1080i vs 1080p on next gen dvds

[Renoir]

Hi guys!

This is my first post (well apart from saying hello in a newbie thread).

I have a simple question regarding next gen dvd and apologize if this has been answered elsewhere. I had a bit of a search but thought I'd get confirmation if possible. This question is from a uk resident interested in the technical details of watching next gen dvds on 50hz displays (I assume most hd displays sold in the uk accept 50hz signals?)

I seem to remember a while back people asking whether the next gen dvd's would have the video data stored on the disc as either 1080i or 1080p and based on my current understanding of deinterlacing It made me wonder whether it matters for a uk resident such as myself. I'm fairly sure that the vast majority of the content I'll be looking to buy in hd will have been either recorded on film at 24fps (and so can be scanned at 1080 lines for the discs) or recorded on hd cameras at 1080p24 (certain american tv shows).

Now does it matter if this particular content is stored on the disc as either 1080i or 1080p? If stored at 1080i could you not just flag the 1080i content as film and then simply weave adjacent fields, add 4% and then frame double when connecting to a 50hz progressive display?

Now what reason would there be for storing 1080p24 native content as 1080i on the disc. ISTR it having something to do with the data size required but would the size of the data not be quite similar between 1080i48/50 and 1080p24/25 therefore rendering it pointless not to encode the data as 1080p24?

I'm sure interlacing 1080p24 content to 1080i50 if necessary (I imagine you may want to do this if your display doesn't support a 1080p input or if for whatever reason your player can't output in 1080p) is just as easy to do if not easier than weaving 1080i to 1080p then adding 4% speedup and frame doubling?

Mr Neal I'm sure you could answer this in your sleep so hows about it!

 

[Stephen Neal]

In no particular order - here are some thoughts :

Will we get 1080/24p discs or 1080/25p discs? The former would allow us to have the same video data as the US/Japan (meaning only one video and audio encode during mastering), the latter would allow the speed-up in the audio to be pitch-corrected with better gear than might be possible in a domestic set-up, and would mean a domestic player wasn't having to resample 48/96kHz audio back to 48/96kHz when it was sped up locally etc. (Imagine 1080/24p material with a 48kHz track, played back at 1080/25p - you'd end up with something like 50kHz audio sampling, which you'd then have to pitch-correct and re-sample back to 48kHz. Quite difficult to do that without decoding to 5.1 I'd have thought - which would mean on-board decoding?)


1080/25p vs 1080/50i. In theory, 1080/50i can carry a 1080/25p signal losslessly, in segmented frame format. (In fact a lot of broadcast gear running in 25p is actually using 25psf to record the two halves of the frame in separate interlaced fields). As you correctly say, a simple weave/field-merge algorithm will re-create the 25p from the 50i. HOWEVER if the 50i is encoded to be displayed on a 50i display - then you might expect 25p material to be vertically pre-filtered to reduce interlace twitter.

Of course the 50i filtering could be performed in the player rather than during mastering, and would possibly be a requirement for 50i playback if the disc were mastered with full vertical-res 25p.

However the number of native interlaced displays in 50Hz countries is quite low - basically CRT sets in Aus (most of which won't be HDCP so won't get full-res HD from a BluRay or HD-DVD anyway), or ALiS plasmas (the 1024x1024 Hitachi etc. panels?) - so pre-filtering may not be an issue for us.

Converting 1080/24p to 1080/50i or 50p without speed-changing would be horrid.

Playing 24p discs at 24p/48i would also be an issue (especially if you have standard def PAL 50 / RGB outputs as well as HD) - though might be an option on some players.

Playing back 24p discs at 60i or 60p with 3:2 pulldown would be a backwards step IMHO - as the 2:2 pulldown you'd get with 25p at 50i or 50p is a nicer system with less unbalanced movement.

My gut feeling is that we may still see 25p/50i mastered discs. Interesting question is whether both HD-DVD and BluRay support on-disc MPEG2 or MPEG4 as actual 24p or 25p video, rather than segmented frame/field within a 48i/60i (with 3:2 flagging) or 50i video. ISTR that BluRay and HD-DVD differed in this regard - with different MPEG2/4/VC-1 subsets being implemented?

 

[Renoir]

Will we get 1080/24p discs or 1080/25p discs? The former would allow us to have the same video data as the US/Japan (meaning only one video and audio encode during mastering), the latter would allow the speed-up in the audio to be pitch-corrected with better gear than might be possible in a domestic set-up, and would mean a domestic player wasn't having to resample 48/96kHz audio back to 48/96kHz when it was sped up locally etc. (Imagine 1080/24p material with a 48kHz track, played back at 1080/25p - you'd end up with something like 50kHz audio sampling, which you'd then have to pitch-correct and re-sample back to 48kHz. Quite difficult to do that without decoding to 5.1 I'd have thought - which would mean on-board decoding?)

