Is compression on HD signals really that bad?

I just splurged on my first quality HD tv.

I can get cable, fiber, or either dish provider in my area.

Should I stay away from comcast till the compression issues are resolved?
Fiber is so expensive, is it really worth it for a poor grad student like me?

I just am at a loss of what to do and a bit disheartened after reading about all these damn problems and wonder if I won't even enjoy my HD.

On a side note, the new turbodish plan costs the same as a DTV plan with more HD channels and 200 SD channels.

That makes no sense, and neither does this thread.

Can anyone guide me in the right direction for a quality picture so I don't piss away the little money that i have?

Wow, this thread got snipped cleanly (post Aug 2) by the AVS data glitch (see data loss memo atop each forum).

Doesn't seem posts can be restored unless some who contributed earlier saved their posts.

Perhaps TVOD and nm88 can restore some of the test patterns they posted just before the data loss? (Think it was here.). TVOD's pattern-generator image (true 1920X1080 resolution wedges) was used by nm88 to show full resolution for his Sharp 1080p LCD panel, with two non-downrezzed zoomed camera shots showing full 1920X1080 resolution. Plus nm88 similarly snapped HDNet's wedge test pattern (http://www.avsforum.com/avs-vb/showthread.php?p=7359152&&#post7359152), via DirecTV's MPEG-4, showing its resolution was roughly 1300 lines maximum horiz, and of ambiguous (IMO) vertical resolution.

Good shots/technique for other comparisons. nm88's sequence, AIUI, used a PC to test a LCD panel for resolution with TVOD's pattern; HDNet's pattern should be captured via DBS or cable to see how compression/filtering along the path, perhaps including STB limitations, influences resolution and likely HD PQ. Some have reported measuring close to 1920X1080 from the HDNet pattern. -- John

Is compression on HD signals really that bad?

I just splurged on my first quality HD tv.

I can get cable, fiber, or either dish provider in my area.

Should I stay away from comcast till the compression issues are resolved?
Fiber is so expensive, is it really worth it for a poor grad student like me?

I just am at a loss of what to do and a bit disheartened after reading about all these damn problems and wonder if I won't even enjoy my HD.

On a side note, the new turbodish plan costs the same as a DTV plan with more HD channels and 200 SD channels.

That makes no sense, and neither does this thread.

Can anyone guide me in the right direction for a quality picture so I don't piss away the little money that i have?

You didn't mention over the air reception. I know there's not much on during the summer, but it's free, and usually considered to be high quality.

I'm an OTA only viewer right now. And yes, over-compression bugs me. Its really shameful that was have to put up with macroblocking during the frickin Olympics. NBC should be ashamed.

And furthermore, I have to say I have never seen a 1080i channel that I don't see at least some blocks. CBS uses their full 19 Mbps but I still see it some, mostly during sports. Both FOX and ABC give the best looking picture here IMO... and they're both 720p. It irritates me greatly that 1080i even exists, 720p (within the limitations of a 19Mbps MPEG-2 bitstream) is far superior. IMO the only reason 1080i still exists is marketing. 1080 is a bigger number than 720 so the 1080i channels can say "look our HD is better"!

This happens on every HD broadcast I've seen since 2001. Any fast movement, fast pans, or bright flashes causes problems. This has always been the case and it's not going to change.
I just use an Algolith FLea to visually decrease the prolems.
I'm an OTA only viewer right now. And yes, over-compression bugs me. Its really shameful that was have to put up with macroblocking during the frickin Olympics. NBC should be ashamed.

And furthermore, I have to say I have never seen a 1080i channel that I don't see at least some blocks. CBS uses their full 19 Mbps but I still see it some, mostly during sports. Both FOX and ABC give the best looking picture here IMO... and they're both 720p. It irritates me greatly that 1080i even exists, 720p (within the limitations of a 19Mbps MPEG-2 bitstream) is far superior. IMO the only reason 1080i still exists is marketing. 1080 is a bigger number than 720 so the 1080i channels can say "look our HD is better"!

And furthermore, I have to say I have never seen a 1080i channel that I don't see at least some blocks. CBS uses their full 19 Mbps but I still see it some, mostly during sports.
Who is your local CBS affiliate?

I'm an OTA only viewer right now. And yes, over-compression bugs me. Its really shameful that was have to put up with macroblocking during the frickin Olympics. NBC should be ashamed.

And furthermore, I have to say I have never seen a 1080i channel that I don't see at least some blocks. CBS uses their full 19 Mbps but I still see it some, mostly during sports. Both FOX and ABC give the best looking picture here IMO... and they're both 720p. It irritates me greatly that 1080i even exists, 720p (within the limitations of a 19Mbps MPEG-2 bitstream) is far superior. IMO the only reason 1080i still exists is marketing. 1080 is a bigger number than 720 so the 1080i channels can say "look our HD is better"!

Yup - same here w/ TWC in San Diego.

The macroblocking has been horrendous for me - totally unacceptable and defeats the purpose of "HD".

May as well call it "Horribly Defective" :o

Still shots and studio shots are great.

Any motion and the picture starts to macroblock - and if the motion continues then the macroblocking gets worse.

I was watching the HD broadcast of the Monday Night Football Game: Packers vs Bengals - OMG! The picture was glorious and is what I went HD for!

Even for motion - practically NO macroblocking at all - I had to look hard for any.

Same night as the game I would switch to the NBC-HD channel on TWC to watch the Olympics and what do I get???? Horribly big macroblocking on motion.

NBC=Nasty Bit Compression

Erik

1080i and it's predecessor Hi-Vision 1125 (1035i) existed before 720P. Not long ago there was little support for 720P professional equipment compared to 1080i. Now it's common that equipment is multi-standard. High end cameras often use 1080P sensors which can be converted internally to either HD format or SD. It seems in 50hz lands that 1080i is alive and well too.

I agree that 720/60P is less susceptible to blocking artifacts on sports than 1080/60i, although there are those who will disagree. OTH 1080i does quite well with 1080/24P based material. Many will say that 1080i sports look sharper. NBC struggles because of their lower network fronthaul bitrates. They do make an effort to make the backhaul bitrates very high, sometimes over 100Mbs. Conversion from 50hz to 60hz such as the Olympics creates even bigger challenges.

The US being a relative early adopter of digital broadcasting is using MPEG2 for ATSC. It's too bad that as HD is just now becoming more mainstream that better codecs can not be taken advantage of for OTA.

Which leads us to the previous MPEG4 debate in the last version of this thread :D

Perhaps TVOD and nm88 can restore some of the test patterns they posted just before the data loss? Here's the one sourced from uncompressed HD SDI and captured with an AJA card:

http://img232.imageshack.us/img232/1350/hdtp1ky5.th.png (http://img232.imageshack.us/my.php?image=hdtp1ky5.png)

I'll admit this one confuses me as it seems to violate Nyquist limits. I think the markers on the edges indicate overscan.

To repeat, again, my 1080i images from WNBC-DT and NYC's TWC, don't have blocking artifacts, and 1080i images from most sources are noticeably crisper than those from 720p broadcasts--as AVSers have reported since HD began. None of my other 100s of channels via my cable system have consistent blocking artifacts (static or with motion) either. Viewing on a year-2000 64" CRT RPTV.

