't want guesses but I'm going to guess. IFC on will give you 2 (identical) frames of source then 2 (identical) frames of invented interpolation, then 2 frames of source. So a 96Hz picture with a simulated 48Hz source.

2. When 1080i/50 film material is processed by something that can do IVTC, it just converts it to 1080p/25.
Normal video is anything that was "filmed" in 50Hz (or 60Hz) interlaced using a video camera, like sport and most TV shows. A good example is The Wire. You'll often find rips that have converted this to 24fps thinking it was shot on film but they judder like a good un. A 60Hz rip on the other hand is smooth because that's how it was recorded.
No you can't force a setting because the Pannies can't do it. They still give a very good result for the price.

3. It's down to personal interpretation of what people are seeing. Motion resolution is lower than static because the pixels have to change colour as things move and how quickly and accurately they can change colour dictates how much detail is still visible during motion.

[Jeroen1000]

Hi Choddo,

Thanks for taking the time to make me a little smarter

Regarding 2. The Pannies should deinterlace 1080i/50 film content, convert it to 24p and then do a 4:4 pull down but they can't? Or they can but are unable to recognize (watch out I've picked up a new word here) the film cadence?

Let me clarify this a little further:

Video: they deinterlace to 25p and then do a 4:4 pull down = 100hz
refresh rate.
Film : they do the same but actually they should deinterlace to 24p and then do a 4:4 pull down (getting us to 96hz). Question remains whether they just lack the algorithm for it (as you say they can't), or whether they just don't spot the broadcast as being fillm. Shame that a lot of cheaper tv's can do this. I do wonder if I would be able to see the difference between a TV that does this properly and one that does not. After all, at this point its only numbers to me.

Regarding 3 I'm still a bit information hungry. I've noticed that LCD's in some worst case scenario's only get to a shameful 500 lines (that's less than SD television!). Have you per chance come across some good articles explaining the tech. details that go behind this bandwith issue?
Of course, feel free to give your understanding about it!

cheers,

Jeroen

[rewerb]

Originally Posted by Jeroen1000

Regarding 3 I'm still a bit information hungry. I've noticed that LCD's in some worst case scenario's only get to a shameful 500 lines (that's less than SD television!). Have you per chance come across some good articles explaining the tech. details that go behind this bandwith issue?
Of course, feel free to give your understanding about it!

cheers,

Jeroen

TVs lose resolution when movement occurs. LCDs are worse than plasma for this, though the very latest are catching up. Even the Panasonic only claim 900 lines (with IFC enabled), but 500 is good for an LCD, 800-900 for plasma.

[choddo2006]

Originally Posted by Jeroen1000

Regarding 2. The Pannies should deinterlace 1080i/50 film content, convert it to 24p and then do a 4:4 pull down but they can't? Or they can but are unable to recognize (watch out I've picked up a new word here) the film cadence?

Let me clarify this a little further:

Video: they deinterlace to 25p and then do a 4:4 pull down = 100hz
refresh rate.

No. Video should be deinterlaced to 50p as it was recorded at 50fields per second. The screen (or processor) should just fill in the blank lines in each field. The better they do that, the better the end result.

Film : they do the same but actually they should deinterlace to 24p and then do a 4:4 pull down (getting us to 96hz).

No. In this case you want 25p out (with a 2:2 or 4:4 cadence if you like) because you can't convert 25 to 24 cleanly.

Question remains whether they just lack the algorithm for it (as you say they can't), or whether they just don't spot the broadcast as being fillm.

I don't think they try.

Shame that a lot of cheaper tv's can do this. I do wonder if I would be able to see the difference between a TV that does this properly and one that does not. After all, at this point its only numbers to me.

Are you sure cheaper TVs can? And if they can, do they get it right?

Regarding 3 I'm still a bit information hungry. I've noticed that LCD's in some worst case scenario's only get to a shameful 500 lines (that's less than SD television!). Have you per chance come across some good articles explaining the tech. details that go behind this bandwith issue?
Of course, feel free to give your understanding about it!

It's not bandwidth, it's the response speed of the pixel to accurately change colour fast enough.