Very interesting aspect I didn't have in mind when writing my post but is useful to know nonetheless. Yeah I imagine they'd have to decode to 5.1 but ISTR that most of the players will have on board decoding for most of the formats so that multi channel pcm can be sent to a receiver via hdmi which would be useful given that most receivers at mo don't support the latest version of hdmi that will be required to send the next gen audio codecs natively. I would imagine as you say that the pitch correction being done with better gear than in a domestic set up would be the best reason for giving us UKers (and others of course) the 1080p24 content as "treated" 1080p25. Incidently could this argument also be used to suggest that sd content upscaled and released on next gen dvd will be far better than watching a sd dvd on a upscaling dvd player (would studios even release such products)? Bit off topic this last bit so feel free to ignore

1080/25p vs 1080/50i. In theory, 1080/50i can carry a 1080/25p signal losslessly, in segmented frame format. (In fact a lot of broadcast gear running in 25p is actually using 25psf to record the two halves of the frame in separate interlaced fields). As you correctly say, a simple weave/field-merge algorithm will re-create the 25p from the 50i. HOWEVER if the 50i is encoded to be displayed on a 50i display - then you might expect 25p material to be vertically pre-filtered to reduce interlace twitter.

Afraid I'm gonna need some educating before I can respond to this part (well apart from the confirmation that the simple weave would be perfectly acceptable for certain content). I must confess I know nothing about this filtering stuff you mention. Perhaps you could briefly explain it or point me in the right direction to some links where I could learn more?

Of course the 50i filtering could be performed in the player rather than during mastering, and would possibly be a requirement for 50i playback if the disc were mastered with full vertical-res 25p.

However the number of native interlaced displays in 50Hz countries is quite low - basically CRT sets in Aus (most of which won't be HDCP so won't get full-res HD from a BluRay or HD-DVD anyway), or ALiS plasmas (the 1024x1024 Hitachi etc. panels?) - so pre-filtering may not be an issue for us.

As mentioned I don't understand the concept of filtering but like you say it may not be an issue due to so few hd 50i native displays. I guess if it turns out to be a non issue then it brings me back to my point that given the possibility of using simple weaving in combination with storing the data on disc at 1080i50 what difference does it make to us whether it's stored on disc as 1080p25 or 1080i50. As mentioned I wondered if it had anything to do with coding efficiency but wouldn't they be about the same. Obviously storing 1080p50 would take more data than 1080i50 but that's not gonna be the issue given how little content there is out there shot in 1080p50 (I know of none).

Converting 1080/24p to 1080/50i or 50p without speed-changing would be horrid.

Very true. I didn't think I'd suggested such a process though. Could be wrong.

Playing 24p discs at 24p/48i would also be an issue (especially if you have standard def PAL 50 / RGB outputs as well as HD) - though might be an option on some players.

Another interesting scenario I hadn't thought of which actually reminds I still have questions regarding dvd refresh rates and pc playback but's that for another time.

Playing back 24p discs at 60i or 60p with 3:2 pulldown would be a backwards step IMHO - as the 2:2 pulldown you'd get with 25p at 50i or 50p is a nicer system with less unbalanced movement.

Agreed and hopefully won't be an issue for us although as you mention below that's gonna depend on how hd-dvd and blu-ray store the data on the disc etc.

My gut feeling is that we may still see 25p/50i mastered discs.

Just to be clear do you mean speeding the 24p content up to 25p on their end and then storing it on the disc either as is (25p) or 50i with a simple flag that says to weave?

Interesting question is whether both HD-DVD and BluRay support on-disc MPEG2 or MPEG4 as actual 24p or 25p video, rather than segmented frame/field within a 48i/60i (with 3:2 flagging) or 50i video. ISTR that BluRay and HD-DVD differed in this regard - with different MPEG2/4/VC-1 subsets being implemented?

You have answered many of my questions and this last quote really sums up the big issue which will most likely have to wait till we know the final specs for hd-dvd/blu-ray before being answered. It is the fairly general question of how can the data be stored on the disc and which ways will studios choose to do it and why. As you say do both formats even allow the data to be stored as 24/25p or must it be stored interlaced with the appropriate flags (ISTR reading an interview that suggested both formats would support storing native 1080p content in some way). If we assume for arguments sake that content producers could store their 24p content as 25p on the disc what reason would they have for choosing it over 50i with flag or indeed vice versa. I guess what I keep harping on about is that it seems as though (your caveat about filtering issues aside) quality wise it should make no difference if it's stored either way so is there another issue I'm missing like perhaps as mentioned coding efficiencies?

 

[Mr.D]

Bear in mind dvd could store 24p and 25p SD material ...but commercial dvds are always encoded from interlaced masters.