There are variations in PQ between Olympic sporting events, as you'd expect, based on lighting and likely cameras being used at each location.

Sources that consistently broadcast unwatchable OTA signals should be reported to the FCC since they're misusing their spectrum allocation. -- John

EDIT: Thanks for reposting the pattern, TVOD. I've assumed, perhaps incorrectly, that it was either computer generated or from a hardware pattern generator originally (non-sampled) and thus not subject to Nyquist limitations. Also assumed nm88, in the now-missing posts, created a file for a computer input to his Sharp LCD panel (showing full 1920X1080 effective resolution) rather than displaying your pattern from double-clicking it, which is how I viewed it. A right-click for its properties shows 5205 bytes (!?)

My file shows 323K. It's a PNG and the original BMP was around 6 Megs. The thumbnail is 5K.

I know when I used an RP CRT set it couldn't resolve as much detail as my current sets so I saw less probelms since it didn't have the lines of resolution to resolve the full 1920x1080 image. But my RP CRT set only had 7" guns which could only resolve 1600 lines of resolution at best.. I see problems on every single HD broadcast(Thousands) I've seen since 2001 on Cable(Comcast), on satellite(DirecTV), and fiber (FIOS). As well as all the OTA channels in the DC area.
I've just learned to live with them but every program I watch I can see them. The flea reduces it but doesn''t eliminate it. and this is currently with 40 to 50 HD channels I have access to.

And this is whether it's a 32inch HD set or a 67" HD set. I can easily see them.

The US being a relative early adopter of digital broadcasting is using MPEG2 for ATSC. It's too bad that as HD is just now becoming more mainstream that better codecs can not be taken advantage of for OTA.

I think there is still hope in better pre-processing of the signal, itself. While mpeg-2 as a codec may be as mature as it will ever get, there is always room for pre-processing what you feed it. If you could dynamically filter the input in a similarly intelligent and adaptive manner (based on dynamic vs. avg metrics of motion and detail criteria, variable filtering at luminance extremes, on-the-fly up/downscaling) that avc does internally (when it gets in over its head), that will give an easier signal for the mpeg-2 encoder to deal with, altogether. There's no way you can make up for "efficiencies" or retaining detail under insufficient bitrate, but I expect you can get to a point of graceful degradation, similar to avc. That will give an overall experience of "better quality", given that the non-demanding scenes get through fine plus the demanding scenes get through with some degradation but not necessarily in an offensive manner. The bonus is that the entire installed base of atsc/mpeg-2 style decoders in the field can still yield the benefits.

I believe there are some possibilities still yet available if they would continue to play around with (further development) super-fast "piecewise 2-pass" variations of the classic real-time mpeg-2 encoder, as well. Who knows what the sky is the limit is...it doesn't even have to be an "mpeg-2 encoder" that makes the stream. It could be anything in the "black box", just as long as the data in the output is still "mpeg-2 legal".

It's not a perfect replacement for avc and "unlimited bitrate", but I'm sure you could pull-off some amazing things that would make an entire atsc revision to avc compatibility strategy pretty moot. It could be applied to existing digital sd/mpeg-2 based cable/sat services, as well.

I've seen problems with motion on some shows with OTA, but what bothers me more often is the color differences between shows and stations. I can have one station looking pristine, and change the channel and it looks like someone turned up the saturation about 10 clicks. The next show on the same channel may be close to perfect.

In a similar manner, dialnorm is yet another thing that has gone completely out of control on the digital broadcast scene. It's just crazy that the advent of digital and hd facilitate all kinds of great things, yet for all it's worth, even the basics of tv are getting neglected on a frighteningly routine basis.

Who is your local CBS affiliate?

KRQE Albuquerque.

I'd agree that MOST of the time telecined material on 1080i looks quite good if they're using the full 6 MhZ spectrum for 1 channel. One huge exception on CBS though is the opening sequence to "The Big Bang Theory".

I think I'll post some screens of the Olympics tonight, so those of you outside the US can see what we have to put up with.

And I DID see some macro-blocking during NBC's SNF broadcast of the Hall of Game game. Was better than last season though.

KRQE Albuquerque.
Yes! KRQE is in fact one of the few stations left that has no subchannels. A report said they pour almost 18 Mbps into their one channel. That must be nice. Even our stations with one subchannel dump at least 10% of their bits into null packets so we don't get used to quality HDTV. One station is throwing away 40% of their bandwidth. It's incredible!

I think I'll post some screens of the Olympics tonight, so those of you outside the US can see what we have to put up with.
Take a look at the ones I posted in the Olympics thread. You won't be able to beat them.

I think there is still hope in better pre-processing of the signal, itself. While mpeg-2 as a codec may be as mature as it will ever get, there is always room for pre-processing what you feed it.I've thought about that, but with blocking it's already down to the pretty much the lowest frequencies of the DCT block. If one were to filter the video ahead of the encoders, would the block boundaries be less visible? It seems post processing would work better as it would work with any encoder and could smooth out boundary transitions.

In a similar manner, dialnorm is yet another thing that has gone completely out of control on the digital broadcast scene. It's just crazy that the advent of digital and hd facilitate all kinds of great things, yet for all it's worth, even the basics of tv are getting neglected on a frighteningly routine basis.Totally agree. It's misuse not only didn't improve things (yet), it's made things worse with each network using a different static value. It would have better if everyone had just left the setting to the default -27db until there is proper support. It's too easy to get into that discussion again so I'll leave it there.

Perhaps TVOD and nm88 can restore some of the test patterns they posted just before the data loss? (Think it was here.).D* MPEG-4 HDNet:

http://www.avsforum.com/avs-vb/attachment.php?attachmentid=116803&stc=1&d=1218739977

http://www.avsforum.com/avs-vb/attachment.php?attachmentid=116804&stc=1&d=1218739977

http://www.avsforum.com/avs-vb/attachment.php?attachmentid=116805&stc=1&d=1218739977

TVOD's pattern:

http://www.avsforum.com/avs-vb/attachment.php?attachmentid=116806&stc=1&d=1218740091

http://www.avsforum.com/avs-vb/attachment.php?attachmentid=116807&stc=1&d=1218740091

http://www.avsforum.com/avs-vb/attachment.php?attachmentid=116808&stc=1&d=1218740091

And furthermore, I have to say I have never seen a 1080i channel that I don't see at least some blocks.D*'s MPEG-4 1080i channels (excluding locals) do not have any significant blocking artifacts, even during heavy motion. Of course I suspect they're downresed (see above), but still, they don't have those artifacts whereas their MPEG-2 channels did.

OTA is another matter; the ASTC spec is limited to MPEG-2, an archaic codec which is poor at 1080i even at the top OTA bandwidth, and on top of that not all stations use all the bandwidth.It irritates me greatly that 1080i even exists, 720p (within the limitations of a 19Mbps MPEG-2 bitstream) is far superior. IMO the only reason 1080i still exists is marketing. 1080 is a bigger number than 720 so the 1080i channels can say "look our HD is better"!1080i is better; on low motion scenes (or any telecined material) it is 2.25x the resolution of 720p.

99% of TV material is either telecined film or 60i video, both of which will look better on 1080i. Therefore, except for maybe the odd sports program that is actually captured at 60 fps, 720p is almost a total waste, nothing more than a downresed version of 1080i.