[Jeroen1000]

Originally Posted by choddo2006

No. Video should be deinterlaced to 50p as it was recorded at 50fields per second. The screen (or processor) should just fill in the blank lines in each field. The better they do that, the better the end result.

No. In this case you want 25p out (with a 2:2 or 4:4 cadence if you like) because you can't convert 25 to 24 cleanly.

I believe the samsungs in the same size and price range do this correctly.

And I thought we got rid of the old 4 % 'PAL' speed up. But since it should be converted to 25p (I'm rambling about deinterlacing film from 50i to 25p here) it is still there.
However, both items you describe always amount to 100hz and never get us to 96hz? My calculated guess is that only 24p signals are displayed at 96hz? So the feature will only be used when a blue-ray player outputs 24p (or any other source capable of doing this)?

I must admit I do not understand how deinterlaced film from 50i to 50p is any different from deinterlaced film from 50i to 25p as the TV is 100hz. I thought the good Pioneers engineers had invented some magic to get 50i film to 24p and then display it at 72 of 96 or 120hz.
The way I thought it to be working was quite wrong and how you describe it should be done does not make sense to me.

[choddo2006]

We have got rid of 4% speedup with BluRay @ 1080p/24 but with anything broadcast at 50Hz, you're stuck with it.

However, both items you describe always amount to 100hz and never get us to 96hz?

Yes but you shouldn't really care.

My calculated guess is that only 24p signals are displayed at 96hz? So the feature will only be used when a blue-ray player outputs 24p (or any other source capable of doing this)?

HD DVD, HTPC, AppleTV, PopcornHour, TVix.

I must admit I do not understand how deinterlaced film from 50i to 50p is any different from deinterlaced film from 50i to 25p as the TV is 100hz.

I don't think the TV being 100Hz is relevant to be honest. A 50Hz TV would be the same.

As for 50i->50p or 50i->25p being the same ...

if the process is that both fields are weaved together to produce 25p then doubled to display 50p then they're both effectively the same.

If on the other hand, the process is that each field is bobbed (or a derivative) to produce 50 individual frames (which is what you DO want for video sources) then they're not the same. This is what the Pannies seem to do.

I thought the good Pioneers engineers had invented some magic to get 50i film to 24p and then display it at 72 of 96 or 120hz.

Nope. This is difficult because: Assuming you want to keep things smooth and not drop frames... after 1 second, you'd have an extra frame received by the TV from the source which you need to store until you want to display it. After 2 seconds, you're 2 frames behind.

At the end of a 2 hour movie, you're obviously 4% or 4.8 minutes behind = 54Gigabytes of uncompressed video data @1080p/24fps/24bit colour. Where's that going to go?

The way I thought it to be working was quite wrong and how you describe it should be done does not make sense to me.

I don't see why not. 4% speedup is by far the least bad way of handling film at 50Hz.

[Jeroen1000]

Now this is the stuff you never get out of a salesman.

I think I gets it Video has 50 unique frames but is broadcasted as interlaced. So we just want to make it progressive (= 50p @50hz). Showing it as 100hz means we just show each frame twice, which makes a smoother picture. But as you say is quite irrelevant to the process itself.

Film is also broadcasted as 50i but the TV should say hmmm, this is film so I have to do reverse telecine or else I'll be weaving the wrong fields together. In simple words the TV should not treat film as having 50 unique frames. It probably makes up a frame here and there.

What a revelation to me:-)

[choddo2006]

Yep, almost there.

Not IVTCing 50Hz film doesn't mean you weave the wrong fields though, it means you don't weave them at all and you instead make up (as you do with video) the blank lines in each field. It's possible to get a pretty good picture doing this especially with diagonal detection /smoothing. It's not just a simple case of doubling each line which produces nasty stepping on diagonals, especially on SD. Go look in a Sony Centre at one of their cheaper sets for an example

Displaying 50fps material at 100Hz doens't necessarily mean it's smoother (although IFC might claim to do so if turned on) - it just means the picture is more solid and flickers less (although there's an argument that since plasma flashes its pixels on & off really fast in order to produce different colours anyway, it's not quite as clear cut as the CRT days)

Oh and video is normally actually recorded/digitised (as well as broadcast) as interlaced too, to save space.

[Jeroen1000]

Originally Posted by choddo2006

Yep, almost there.