Just because bluray and hd-dvd can store 24p/25p (hd-dvd can simply store progressive material as far as I'm aware...makes sense just data at the end of the day) doesn't necessarily mean they will in spite of what some of the badly worded press releases for both camps suggest.

I doubt you are going to see a significant difference between 1080i and 1080p material when it hits the display regardless of filtering. Its probably less significant than the quality of the mastering.

 

[Stephen Neal]

My comment about 25p carried as 25p vs filtered 25p carried as 50i, for display on interlaced displays is purely an interlace issue.

This because if a 25p signal, exploiting the full 1080p (or 576p in SD terms) vertical resolution, is simply interlaced without filtering, you end up with flicker at the frame, not field, rate on fine detail. This is because the fine vertical detail differs between fields and thus flickers at field, not frame rate. If there is a lot of HF vertical detail, you end up with annoying flicker on fine detail. You see this on cheap computer graphics systems used by some broadcasters - where broadcast gear would do proper filtering.

The most extreme case to consider is a 1080p signal which is alternate frame lines of black and white - 540 white lines and 540 black lines, alternating. If you interlace this you end up with 2 fields, one entirely white and one entirely black. If you fed this into an interlaced display you'd get a screen flashing at frame rate - and it would be horrible. With a bit of vertical filtering, you'd lose the fine detail, but you'd get a mid-grey frame, which would interlace to two mid grey fields, and no flicker.

This is an extreme example but the principle kind of still holds.

To reduce the levels of flicker vertical information is filtered (either in the video camera or in the telecine process) to reduce the levels of detail that would cause flicker. However the reduction is less extreme than using a 540/60p (or 288/60p system) so you still get a vertical resolution benefit from interlacing.

If you were carrying 25p material as 25psf within a 50i stream, but not aiming for it to be displayed on a 50i display in interlaced form, you wouldn't need to vertically filter - and that is how some broadcast 25p VTRs actually work. However if you are creating a 50i interlaced signal to actually display interlaced on an interlaced display, then you should pre-filter the 25p. This means that even if you de-interlace using a WEAVE or similar, you only get back the filtered 25p signal, not the unfiltered one.

 

[Renoir]

Ok I see.

Like you say there will most likely be very little incentive to pre-filter the 25p given how small the market is for displaying the 1080 signal on an interlaced display.

So again if we assume the filtering will not be an issue what reasons are there for choosing 25p over 50i or vice versa?

Mr. D points out an interesting observation about sd dvd having the ability to store 24/25p material on the disc natively but it rarely being used with content producers prefering to store it as 50i. Although I imagine this decision was simple given that the VAST majority of people watch sd dvds on interlaced displays (Is most content on sd dvds pre-filtered?). This as mentioned above is unlikely to be the case for next gen dvd.

So I still wander whether there's a coding issue for choosing 25p over 50i or 50i over 25p. This is of course assuming both formats can store the 24/25p signal natively which based on Mr. D's comment and the stuff I've read seems to be the case. Perhaps there's another reason for choosing one over the other I haven't thought of?

Thanks for the replies guys I think I'm gonna be learning a lot around here!

 

[richard plumb]

on a 1080p display (I'm assuming many of us will have one of these in the next 5-7 years), I'd expect native 1080p encoded content to look a lot better than 1080i with prefiltering.

Didn't someone say that 1080i with filtering contains approximately the equivalent information of 800p?

It'd certainly be interesting if Europe got pure 1080p mastered content and the US got prefiltered (because of their higher install base of interlaced HD displays). However the US also has a higher demand for HD, so maybe consumer pressure would mean 1080p masters there too.

anyway, my hope is for 1080p masters fullstop. 24/25 I don't mind really. probably lean towards 25p as I find NTSC juddering annoying but don't notice the pitch change.

Is there a 72Hz option for 1080p displays? that'd be ideal

 

[Renoir]

on a 1080p display (I'm assuming many of us will have one of these in the next 5-7 years), I'd expect native 1080p encoded content to look a lot better than 1080i with prefiltering.

The prefiltering would hinder picture quality on a 1080p display but my question is taking prefiltering out of the equation because as mentioned the installed base of interlaced hd displays is low in europe why choose 25p over 50i without prefiltering or vice versa?

It'd certainly be interesting if Europe got pure 1080p mastered content and the US got prefiltered (because of their higher install base of interlaced HD displays). However the US also has a higher demand for HD, so maybe consumer pressure would mean 1080p masters there too.

Indeed an interesting question as to what they'll produce in the US. Like you mention people with 1080p displays will get lower PQ if the content is prefiltered but if it's not then people with interlaced displays could have problems with flicker if I understand Stephen correctly. It'll be interesting to see but I'm only really interested from a practical point of view what they do for discs in Europe considering that as with current dvds I don't intend to buy any discs from the US.

anyway, my hope is for 1080p masters fullstop. 24/25 I don't mind really. probably lean towards 25p as I find NTSC juddering annoying but don't notice the pitch change.