I've thought about that, but with blocking it's already down to the pretty much the lowest frequencies of the DCT block. If one were to filter the video ahead of the encoders, would the block boundaries be less visible? It seems post processing would work better as it would work with any encoder and could smooth out boundary transitions.

Yes, post processing is just as relevant a step in the process. It's really just a discrete implementation of the same things that happen inside avc. It looks ahead AND behind of the actual encoding. If there is stuff that can be changed to make encoding easier (less prone to inducing blocking, in the first place), it will do it...and if something didn't quite work out right after the process, it will address that, as well. Naturally, avc is far more integrated, and the various tricks can be coordinated in a more integrated manner, right down to direct feedback and dynamic tweaking of the encoder operation, itself. It's just worth noting that even mpeg-2 can achieve some feats, once you can get to the point of watching/processing what goes in, as well as watching/processing what comes out. For the US market, this has special importance, since it can leverage all of the existing mpeg-2 decoding equipment already in the field.

99% of TV material is either telecined film or 60i video, both of which will look better on 1080i.

What's the last 1%? :confused:

psychotic subliminal signals :eek:

What's the last 1%? :confused:I'm not sure about this, but I recall reading somewhere that some sports broadcasts make use of 720p/60. I don't watch them so I wouldn't know. The 1% isn't an exact figure, just an approximation for "very small percentage".

99% of TV material is either telecined film or 60i video, both of which will look better on 1080i. Therefore, except for maybe the odd sports program that is actually captured at 60 fps, 720p is almost a total waste, nothing more than a downresed version of 1080i.Actually many high end HD cameras use 1920 horizontal CCDs and scale down for 1280. Some cameras like Sony use 1080p vertical and scale to 720p while others like Thomson GV use pixel grouping to create native 720p vertical. In either case the 720p is not derived from 1080i. Exceptions to this are when shows are created in 1080i and cross-converted to 720p. Some examples are 720p PBS stations and some HD syndication.

Actually many high end HD cameras use 1920 horizontal CCDs and scale down for 1280. Some cameras like Sony use 1080p vertical and scale to 720p while others like Thomson GV use pixel grouping to create native 720p vertical. In either case the 720p is not derived from 1080i. Exceptions to this are when shows are created in 1080i and cross-converted to 720p. Some examples are 720p PBS stations and some HD syndication.Oh, I'm not implying that 1080i is always the source. Just that the final output on a 720p station is going to be little more than a downresed version of the final output on a 1080i station, for virtually all programming.

720p's only advantage is with true 60p programming, and I can't recall ever seeing that (I don't watch sports). That's why I hope stations don't move to 720p, and I in fact hope they all move away from 720p so I stop seeing downresed telecined film.

I'm not sure about this, but I recall reading somewhere that some sports broadcasts make use of 720p/60. I don't watch them so I wouldn't know. The 1% isn't an exact figure, just an approximation for "very small percentage".
All HD video production for 720p channels is 720p/60. That includes all HD sports on the ESPNs, ABC, and Fox as well as shows like American Idol, Good Morning America and Dancing with the Stars and surely many more to come (like ABC Nightly News). When you add up all the football and baseball games on these networks, it's certainly more than a "very small percentage".

720p's only advantage is with true 60p programming, and I can't recall ever seeing that (I don't watch sports).
Have you seen American Idol or Good Morning America?

All HD video production for 720p channels is 720p/60. That includes all HD sports on the ESPNs, ABC, and Fox as well as shows like American Idol, Good Morning America and Dancing with the Stars and surely many more to come (like ABC Nightly News). When you add up all the football and baseball games on these networks, it's certainly more than a "very small percentage".

ESPNs HD broadcasts for Monday Night Football have been exceptionally good based on my viewing experience.

Not so with CBS Sunday Football broadcasts in 1080i where, comparatively speaking, the macroblocking is more noticable for CBS.

This may have nothing to do with 1080i or 720p.

just stirring the pot...

Erik

Have you seen American Idol or Good Morning America?No. And I'll take your word for it that these two shows and some sports are shot at 60p.

But even if you add all of these up, it's still a very small percentage of the total number. I mean, you're talking about a few stations and two non-sports shows, compared to dozens of other HD stations and countless other shows.

Also, outside of sports or other programming where high motion detail is important, do you really care if it's 60p or 60i? A guy sitting in a chair talking on nightly news isn't going to look very different between the two. However, the 2.25x resolution advantage of 1080i will make a big difference in detail.

No. And I'll take your word for it that these two shows and some sports are shot at 60p.
Why don't you just check the programming that's on the 720p networks and see what shows are being produced in 720p video?

But even if you add all of these up, it's still a very small percentage of the total number. I mean, you're talking about a few stations and two non-sports shows, compared to dozens of other HD stations and countless other shows.
That produce what shows in 1080i video?

When I talk about 720p video production I'm talking about one third of the NFL games on network television, a large portion of NCAA football including several bowl games, a lot of NBA basketball including the playoffs and championship, a whole lot of Major League Baseball including both division playoffs and the World Series, the most popular show on television, another very popular show (Dancing with the Stars), and two 24 hour sports networks that broadcast hundreds of sporting events in HD every year. Although there is certainly more 1080i video production every year, the amount of 720p/60 production happening is certainly comparable.

Also, outside of sports or other programming where high motion detail is important, do you really care if it's 60p or 60i? A guy sitting in a chair talking on nightly news isn't going to look very different between the two. However, the 2.25x resolution advantage of 1080i will make a big difference in detail.
What you or I or others care about is another topic.

But even if you add all of these up, it's still a very small percentage of the total number. I mean, you're talking about a few stations and two non-sports shows, compared to dozens of other HD stations and countless other shows.Not really. Just looking at the OTA stations, you have 3 using 720p (ABC, FOX, MyNetworkTV) and 4 that use 1080i (CBS, NBC, PBS, The CW). Of those, you essentially have 4 networks that do quite a bit of live events (ABC, CBS, FOX, NBC), with 2 using 720p & 2 using 1080i.

When you look at the cable channels providing live events, ESPN HD & ESPN2 HD probably do more than all the rest put together, and they are 720p.


However, the 2.25x resolution advantage of 1080i will make a big difference in detail.The more accurate way to look at the difference between 1080i & 720p is the bandwidth used for each, where 1080i has approximately a 12% advantage. Not exactly earth shaking.

After years and years of evaluating the two, forum consensus makes it very clear there are many other factors affecting the end product, so many that in the end either 720p or 1080i are capable of great or poor HDTV. Making a generalization that one is 'better' than the other is simply wrong.

When I talk about 720p video production I'm talking about one third of the NFL games on network television, a large portion of NCAA football including several bowl games, a lot of NBA basketball including the playoffs and championship, a whole lot of Major League Baseball including both division playoffs and the World Series, the most popular show on television, another very popular show (Dancing with the Stars), and two 24 hour sports networks that broadcast hundreds of sporting events in HD every year.Ok. But that's still a tiny minority of the total programming on all HD channels.

If 720p channels would actually consistently broadcast 720p/60, I'd be more enthusiastic about them; but as it stands, on ones I watch (ABC, Fox, History HD, NGCHD) I have never seen a 720p/60 program. Just downresed material.