Not IVTCing 50Hz film doesn't mean you weave the wrong fields though, it means you don't weave them at all and you instead make up (as you do with video) the blank lines in each field. It's possible to get a pretty good picture doing this especially with diagonal detection /smoothing. It's not just a simple case of doubling each line which produces nasty stepping on diagonals, especially on SD. Go look in a Sony Centre at one of their cheaper sets for an example

.

I do not believe Pannies make up half the field of a frame when it comes to deinterlacing 50i video. They take the first field and then the second, and deinterlace them (weaving them together). I'll have to rest my brain for a while know:d
Or were you pointing out somthing else?



This comes not from my own knowlegde but from many many internetsources I've browsed this week.

[choddo2006]

They absolutely don't do that. You're getting it the wrong way round. That's what Pioneer's deinterlacer does with film.

Anything that did that with video (e.g. SkyHD's deinterlacer when trying to handle bloomberg tickers) would mess up big style and you'd have horrible combing on the edges of anything moving that was remotely contrasting colours. Think about it. You've got different fields captured 1/50th of a seconds apart. If something has moved in that time, and you try to just weave together the two fields... they don't line up! Whereas with film you've got a single frame which has been split into two fields, so they do match up when you weave them back together.

(edit: Worth mentioning that most video processors can detect motion and decide to weave video in part of a frame where there isn't any but I'd be amazed if the Panny doesn't do this and not sure on the Pioneer)

The question is; do Pannies treat film as video as well, or do they do what Pioneer do? Every technical test I've seen suggests they treat it as video because test disks with single pixel height lines (i.e. lines which exist in one field but not in the next) shimmer on & off when deinterlaced, but I'm more than interested to know if they've changed that with VReal3. They're desperately short on technical detail when you dig.

[rewerb]

Originally Posted by choddo2006

They absolutely don't do that. You're getting it the wrong way round. That's what Pioneer's deinterlacer does with film.

Yes, you are correct. All the technical reviews say the panny (along with lots of others) does not deinterlace broadcast film material correctly. But it's video processing is fine.

To show the film based material correctly you have to combine (deinterlace) two 50hz frames together to make one 1080p image. Also note that HDTV Test said that it wasnt actually a big deal, you do get more "noise" - jaggies and moire effects.

[Jeroen1000]

Choddo, your patience with noobs is really exemplatory!
To be honest I missed the key fact that the fields in progressive material represent the same instance in time. And with interlaced this is not the case. Now all your replies fall into place for me.

Having read up on the matter, it would seem that deinterlacing 50i film is a lot easier than 60i film ('PAL' vs 'NTSC').
Can you confirm that this is called 2:2 cadence with 50i and 3:2 cadence with 60i? And is this terminology equal to 2:2 pulldown and 3:2 pulldown?

[Kalos Geros]

The term "cadence" describes the structure of video, "pullup" and "pulldown" are actual processes that create / recreate frames to/from cadences...

Panasonic Tvs are notorius for discarding the whole field on 2:2 and 3:2 signal...at least they could force BOB if processor is unsure instead of discarding half a resolution...curiously, I believe they actually use weave when they detect the scene is not moving, but when it is, the whole field gets removed...but don't be sad previous Philips series 5 is even worse in that on the same shot it weaves the picture until something passes in front of the camera,then it goes haywire and drops the whole field, after the object had passed through the shot, it reverses to weave but remains unstable, constantly droping in/and out of the weave...

Pioneer 490 DVD players, just like Panny plasmas, discard the whole field in moving scenes, but try to weave when they detect no motion...

a neverending game this is...

[Jeroen1000]

Although I'm getting what it should do with 50i film, I'm not yet entirely on board with what it is doing wrongly.

(that'll be for my evening google session)

That BOB deinterlacing, is it exactly the same as in deinterlacing method C on this site Interlacing - Luke's Video Guide

So it is using this BOB deinterlacing on film AND on video? You say one whole field (out of 2) per frame is discarded, but going on the description on the aforementioned website, every field is stretched to fill up an entire frame (effectively doubling the framerate) thus not removing any fields.

EDIT: Oh, I think I got it. It throws aways 1 field of every frame when deinterlacing 50i film??