The "NTSC" juddering as I understand it comes from the 3-2 frame repetition so if viewing on a 50hz this wouldn't be an issue whether 24p or 25p. Not noticing the pitch change however is perfectly valid. As for hoping for 1080p masters full stop that again relates to my question of what is the benefit of 25p over 50i with a simple "weave" flag. It would seem from Stephen's earlier post and others I've seen from him elsewhere that inverse 2-2 pulldown is a fairly simple process meaning that it may not make any difference whether the disc is 1080p or 1080i with a flag. Unless of course there are other issues such as coding efficiencies. I suppose recognising the flag and then weaving adds an extra step of complexity but how hard is it relative to any advantages to storing the data as 1080i?

Is there a 72Hz option for 1080p displays? that'd be ideal

That certainly is an interesting topic but you'd need both a player and display that are 72hz capable (or 75hz if the disc has already been sped up to 25p) which could be mighty difficult to find. Would be cool though! Having a 1080p24 master frame repeated to 72p would have no speed up, no pitch change and a very smooth image. Arguably the holy grail for 1080p24 native content.

 

[Stephen Neal]

The prefiltering would hinder picture quality on a 1080p display but my question is taking prefiltering out of the equation because as mentioned the installed base of interlaced hd displays is low in europe why choose 25p over 50i without prefiltering or vice versa?

Not sure - though it could be that some films are still transferred in 50i telecines, rather than in 25p I guess? Less likely these days I would have thought - especially in Europe where HD gear will have been bought more recently. (Early gear sold in the US was strictly 1080/60i or 720/60p - no 24p modes)

However - I'm not sure it is really correct to describe a 50i signal derived from a 25p source with no filtering as 50i - as it is not technically an interlaced signal (as it is not compatible with interlaced displays) Instead it should probably be referred to as a 25psf (25 frame progressive but segmented frames)

The "NTSC" juddering as I understand it comes from the 3-2 frame repetition so if viewing on a 50hz this wouldn't be an issue whether 24p or 25p.

No - we'd use 2:2 pull down for 25p rather than 3:2 - so both source frames are displayed for the same number of 50i fields or 50p frames (if you are outputting 1080/720 50p)

For 24p you'd have to do the speed up to 25p prior to 2:2 though - otherwise you'd be in the realms of standards conversion.

Not noticing the pitch change however is perfectly valid. As for hoping for 1080p masters full stop that again relates to my question of what is the benefit of 25p over 50i with a simple "weave" flag. It would seem from Stephen's earlier post and others I've seen from him elsewhere that inverse 2-2 pulldown is a fairly simple process meaning that it may not make any difference whether the disc is 1080p or 1080i with a flag. Unless of course there are other issues such as coding efficiencies. I suppose recognising the flag and then weaving adds an extra step of complexity but how hard is it relative to any advantages to storing the data as 1080i?

The biggest problem with 25p carried as 50i comes if it is edited in 50i, and there are edits between fields in the same frame. This shouldn't be an issue if the source itself is a 25p edit, but it could be if the edit has taken place in the 50i domain. (Though it SHOULDN'T...)

This is a much bigger problem for 3:2 60i edits - as in these you DO have to edit at field boundaries to keep the cadence.

I think there are 3 options here :

25p as 25p
25p as 50i but MPEG flagged as progressive
25p as 50i but MPEG flagged as interlaced

I'm not sure if there is a huge difference between 1 and 2 - but there would be between 1/2 and 3.

That certainly is an interesting topic but you'd need both a player and display that are 72hz capable (or 75hz if the disc has already been sped up to 25p) which could be mighty difficult to find. Would be cool though! Having a 1080p24 master frame repeated to 72p would have no speed up, no pitch change and a very smooth image. Arguably the holy grail for 1080p24 native content.

Some plasmas in the US do a reverse 3:2 then a 3:3 to convert 24p carried as 60i to 72p for display with no judder.

 

[Stephen Neal]

on a 1080p display (I'm assuming many of us will have one of these in the next 5-7 years), I'd expect native 1080p encoded content to look a lot better than 1080i with prefiltering.

Didn't someone say that 1080i with filtering contains approximately the equivalent information of 800p?

I have used that rough figure before - but I think it may be slightly skewed to video origination. It is one of the arguments for 720/60p not being as bad compared to 1080/60i vertically as many at first think.

It'd certainly be interesting if Europe got pure 1080p mastered content and the US got prefiltered (because of their higher install base of interlaced HD displays). However the US also has a higher demand for HD, so maybe consumer pressure would mean 1080p masters there too.

There is a large installed base of interlaced CRT HD sets in the US - so I suspect pre-filtering or playback filtering would be required still.