However, on the 1080i stations I watch, every HD program is making full or nearly full use of the bandwidth -- it's either 1080i video, or 1080p/24 telecined film -- and the image is more detailed than the image on 720p stations.

Not really. Just looking at the OTA stations, you have 3 using 720p (ABC, FOX, MyNetworkTV) and 4 that use 1080i (CBS, NBC, PBS, The CW). Of those, you essentially have 4 networks that do quite a bit of live events (ABC, CBS, FOX, NBC), with 2 using 720p & 2 using 1080i.Live events are only a small part of the total programming. If I watch an episode of Lost on ABC, all I'm seeing is a downresed version of what I'd get on the BD disc. If it were 1080i, I'd be seeing only a more heavily compressed version of the BD disc.The more accurate way to look at the difference between 1080i & 720p is the bandwidth used for each, where 1080i has approximately a 12% advantage. Not exactly earth shaking.Well, the most accurate way to look at the difference is the bandwidth actually being used for each.

For the most common source, telecined 24 fps material, 1080i has a 225% bandwidth advantage over 720p.

A large number of progressive-scan displays don't properly remove 2:3 pulldown, so any resolution gained by going to 1080i on 24p content is effectively thrown out by the TV since it's limited to 1920x540 before taking the content itself into account. Then, you take into account providers like Comcast who rape the picture to fit it into 12Mbit/s, leaving 24p content in 1080i a blocky mess while leaving 24p content in 720p relatively block-free. 720p still holds the overall quality advantage in this situation, especially as more and more providers go to 3 HD channels per QAM. Comcast, TWC, Cox, and Charter are already doing it.

All HD video production for 720p channels is 720p/60. That includes all HD sports on the ESPNs, ABC, and Fox as well as shows like American Idol, Good Morning America and Dancing with the Stars and surely many more to come (like ABC Nightly News). When you add up all the football and baseball games on these networks, it's certainly more than a "very small percentage".

I agree with all your points - except that isn't some drama now shot on video at 1080/24p or 720/24p these days - so not ALL video shot for the 720p nets is 720/60p?

Or are all Fox and ABC scripted dramas and sitcoms still shot on film?

(ISTR Arrested Development was shot 720/24p native video on Varicams - though is no now longer in production?)

Ok. But that's still a tiny minority of the total programming on all HD channels.

If 720p channels would actually consistently broadcast 720p/60, I'd be more enthusiastic about them; but as it stands, on ones I watch (ABC, Fox, History HD, NGCHD) I have never seen a 720p/60 program. Just downresed material.

However, on the 1080i stations I watch, every HD program is making full or nearly full use of the bandwidth -- it's either 1080i video, or 1080p/24 telecined film -- and the image is more detailed than the image on 720p stations.

The genres of production that you watch in 1080/60i native video on NBC, PBS and CBS will be the same genres of production that are shot 720/60p for ABC, Fox, ESPN.

Or are you saying you only watch those genres of programming (Sport, Studio Entertainment etc.) on the 1080i networks?

Almost every sport coverage or entertainment show shown on the 720p nets will have been shot 720/60p - not 1080/60i and cross-converted.

Drama IS usually post-produced in the 1080/24p domain rather than 720/24p. This is because most US dramas have a life in syndication and make significant income from overseas sales and optical media release, so shooting 1080/24p is a more future-proof production format. 1080/24p can be converted to 720/24p with no issue for 720p viewers (in fact it may look marginally better than stuff shot 720/24p native because of oversampling) The same is not true for 60Hz content - as 1080/60i cross-converted to 720/60p will have an issue with resolution on moving content, and can perform less well in this regard than 720/60p native content.

There is no major benefit (apart from cost) in shooting drama 720/24p over 1080/24p - so no reason to chose 720/24p even if you are a 720p broadcaster. However there ARE reasons to shoot 720/60p over 1080/60i if you broadcast in 720/60p. They are potential benefits not guaranteed ones - but they are big enough for some broadcasters to chose the 720p production standard.

This is why most studios and mobile trucks sold on the open market are switchable between 1080/60i and 720/60p native production - rather than just cross 1080i converted to 720p on their output (though in the case of some sporting events shared between 1080i and 720p nets cross conversion is still an issue) It is also a reason why Sony had problems selling their earlier HD camera range in this market segment - as, unlike Philips at the time who had a native switchable camera based on a 4320 line CCD that could line group, Sony only had 1080i native sensors. Sony's current range now have 1080p native sensors so can successfully sell to this market segment.

In Europe, the UK is entirely 1080/50i for HD transmission (as is Germany I believe), but the SVT HD (main Swedish HD service and the only current formal HD OTA service in Europe) is 720/50p as is Sky Italia in Italy.

The European Broadcasting Union has recommended 720/50p over 1080/50i - as a stepping stone to 1080/50p. This was a result of research suggesting that at normal viewing distances the benefit of 1080/50i in resolution terms wasn't apparent until you hit a 50" screen. It has also been noted that many de-interlacers don't deliver the full potential resolution of 1080i - instead treating it as 540p constantly (and thus when showing stuff shot on HDCam or DVCProHD will only deliver a 1440x540 res picture - not a million miles from 1280x720). (Interesting that some research has found that some people don't respond well to 50p captured video finding the lack of resolution reduction on movement a bit "odd", and prefer the resolution reduction that 50i delivers on moving stuff!)

However many European broadcasters are ignoring the 720p recommendation - partially as 1080 is a bigger number than 720 (and Joe Public will think automatically that 1080 is better than 720) - and Full 1080 panels are now widespread, and partially because in real world tests people responded better to the 1080i picture. I believe this is why Sky in the UK (who were expected to chose 720/50p for their sports networks - as they are closely linked with Fox in the US who are 720/60p - but research caused them to chose 1080/50i - partially because they are a satellite rather than OTA broadcaster who have more bandwith available to them)

Wow, this thread got snipped cleanly (post Aug 2) by the AVS data glitch (see data loss memo atop each forum).

Doesn't seem posts can be restored unless some who contributed earlier saved their posts.

Well, my post essentially said "yes". Of course, it was with the qualifier that I work with 40Mb/s video in post and am able to see network backhauls and truck feeds prior to compression for home viewing.

This means, I'm pretty much always somewhat disappointed with HD I view at home on TV. That's not to say there isn't some that looks very good to me within those limitations. It's just I know how much better it can look from the source.

The problem I have is when I see severely bit-starved broadcasts that break up under the least little bit of complication in the image. That really pulls me out of a show. Consistant levels of compression I can ignore - sudden infusions of macroblocking during intense lighting, explosions and fast movement drive me nuts.

BTW: the one thing I'm getting from this thread is perhaps:

Bicker1 is to Stephen King as nm88 is to Richard Bachman...???? ;)

Well, my post essentially said "yes". Of course, it was with the qualifier that I work with 40Mb/s video in post and am able to see network backhauls and truck feeds prior to compression for home viewing.

This means, I'm pretty much always somewhat disappointed with HD I view at home on TV. That's not to say there isn't some that looks very good to me within those limitations. It's just I know how much better it can look from the source.

The problem I have is when I see severely bit-starved broadcasts that break up under the least little bit of complication in the image. That really pulls me out of a show. Consistant levels of compression I can ignore - sudden infusions of macroblocking during intense lighting, explosions and fast movement drive me nuts.

BTW: the one thing I'm getting from this thread is perhaps:

Bicker1 is to Stephen King as nm88 is to Richard Bachman...???? ;)

Out of interest - when dealing with HD in post, what kind of data rate is used to encode content in the US market?

The BBC mandate a minimum of a 160Mbs codec to be used for post in their HD shows (anything edited below that is consired SD and comes out of the SD quota for the show - as does 16mm, HDV and live content backhauled at less than 60Mbs)

Whether this is ever enforced...

Out of interest - when dealing with HD in post, what kind of data rate is used to encode content in the US market?

The BBC mandate a minimum of a 160Mbs codec to be used for post in their HD shows (anything edited below that is consired SD and comes out of the SD quota for the show - as does 16mm, HDV and live content backhauled at less than 60Mbs)

Whether this is ever enforced...
It depends on the use and the source of the material.

Editing of material from live events is much lower. Editing from source material from camera tapes tends to be higher, as does film (which I don't work with).

The problem is, every time you do something to the video in the editing system, there's compression taking place. It's like editing the original JPEG image in Photoshop. Every version of the file results in lower quality - most of which is invisible to the eye (to a point). In a fast-turnaround broadcast environment, "lossless" isn't an option. Longer form post projects can afford the longer renders that maintain the higher image quality, but when it's 10 minutes to air time, being willing to lose a bit of fidelity is usually necessary.

Of course, some systems are better than others about it. Having hardware-based controls for common tasks (such as dissolves and color correction) avoids a lot of software rendering than can affect image quality. In addition, some systems (like the Quantel EQs) work almost entirely in a compressed form (unlike the Quantel QEPs). Avid and Final cut (I believe) tend to also do better with that sort of thing despite the use of "soft" controls.

Of course, other variables like whether the material is being brought via a shared server or is ingested to the local machine can affect this, too. Server-based clips with higher data rates would make moving media almost impossible in a timely manner.

Thanks, nm88, for restoring your Aug 11-lost, zoomed camera shots with HDNet's resolution wedges comparing D*'s HD Lite via MPEG2, MPEG 4 apparently at 1440X1080, and TVOD's full-res test pattern on your display.

Might be putting too fine a point on it, but BeachComber restored some of these images, too, in the DirecTV-rez thread (http://www.avsforum.com/avs-vb/showpost.php?p=14445203&postcount=19). He had them stored from subscribing to this thread, and reposted the images using Imageshack. Curiously, his restored D* MPEG4 horiz rez shot you zoomed in on is noticeably more distinct than yours above, making a resolvable 1440 reading more feasible, although both are ambiguous.

Must have missed it posted elsewhere, but if D*'s MPEG4 is now 1440X1080 (just HDNet, all channels?), shouldn't this be clear from MPEG analysis software? Surely this isn't just a conclusion from these HDNet resolution readings?
**************************

My 2 cents on 1080i vs 720p just above: Don't see blocking artifacts, consistently, on any digital channel from NYC's TWC, which like many cable systems seems to max out horizontal rez at 1290 lines (http://www.avsforum.com/avs-vb/showthread.php?p=6005155&&#post6005155)--at least on HDNet's channel. 1080i programming, on a 9"-gun 64" @8' CRT RPTV, usually shows crisper images than with 720p sources.

But, AIUI, 1080/60p cameras that can downconvert internally to 720/60p (Sony's 1500 series), or sports/location trucks converting 1080/60i capture to 720p, can boost 720p's delivered effective resolution much closer to 1280X720 than sampling images at 720p. Similarly, 1080/24p master tapes, used for most HD broadcasting, downconverted to 720p, boost effective rez closer to format rez.

By contrast, consultant Michael Robin's Broadcast Engineering article (http://broadcastengineering.com/aps/acquisition/broadcasting_hdtv_format/index.html) about 720p with standard sampling gives its limiting resolution (http://archive2.avsforum.com/avs-vb/showthread.php?p=5667245&&#post5667245) as 1035X504, where he gives horiz rez at LPH (582X1.78), not lines per 16X9 HD picture width. His similar article (http://broadcastengineering.com/aps/transmission/broadcasting_xi_hdtv_format/index.html) for 1080i gives estimated resolution as 1549X746. Measured resolutions during mid-'90s ATSC approval tests were 1638X800 (http://archive.avsforum.com/avs-vb/showthread.php?postid=326565#post326565) for 1080/60i and 1139X550 for 720p; this linked simplified table is derived from table 2.3 (http://www.atsc.org/news_information/papers/1995_acats/tsreport.pdf) at ATSC.org.

So, while many AVSers report crisper images from 1080/60i programming, 720p heavily based on 1080 sampling/downconversion boosts its resolution (although limited to 1280X720) and seems to 'sharpen' 720p. Presumably, 1080/60i derived from 4k oversampling/downconversion could be similarly enhanced.

'Sharpening' here mostly means boosting coarser lower resolutions/frequencies, the result of oversampling/downconversion. Sharper in this sense isn't necessarily higher resolution, (only better-contrast lower resolution). An Arri expert outlines this sharpness-resolution distinction in a ~7-MB pdf paper (http://www.arri.com/infodown/cam/broch/2008%2003-25a%204K+%20Technology%20Brochure.pdf) published (edited) in a recent SMPTE Imaging Journal issue. TVOD reposted/analyzed some of the Arri diagrams here (can't find them), but page 9 photos compare a 'sharp' image of a Arri-model camera with another, seemingly duller, image of the same camera but at higher resolution, revealed by the insets showing knob detail. Page 11 modulation transfer function (MTF) graphs--resolution vs modulation or contrast--of the two cameras shows the boosted coarser resolutions that make one image appear 'sharper'.

Perhaps similar comparisons can be made between live 1080i and 720p (downconverted) sports broadcasts. CBS's or NBC's live 1080i often has a certain higher-res but seemingly 'duller' appearance compared to, say, a 'sharp' 720p sports broadcast that seems to be using 1080 downconversion. But the 1080 images clearly show finer details, like faces or clothing in distant crowds or lettering on signs. -- John

Might be putting too fine a point on it, but BeachComber restored some of these images, too, in the DirecTV-rez thread (http://www.avsforum.com/avs-vb/showpost.php?p=14445203&postcount=19). He had them stored from subscribing to this thread, and reposted the images using Imageshack. Curiously, his restored D* MPEG4 horiz rez shot you zoomed in on is noticeably more distinct than yours above, making a resolvable 1440 reading more feasible, although both are ambiguous.Maybe because in that thread I posted a downsized version of the whole screen, which may appear sharper, whereas here I posted 100% crops from the camera, which displays the pixels very close up. But they are from the same raw file from the camera.

As I look at them more carefully I think you're right, 1440x1080 (as I originally said) would be slightly overstating the resolution. But it's certainly a step up from the 1280x1080 MPEG-2 days.

Although I have no proof, it sure looks like D* is employing some sort of blurring/denoising type algorithm to improve compressibility. The MPEG-4 channels have always looked a bit soft, though cleaner than the MPEG-2 channels.

A large number of progressive-scan displays don't properly remove 2:3 pulldown, so any resolution gained by going to 1080i on 24p content is effectively thrown out by the TVOk, if your TV is defective, that's a different matter. You could say the same thing about 1366x768 HDTVs that do a horrible job of scaling. But that's not a limitation inherent to 1080i.