 

[Kalos Geros]

I'll jump in if you don't mind...

A thought about interlaced HD - in my view there is NO and will never be a DVD or HD-DVD with content that is stored in inetrlaced manner - that would mean that bytes are PHYSICALLY interlaced on the media i.e. you'd have data for every even line followed by data for every odd line...we all know that is not so, it is so only with tape meda where fields trully ARE physically magnetically recorded in fields...DVD or HD-DVD stores full frames with flags indicating the playback device whether the lines stored within it are to be output inetrlaced i.e. frame is buffered and its even lines are sent to the decoder first followed by odd lines for compatibility with interlaced CRt technology...ergo, any film material telecined even by interlaced telecine ends up PHYSICALLY as progressive on a DVD/HD-DVD because both telecine fields come from the same frame/moment in time...so there is absolutely no reason in my opinion to fear that there would be any artefacts rising from the fact that a film source material would be flagged as interlaced...I also don't see why it wouldn't be stored and run as original 24fps because I'm under impression that HD displays are multisync devices that can match any refresh rate between their min and max refresh specs so 72 Hz should be possible...projectors happily run at any refresh fed to them...

I can only see real-life problems in the fact that the display reolution may not match the source resolution so a 1080 source, say, would be scaled to 720 or 768 on a field by field basis so scaling would take place between every other line instead of every line which would yield results closer to the original film frame...

Unforunatelly once again has industry crumbled under the bandwidth problem and accepted a 1080i as de-facto a standard and like with original DVD that had to compromise for interlaced displays, the 1080i for compatibilty issue will stay to haunt us forever...I just hope that playback device will be smart enough to communicate via HDMI with display device and switch to 1080p mode where available, if authoring studios had done proper job with flags...too many ifs, again...

 

[Mr.D]

I'll jump in if you don't mind...

A thought about interlaced HD - in my view there is NO and will never be a DVD or HD-DVD with content that is stored in inetrlaced manner -

I don't know of any commercial dvd that is not derived from an interlaced master. The data contained on the dvd decodes to a interlaced video image.
The video then has to be deinterlaced correctly.

The flagging on a dvd is there to assist the encoding its not there as a guide to deinterlacing hence all the problems with flag reading deinterlace systems (although I have my doubts as to how widespread this particular issue actually is).

The discussion was about the ramifications on the new hd formats being 1080p native and how this would differ from 1080i.

Commercial dvd is most certainly an interlaced format. We don't yet know about the HD formats but I personally suspect we are going to get 1080i on the discs in exactly the same way that we get interlaced material on SD dvd.

If you read the thread you will find reasons as to why 1080p is not the same as 1080i deinterlaced to 1080p...what the actual difference will be on domestic displays is entirely open to conjecture.

I'm not actually worried about this issue personally. 1080i deinterlaced will look good 1080p might have looked a little more detailed on very high end 1080p displays , then again given the tolerances on the sort of afordable 1080p panels we will see I doubt there will be much in it.

 

[Kalos Geros]

If you read the thread you will find reasons as to why 1080p is not the same as 1080i deinterlaced to 1080p...what the actual difference will be on domestic displays is entirely open to conjecture.

Of course it is not the same...I tried to point out that HD DVD source, even when derived from an 1080 interlaced master of film material, could, by proper 1080p implementation on both display and playback side eventually end up as if were true 1080p transfer...

 

[Mr.D]

Of course it is not the same...I tried to point out that HD DVD source, even when derived from an 1080 interlaced master of film material, could, by proper 1080p implementation on both display and playback side eventually end up as if were true 1080p transfer...

No it can't read the thread. Note the vertical filtering issue which is really the crux of the matter. Whether its a issue of note is something we will have to wait and see. We are not talking about inverse telecine issues which are frankly facile.

 

[Stephen Neal]

I'll jump in if you don't mind...

A thought about interlaced HD - in my view there is NO and will never be a DVD or HD-DVD with content that is stored in inetrlaced manner - that would mean that bytes are PHYSICALLY interlaced on the media i.e. you'd have data for every even line followed by data for every odd line...we all know that is not so, it is so only with tape meda where fields trully ARE physically magnetically recorded in fields...

Not sure I follow - how you store the data on the media is not relevant directly to the signal. Interlaced video is stored in all manner of ways, from analogue line-sequential recording, to interleaved uncompressed digital (where interlaced video is spread around tape in a manner that means drop out is less visible), via DCT compressed MPEG and DV/MJPEG. Just because the data isn't interleaved, it doesn't mean the video isn't interlaced.

However field-based video sometimes compresses less well than frame-based video when it comes to temporal/spatial compression systems.