My Sharp LCD does a fine job of IVTC, and detects it automatically.Then, you take into account providers like Comcast who rape the picture to fit it into 12Mbit/s, leaving 24p content in 1080i a blocky mess while leaving 24p content in 720p relatively block-free.D* uses less than that (best estimate, 7-8 Mbit) for the 1080i MPEG-4 channels and there is no blocking. Even really tough scenes like rain, flocks of birds, or fast detailed pans don't show any noticeable blocking.

Again, this is a flaw in the provider, not a flaw inherent to 1080i. Perhaps you should consider switching to D*.

Although I have no proof, it sure looks like D* is employing some sort of blurring/denoising type algorithm to improve compressibility. The MPEG-4 channels have always looked a bit soft, though cleaner than the MPEG-2 channels.
You probably have that backwards. I would bet any softening you detect is a product of the compression itself, not the other way around.

H.264 tends to soften prior to actually inducing macroblocking or artifacting as bit rate is reduced.

It depends on the use and the source of the material.

Editing of material from live events is much lower. Editing from source material from camera tapes tends to be higher, as does film (which I don't work with).



Yep - I guess it depends on your workflows and infrastructure, and timescales.



Of course, some systems are better than others about it. Having hardware-based controls for common tasks (such as dissolves and color correction) avoids a lot of software rendering than can affect image quality. In addition, some systems (like the Quantel EQs) work almost entirely in a compressed form (unlike the Quantel QEPs). Avid and Final cut (I believe) tend to also do better with that sort of thing despite the use of "soft" controls.


I have spent a bit of time working in the SD domain with 30Mbs MPEG Quantel servers editing stuff that would otherwise have been on 25Mbs DV codec kit. The Quantel stuff certainly doesn't do the pictures much damage - I'd say it probably survives better than in the previous Avid Newscutters/Media Composers and Beta SP suites. (16:9 Beta SP with BVW75s old enough to vote and buy a drink was not a recipe for great pictures though)


Of course, other variables like whether the material is being brought via a shared server or is ingested to the local machine can affect this, too. Server-based clips with higher data rates would make moving media almost impossible in a timely manner.

Yep - explaining the timescales required for live TV production (news, sport, entertainment, current affairs etc.) is always fun. People who aren't used to that environment never believe that 15 seconds can mean the difference between an item airing or never being seen...

The genres of production that you watch in 1080/60i native video on NBC, PBS and CBS will be the same genres of production that are shot 720/60p for ABC, Fox, ESPN.I'll take your word for it, since I only watch two shows on ABC and Fox and never ESPN, and the (non-live) shows I watch are of course shot at 24 fps.

But I have seen quite a number of documentaries on 1080i channels shot using native 1080i video, though never any documentaries on 720p channels shot using native 720p video.

All I see on 720p channels are downresed versions of telecined film, or even worse, downresed, poorly deinterlaced video full of judder and blending artifacts.

This is particularly bad on, say, 720p HISTHD, where they show quite a number of programs that were obviously captured at least in part at 1080i and then downresed and horribly deinterlaced and resized to 720p.

Don't misunderstand me; if there were real 720p/60 being shown for a majority of the time on these 720p channels, I would see the benefit. But right now they're just a waste.It has also been noted that many de-interlacers don't deliver the full potential resolution of 1080i - instead treating it as 540p constantly (and thus when showing stuff shot on HDCam or DVCProHD will only deliver a 1440x540 res picture - not a million miles from 1280x720).As I said in the post above - if the TV is defective, that's a problem with the TV, not a problem with the transmission format. There is a good selection of TVs that can properly deinterlace/IVTC and no reason not to pick one of them.

You probably have that backwards. I would bet any softening you detect is a product of the compression itself, not the other way around.

H.264 tends to soften prior to actually inducing macroblocking or artifacting as bit rate is reduced.A static resolution chart (e.g., HD Net's) requires almost zero bandwidth, so that can't be the case.

Although I have no proof, it sure looks like D* is employing some sort of blurring/denoising type algorithm to improve compressibility. The MPEG-4 channels have always looked a bit soft, though cleaner than the MPEG-2 channels.Here's a recent post (http://www.avsforum.com/avs-vb/showpost.php?p=14450512&postcount=20) from someone in the D*-rez thread, stating from a D* insider, that they're using 1920X1080 for MPEG4. Not unexpected. :). But again, with all the computer-stored/analyzed MPEG4 images, including your HDNet MPEG4 frame, it's not clear why someone hasn't found a 1920X1080 or 1440X1080 MPEG tag (or whatever appears) with analysis software. -- John

Here's a recent post (http://www.avsforum.com/avs-vb/showpost.php?p=14450512&postcount=20) from someone in the D*-rez thread, stating from a D* insider, that they're using 1920X1080 for MPEG4.Forgive me if I don't put much stock in vague hearsay from an unnamed D* "insider". If I can't see the visual proof, it isn't there.

As for analyzing tags and such, well, you're assuming that D* is transmitting in a matter conforming to a known standard (like H.264, etc.). However since it's their own proprietary network they can transmit in any format they like. Of course I'm not up on the technical details; is their encoding method well documented publicly?

Forgive me if I don't put much stock in vague hearsay from an unnamed D* "insider". If I can't see the visual proof, it isn't there.

As for analyzing tags and such, well, you're assuming that D* is transmitting in a matter conforming to a known standard (like H.264, etc.). However since it's their own proprietary network they can transmit in any format they like. Of course I'm not up on the technical details; is their encoding method well documented publicly?

Are any D* receivers modded for bitstream output - as others have been - using the 169time or R5000 systems?

If they have been then simply analysing the video streams will easily tell you.

This is possible on Sky HD in the UK - which uses DVB-S2, Videoguard and H264 (HD) and MPEG2 (SD) for broadcast. Early SD receivers can be persuaded to deliver a decrypted transport stream (which a few HD channels broadcast in DVB-S rather than S2 can be analysed from) via a digital interface slot (never used in anger) on the back. For DVB-S2 stuff you need to use a PC with a DVB-S2 receiver card, and a reverse engineered conditional access module (either hardware or software emulated). This is NOT to allow you to get free pay-TV - you still need a valid subscription and viewing card.

It is thus quite easy to work out what is going in on the UK HD satellite services.

Has anyone done this for D*. (Don't they also use Videoguard - or am I confusing my abbreviations?)

D*'s MPEG-4 1080i channels (excluding locals) do not have any significant blocking artifacts, even during heavy motion. Of course I suspect they're downresed (see above), but still, they don't have those artifacts whereas their MPEG-2 channels did.

I said that I have never seen a 1080i chan with no blocking. I've actually never seen an Mpeg-4 1080i chan. I should have been more specific in saying MPEG-2 1080i. I have little doubt that MPEG-4 AVC is capable of handling 1080i.

For the most common source, telecined 24 fps material, 1080i has a 225% bandwidth advantage over 720p.

No way. I don't buy that for 1 second. I have seen macroblocking on 1080i30 telecined material including on my local CBS affiliate which uses its entire spectrum. 1080i30 telecined is NOT the same as broadcasting @ 1080p24. And I BELIEVE from what I've heard is that CBS-KRQE has the best most modern equipment in this DMA.

Furthermore whats the rationale in calling it 1080i60? Isn't it standard to list the frame rate, and not the field rate?