DVD or HD-DVD stores full frames with flags indicating the playback device whether the lines stored within it are to be output inetrlaced i.e. frame is buffered and its even lines are sent to the decoder first followed by odd lines for compatibility with interlaced CRt technology...ergo, any film material telecined even by interlaced telecine ends up PHYSICALLY as progressive on a DVD/HD-DVD because both telecine fields come from the same frame/moment in time...so there is absolutely no reason in my opinion to fear that there would be any artefacts rising from the fact that a film source material would be flagged as interlaced...I also don't see why it wouldn't be stored and run as original 24fps because I'm under impression that HD displays are multisync devices that can match any refresh rate between their min and max refresh specs so 72 Hz should be possible...projectors happily run at any refresh fed to them...

You're ignoring vertical pre-filtering to reduce interlace flicker - and you're also assuming all discs are mastered properly.

If a lazy or cheap mastering facility takes a 25p film transfer delivered on a 50i VT - they may well just leave it flagged as 50i. Sure the MPEG encoding system may detect the lack of motion between fields in the same frame and switch to progressive encoding.

However quite a lot of 25p productions are mastered in 25p, then have 50i caption rolls or crawls added. That will be an interesting one...

I can only see real-life problems in the fact that the display reolution may not match the source resolution so a 1080 source, say, would be scaled to 720 or 768 on a field by field basis so scaling would take place between every other line instead of every line which would yield results closer to the original film frame...

Unforunatelly once again has industry crumbled under the bandwidth problem and accepted a 1080i as de-facto a standard and like with original DVD that had to compromise for interlaced displays, the 1080i for compatibilty issue will stay to haunt us forever...I just hope that playback device will be smart enough to communicate via HDMI with display device and switch to 1080p mode where available, if authoring studios had done proper job with flags...too many ifs, again...

Though 1080i is currently the only practical standard for fluid motion.

Sure 1080/25p is practical for film-effect or film-transfer stuff - with the lower temporal resolution - but as 1080/50p is not yet a viable production system, 1080/50i is the only solution in the 1080 sphere for fluid motion... (Sports, Entertainment etc.)

 

[Stephen Neal]

Of course it is not the same...I tried to point out that HD DVD source, even when derived from an 1080 interlaced master of film material, could, by proper 1080p implementation on both display and playback side eventually end up as if were true 1080p transfer...

Except that a proper 1080/50i master from a 1080/25p transfer will have a lower vertical resolution, as the interlacing process should have been accompanied by a vertical pre-filter to reduce interline twitter (caused by interlace flicker on fine detail) on interlaced displays.

Sure you can omit the pre-filtering - however if you do, then you are really dealign with 25psf not 50i.

 

[richard plumb]

Some of the placeholder boxes on show at CES had '1080p' printed on them. Wonder what the studios mean with that, or whether its just a marketing thing?

 

[Mr.D]

Some of the placeholder boxes on show at CES had '1080p' printed on them. Wonder what the studios mean with that, or whether its just a marketing thing?

I am not going to hold my breath for 1080p on these formats. Trouble with issues like this is the marketing people often don't have a clue what it is they are describing.

1080p for example might just mean the player can deinterlace 1080i rather than the format itself being 1080p native. I reckon the studios will do one encode at 1080i and use it on both formats. I'm not expecting the picture quality of bluray to be better than HD-DVD. So far the bluray blurb seems to emphasise all the extras you can put on it....I can't actually remember the last time I watched an extra off a dvd. All I want is a good transfer.

Instead of criticising transfers for having too much edge enhancement or lacking DTS we'll be moaning about 1080i versions relative to 1080p versions.

 

[richard plumb]

from a blog linked to in the 'HDMI downsampling' thread

During the Q&A session, Sony's Don Eklund confirmed that Sony and MGM titles would be encoded on the discs at 1080/24p. The user will set the player to convert this native resolution as required to match the capability of his or her display. For most displays, the user will set the player for 1080/30i or 1080/60p. But if your display will accept 1080p/24 you can set it for that output, and the display will presumably double or (preferably) triple the frame rate to eliminate flicker. A display rate of 1080p/48 or 1080p/72 would eliminate the need for 3/2 pulldown, producing smoother motion and minimizing artifacts

seems like good news. Sounds like the players deal with interlacing - can they filter before outputting? Anyway, seems ideal.

They also said the launch Sony/MGM titles won't be using the ICT flag, so they'll play full Hidef through component

oh, and uncompressed 5.1 PCM audio.

 

[Mr.D]

Yep I've read that one. I'm still not betting on there actually being any 1080p discs though.
I hope they will be but the term is being bandied about by people in terms too unspecified at the moment. I could imagine these people almost making the same sorts of comments if someone passed them a paper on dvds containing 60i with 3:2 pulldown on players that inverse telecine correctly and incorrectly surmising the material is 24p on the discs ( as so many people on here do).

Wait and see methinks.