My 2 cents on 720P (ok maybe 3):

First off video is two dimensional (H & V) so there is different resolution for each. There are 2/3 the pixels in each direction so the spacial difference is closer to 2/3 than 1/2 compared to 1080.

Second issue that the perception of sharpness has more to do with mid to upper mid frequency response than how high the uppermost frequency is. From the ARRI paper that John posted it can be shown that an image with flat response in the mid frequencies can look sharper than one with gradual rolloff but extends farther. 720P cuts off at the 2/3 point so it preserves much of the perception of sharpness.

CCDs have a gradual rolloff curve due to the light gathering area of the elements nearly touching each other. An optical low pass filter is also required. Optics also have losses. Telecine transfers generally use zooming to properly frame an image so the entire resolution of the sensors are not used. All these factors compromise the depth of modulation at the higher spacial frequencies. The filter to limit to 1280 pixels is much steeper.

A third issue is how 1080i cameras operate. When used in interlace mode the target pixels are made to be twice the vertical size of 1080p pixels by the use of CCD pixel grouping. The target pixels in one field overlaps the top and bottom of the row in between on other field. This grouping is necessary to keep diagonals smooth and to minimize vertical aliasing. These larger pixels also reduce the depth of modulation in the higher vertical frequencies. On the other hand little if any overlap is used for progressive formats so this loss is not experienced. Furthermore any motion in an interlaced system effectively lowers the vertical resolution in half, so it can be argued that 720P images can have greater resolution and depth of modulation than a 1080/60i formated image.

Another factor to be considered is the efficiency of progressive encoding vs interlaced encoding. While it's been argued here that such an advantage does not exist, it's still a widely held belief and appears to have support from empirical observations. While nearly being the same bitrate as 1080i (around 12% lower) it seems to be better with less artifacts than 1080i at a given bitrate. This can be attributed to pixels having tighter temporal relationship as a given pixel repeats position in half the time that interlaced ones do. Another factor may be that they are also always adjacent in the vertical axis where as interlaced are separated by the row in the opposite field. Interlace can use a modified DCT zig-zag readout pattern called alternate scan to help with efficiency, but the fact that it exists demonstrates a need for improvement. Decreased encoder efficiency can cause greater compression which means that some of the higher static spatial response of 1080i can be lost.

My own choice if we had to have just one or the other is still 1080i. I think the resolution on 24P material can be better conveyed. A single encode at ATSC bitrates can look very good as may have seen on channels like HDNet.

I'm not sure what over-compression is. Is it where artifacts become visible? As noted in this thread there are situations even under the best conditions where artifacts will occur. Blocking is probably the most noticeable artifact.

After years and years of evaluating the two, forum consensus makes it very clear there are many other factors affecting the end product, so many that in the end either 720p or 1080i are capable of great or poor HDTV. Making a generalization that one is 'better' than the other is simply wrong.Yeah that too.

Are any D* receivers modded for bitstream output - as others have been - using the 169time or R5000 systems?

If they have been then simply analysing the video streams will easily tell you.

This is possible on Sky HD in the UK - which uses DVB-S2, Videoguard and H264 (HD) and MPEG2 (SD) for broadcast. Early SD receivers can be persuaded to deliver a decrypted transport stream (which a few HD channels broadcast in DVB-S rather than S2 can be analysed from) via a digital interface slot (never used in anger) on the back. For DVB-S2 stuff you need to use a PC with a DVB-S2 receiver card, and a reverse engineered conditional access module (either hardware or software emulated). This is NOT to allow you to get free pay-TV - you still need a valid subscription and viewing card.

It is thus quite easy to work out what is going in on the UK HD satellite services.

Has anyone done this for D*. (Don't they also use Videoguard - or am I confusing my abbreviations?)
No such thing exists for D*'s MPEG4-capable receivers. E*'s boxes can be modded with an R5000 (capture is also possible with certain DVB cards since E*'s encryption has been cracked), but the resulting MPEG4 stream can't be edited with any of the tools commonly used for EU MPEG4 streams because E* is using a somewhat different format (PAFF vs. MBAFF I believe).

Furthermore whats the rationale in calling it 1080i60? Isn't it standard to list the frame rate, and not the field rate?

Think of it as a distinction of frequency (Hz) at which something new happens, rather than simply "frames per second". Under high motion, there is a pretty big difference between 30 Hz and 60 Hz motion, regardless of whether it is interlaced or progressive. Try watching a video of a sports game (basketball, football?) or a nascar race encoded at "30 Hz progressive" vs. "60 Hz interlaced". There's a big difference, and that is why it is important to retain that distinction.

"i60" tells you exactly what the content is...interlaced motion at 60 Hz,

"p30" used in its place would be needlessly ambiguous...you don't know if it is simple progressive frames that show motion at 30 Hz intervals or is it actually interlaced fields that depict motion in 60 Hz intervals, where 2-fields are informally considered a "frame" (which is also not an exactly accurate distinction),

"i30" would be something entirely different...it's interlaced, but the motion would be no better than an actual 30 fps video.

In the hdtv broadcast realm, there really is no such thing as "p30", anyway. It is either 60i or 60p. The only exception is if somebody has taken 60i material and deinterlaced it (yes, there are some exceptions- I know), which would then yield an actual 30 fps (which would likely be rendered in repeat-frame cadence for 60p broadcast). Once that has been done, it may be in a nice progressive frame result, but it will never be able to duplicate the motion rendition of the original 60 Hz source.

Now if you move to videogames, you can get all sorts of possibilities. Sometimes the game is output @ 60 Hz, but in actuality, the motion is only updating at 30 Hz.

There was some debate a while back regarding the best nomenclature for video standards. It was sort of agreed upon that we would use:

<active lines> / <frames per sec for progressive or fields per sec for interlace><i or p>

For example the two common standards in the US would be 1080/60i and 720/60p. Combinations would include 480 (or 483), 575, 720, 1080 for active lines (1035 and 1250 being outdated) and 50i, 60i, 24p, 25p, 30p and 60p for fps. The .1% pulldown is usually not considered when stating these standards here. Even though 525 and 625 are still identifiers on professional equipment they don't see much use here.

Another method is: <active lines><i or p><frame rate>

For example 1080i30 would mean the same as 1080/60i. 1080i60 could imply a field rate of 120hz. Author Charles Poynton uses this approach.

I think the confusion (or debate) comes about because we instinctively want to put things in nice, simple categories. Unfortunately, interlaced video is somewhat of a strange beast that defies easy categorization. Depending on the nature of the scene, it might be delivering a complete image resolution of 1080 vertical pixels, or it might be delivering half that, but updating the motion in smooth 60 Hz intervals, or it might be delivering anything in between those 2 points as far as resolution. It could be one thing in one moment, and something else in another, or even both at the same time in different parts of the image, depending on the sophistication of the deinterlacing operation. No one designation really fits the bill, all of the time, and if we settle on one hard designation, chances are it will be wrong most of the time, anyway. ;)

I prefer the notion that "1080i60" is simply stating the "best case scenario" that is possible over 2 metrics- resolution and refresh. In the best case scenario for resolution (a still or nearly motionless image), it will indeed be 1080 pixels of vertical resolution. In the best case scenario for refresh, it will indeed render smooth motion at 60 Hz. There is also an assumed presumption that what you actually get will always be somewhere in between those 2 goalposts, based on the nature of the image and the deinterlacing scheme.