 

[Renoir]

Boy are you guys in for a long post. Enjoy!

Not sure - though it could be that some films are still transferred in 50i telecines, rather than in 25p I guess? Less likely these days I would have thought - especially in Europe where HD gear will have been bought more recently. (Early gear sold in the US was strictly 1080/60i or 720/60p - no 24p modes)

Ok so 50i may be chosen over 25p for some content transferred using older equipment. Sounds reasonable

However - I'm not sure it is really correct to describe a 50i signal derived from a 25p source with no filtering as 50i - as it is not technically an interlaced signal (as it is not compatible with interlaced displays) Instead it should probably be referred to as a 25psf (25 frame progressive but segmented frames)

Duly noted. I will from now on try to refer to 50i derived from an unfiltered 25p source as 25psf and 50i derived from a prefiltered one as 50i.

The "NTSC" juddering as I understand it comes from the 3-2 frame repetition so if viewing on a 50hz this wouldn't be an issue whether 24p or 25p.

No - we'd use 2:2 pull down for 25p rather than 3:2 - so both source frames are displayed for the same number of 50i fields or 50p frames (if you are outputting 1080/720 50p)

For 24p you'd have to do the speed up to 25p prior to 2:2 though - otherwise you'd be in the realms of standards conversion.

My point exactly. The NTSC juddering wouldn't be an issue on a 50hz display.

The biggest problem with 25p carried as 50i comes if it is edited in 50i, and there are edits between fields in the same frame. This shouldn't be an issue if the source itself is a 25p edit, but it could be if the edit has taken place in the 50i domain. (Though it SHOULDN'T...)

This is an interesting point but would it not have the same impact on PQ whether it's stored in 25p or 50i (25psf) when displaying on a progressive display. The weaving process wouldn't have any bearing on the fact that the content was edited in the 50i domain versus having it stored as 25p would it? I imagine editing in the 50i domain would end up sucking either way if it's eventually shown at 50p.

I think there are 3 options here :

25p as 25p
25p as 50i but MPEG flagged as progressive
25p as 50i but MPEG flagged as interlaced

I'm not sure if there is a huge difference between 1 and 2 - but there would be between 1/2 and 3.

The difference between 1 and 2 is the crux of my original post. Based on the comments in this thread it seems pretty clear that if handled correctly number 1/2 should have identical picture quality when viewed on a progressive display leaving other reasons for studios to choose one over the other.

Just so I understand you correctly does number 3 refer to 50i from a prefiltered 25p source?

Some plasmas in the US do a reverse 3:2 then a 3:3 to convert 24p carried as 60i to 72p for display with no judder.

Sweet!

There is a large installed base of interlaced CRT HD sets in the US - so I suspect pre-filtering or playback filtering would be required still.

If a disc in the US had prefiltered 1080i so that it was compatible with interlaced displays I assume that would penalise those with a 1080p display (lower vertical resolution). Would it therefore make sense to rely on playback filtering as and when needed rather than the studio prefiltering it so that 1080p owners (who will soon be the majority) get max quality while compatibility is still maintained with the installed base of hd crts?

I don't know of any commercial dvd that is not derived from an interlaced master. The data contained on the dvd decodes to a interlaced video image.
The video then has to be deinterlaced correctly.

I was under the same impression and to be honest what the "dvd decodes to" is what's most relevant to this discussion and not how the bits are physically stored on the disc.

The flagging on a dvd is there to assist the encoding its not there as a guide to deinterlacing hence all the problems with flag reading deinterlace systems (although I have my doubts as to how widespread this particular issue actually is).

Interesting. I'd heard about there being problems with flag reading but really know nothing about it. I wasn't aware the flag was to assist the encode I thought it was there to help guide the deinterlacing. Perhaps Mr. D you could point me to some relevant info?

If you read the thread you will find reasons as to why 1080p is not the same as 1080i deinterlaced to 1080p...what the actual difference will be on domestic displays is entirely open to conjecture.

If I have understood the thread so far the only reason it would be different is if the 1080i were from a prefiltered 25p source and the 25p native was not.

I'm not actually worried about this issue personally. 1080i deinterlaced will look good 1080p might have looked a little more detailed on very high end 1080p displays , then again given the tolerances on the sort of afordable 1080p panels we will see I doubt there will be much in it.

I take it with this statement you are assuming the 1080i is prefiltered and hence would have lower vertical resolution than unfiltered 1080p?

However field-based video sometimes compresses less well than frame-based video when it comes to temporal/spatial compression systems.

Now that's the kind of info I'm interested in. If it turns out that 25p compresses better than 25psf as 50i or 50i then that's another good reason to just dump the data as 25p on the disc. You could then have the player interlace to 50i and add vertical filtering if outputting to an interlaced display (how hard would this be for the player from a technical and practical point of view?).

However quite a lot of 25p productions are mastered in 25p, then have 50i caption rolls or crawls added. That will be an interesting one...

Not sure I understand this could you explain.

Though 1080i is currently the only practical standard for fluid motion.

Sure 1080/25p is practical for film-effect or film-transfer stuff - with the lower temporal resolution - but as 1080/50p is not yet a viable production system, 1080/50i is the only solution in the 1080 sphere for fluid motion... (Sports, Entertainment etc.)

Very good point! I'm mainly interested in 24/25p sourced content (films and certain TV shows) but this is an interesting technical tidbit nonetheless.

Except that a proper 1080/50i master from a 1080/25p transfer will have a lower vertical resolution, as the interlacing process should have been accompanied by a vertical pre-filter to reduce interline twitter (caused by interlace flicker on fine detail) on interlaced displays.

Sure you can omit the pre-filtering - however if you do, then you are really dealign with 25psf not 50i.

Should 1080i be chosen then it'll be interesting to see whether they go for 50i or 25psf. As mentioned I wander whether performing the filtering on playback if necessary is the best compromise to satisfy everyone?

seems like good news. Sounds like the players deal with interlacing - can they filter before outputting? Anyway, seems ideal.

Given that most people will be using progressive displays to watch hd content it makes sense for progressive to be the default and to have the player do interlacing if required although whether the filtering can be done on the fly may well be a key factor in deciding if that happens.

Yep I've read that one. I'm still not betting on there actually being any 1080p discs though.
I hope they will be but the term is being bandied about by people in terms too unspecified at the moment. I could imagine these people almost making the same sorts of comments if someone passed them a paper on dvds containing 60i with 3:2 pulldown on players that inverse telecine correctly and incorrectly surmising the material is 24p on the discs ( as so many people on here do).

Wait and see methinks.

There certainly seems to be a lot of confusion about the various terminology and dodgy marketing doesn't help.

Well done if you read through all that

 

[Stephen Neal]

To clarify my 25p material with 50i caption comment. It is quite common in the UK for 25p film sourced shows, or 25p post-produced/originated video, to have 50i captions added (or 25p film credits sped up to 50i!) as 25p motion is pretty nasty.

You can therefore end up with background video with a 25p motion characteristic (i.e. no motion within the two fields in each frame), but superimposed on this a 50i crawl (with motion between the two fields in each frame). This can cause all sorts of issues both in encoding (as some MPEG encoders decide the whole frame is progressive) and de-interlacing.

This can also happen on broadcasts where you have 25p programme material with a small portion of added 50i signing. I've seen some de-interlacers decide the whole frame is 25p (because the bulk of it is) - and turn the signer into flickery, comby mush. I guess the only way round this is if you have de-interlacing done on a block by block basis, and with a conservative drop to 50i if there is any confusion.

 

[Kalos Geros]

Yes, Stephen, this is an issue, captions get the combing effect because de-interlacer weaves the fields together which can make scrolling text unreadable if it moves too fast...but how many scrolling text moves at such a speed that combing would make it unreadable except stock market tickers...LOL

block/by/block de-interlacing...is any of us rich enough to get such a baby withot robbing The US Federal Reserve!!!

 

[Tarbat]

You can therefore end up with background video with a 25p motion characteristic (i.e. no motion within the two fields in each frame), but superimposed on this a 50i crawl (with motion between the two fields in each frame). This can cause all sorts of issues both in encoding (as some MPEG encoders decide the whole frame is progressive) and de-interlacing.

Thanks for this description, it helps explain something for me. I see this happening all the time on TV broadcasts. My TV has a "Film Mode" that detects if the video is progressive and display it as progressive. So, on a program that was filmed on film, it detects film and displays as progressive. It then gets to the scrolling titles at the end and the titles judder up the screen and you can see interlace tearing. If I then pause (on Sky+) and play, the TV detects interlace and the titles scroll smoothly. Also happens when they vertically mix the scrolling titles and a trailer for the next programme.

 

[Stephen Neal]

Yes, Stephen, this is an issue, captions get the combing effect because de-interlacer weaves the fields together which can make scrolling text unreadable if it moves too fast...but how many scrolling text moves at such a speed that combing would make it unreadable except stock market tickers...LOL

block/by/block de-interlacing...is any of us rich enough to get such a baby withot robbing The US Federal Reserve!!!

To be honest - this is why I really regret the rise of progressive displays whilst we still broadcast in interlaced format. If I could buy a decent HD Ready CRT set I would...

Until de-interlacing actually delivers decent results on all interlaced broadcasts it seems a flawed concept.

I know PCs can provide quite high quality de-interlacing solutions (including some block-based de-interlacing) - but they aren't feasible to introduce into an HDMI path - and are really only suitable for SD de-interlacing and scaling apps?