You can't really consider it half-resolution at x progressive fps or full-resolution at half-progressive fps, because that just doesn't cover what it can actually deliver. It is certainly not restricted to producing a half-resolution image, and never does a truly progressive frame exist in the stream. It can certainly be shoehorned into such a condition (with the expected degradations), but that is still a different "thing" then what you started with.

I said that I have never seen a 1080i chan with no blocking. I've actually never seen an Mpeg-4 1080i chan. I should have been more specific in saying MPEG-2 1080i. I have little doubt that MPEG-4 AVC is capable of handling 1080i.That's what I was saying. Maybe you should switch to D* if Comcast is so bad, because D*'s 1080i channels are MPEG-4 and do not have blocking.No way. I don't buy that for 1 second. I have seen macroblocking on 1080i30 telecined material including on my local CBS affiliate which uses its entire spectrum.Are you watching OTA? OTA uses MPEG-2, and even at the theoretical top bandwidth, it can have plenty of blocking. 1080i30 telecined is NOT the same as broadcasting @ 1080p24.Yes it is. You can take any telecined program and, errors aside, precisely reconstruct the original 24p source using a very simple algorithm. Most TVs have this built in, or have a switch to turn it on or off.

Now, telecined material may not compress identically to (i.e., as cleanly as) a 1080p/24 broadcast, because interlacing throws a little wrench in the compression, but compression issues aside, the detail and the image is the same.

You can take any telecined program and, errors aside, precisely reconstruct the original 24p source using a very simple algorithm. Most TVs have this built in, or have a switch to turn it on or off.

Now, telecined material may not compress identically to (i.e., as cleanly as) a 1080p/24 broadcast, because interlacing throws a little wrench in the compression, but compression issues aside, the detail and the image is the same.
Interlaced broadcasts (including 24p content pulled down to 60i) are usually filtered vertically to prevent artifacts caused by fine detail when watching on interlaced displays. It is not possible to recover the detail lost because of this filtering. Proper pulldown removal will give a good approximation of the original 24p content, but it's not the same. Also, pulldown removal is far more complex than blindly matching fields since editing is often done after pulldown is applied. Each cut has the potential to throw off the pattern, and the device need to be able to detect this and either find a better match or simply deinterlace the frame to prevent combing from appearing in the output.

That's what I was saying. Maybe you should switch to D* if Comcast is so bad, because D*'s 1080i channels are MPEG-4 and do not have blocking.

You clearly did not see Hollywood Green on Planet Green the other night.

but the resulting MPEG4 stream can't be edited with any of the tools commonly used for EU MPEG4 streams because E* is using a somewhat different format (PAFF vs. MBAFF I believe).

Hmm - AIUI Sky HD uses PAFF but BBC HD uses MBAFF - so both flavours should be knocking around in Europe.

However the only - very crude - method of trimming recordings I have currently is to use TS Remux, which allows you to trim the start and end from a transport stream file (and remove the superfluous audio description streams to save a bit of space) That said I haven't looked for H264 editing tools in earnest.

Don't misunderstand me; if there were real 720p/60 being shown for a majority of the time on these 720p channels, I would see the benefit. But right now they're just a waste.As I said in the post above - if the TV is defective, that's a problem with the TV, not a problem with the transmission format. There is a good selection of TVs that can properly deinterlace/IVTC and no reason not to pick one of them.

Yep - though whilst good de-interlacing is available - it is never going to be able to fully differentiate between some types of content as a result of the spectrum folding (vertical detail vs temporal detail) inherent in interlacing, and many fall over with mixed content (where you have both native interlaced and native progressive content on-screen at the same time)

The best de-interlacers currently around use Phase Correlation AIUI (mainly used to extract the maximum resolution from interlaced SD - including stuff with mixed sources - such as interlaced moving captions over film content etc.) - but these are not available at a consumer price point (and the techniques are patented and licensed AIUI)

Progressive capture and native broadcast removes this process - ensuring full quality for all (and removing the requirement to chuck a lot of processing into every display).

That said - I'm not advocating 720p over 1080i - just pointing out that there are still major issues with continued use of interlaced capture (interlaced broadcasts of pure 24p content are a different matter)

There was some debate a while back regarding the best nomenclature for video standards. It was sort of agreed upon that we would use:

<active lines> / <frames per sec for progressive or fields per sec for interlace><i or p>

For example the two common standards in the US would be 1080/60i and 720/60p. Combinations would include 480 (or 483), 575, 720, 1080 for active lines (1035 and 1250 being outdated) and 50i, 60i, 24p, 25p, 30p and 60p for fps. The .1% pulldown is usually not considered when stating these standards here. Even though 525 and 625 are still identifiers on professional equipment they don't see much use here.

Another method is: <active lines><i or p><frame rate>

For example 1080i30 would mean the same as 1080/60i. 1080i60 could imply a field rate of 120hz. Author Charles Poynton uses this approach.

Yep - the two standards are widespread.

If the letter (i or p) is the last character in the description then the number before it (usually 24,25,30,50 or 60 (*)) indicates the field rate if interlaced and the frame rate if progressive.

If the letter is before the second set of numbers then it is always the frame rate that is referenced, irrespective of whether the source is interlaced or progressive. (This is the EBU standard)

I prefer the former - where you list the field rate for interlaced and the frame rate for progressive - as interlaced native content has frames consisting of two fields captured at different moments in time. As a result a 30 frame per second INTERLACED native source is delivery 60 images per second captured at 60 different points in time, and thus has motion rendition similar to a 60 frame per second progressive native source NOT a 30 frame per second source.

People who see interlaced systems defined purely in terms of their frame rate (rather than their field rate) don't appreciate that the two fields that make a frame can (and are in native interlaced video) be captured at different times.

I find the 1080/60i format more useful when discussing with non-experts - but can see why the 1080i30 equivalent is also useful in an academic context.

Interlaced broadcasts (including 24p content pulled down to 60i) are usually filtered vertically to prevent artifacts caused by fine detail when watching on interlaced displays. It is not possible to recover the detail lost because of this filtering. Proper pulldown removal will give a good approximation of the original 24p content, but it's not the same. Also, pulldown removal is far more complex than blindly matching fields since editing is often done after pulldown is applied. Each cut has the potential to throw off the pattern, and the device need to be able to detect this and either find a better match or simply deinterlace the frame to prevent combing from appearing in the output.
Interesting point (vertical filtering), especially here where I'm still using a CRT RPTV bought 8 years ago. So, with my cable STB outputting 1080/60i for everything, I'm viewing 24p material with 2-3 pulldown plus the vertical filtering--smearing lines vertically: an anti-twitter measure to minimize the twittering or apparent 'jumping' of higher-contrast fine details that would otherwise be exaggerated when displayed (by CRTs) first by one TV field then the other field.

Someone here recently implied such vertical filtering may not be taking place that much these days with 1080/60i 2-3 pulldown. Not sure how that could be checked. But the recent introduction of 1080p VOD movies for both Echostar and DirecTV (see current threads), it would seem, should permit 24p delivery without the extra vertical filtering that must reduce some PQ.

Recall someone posting in the E* Turbo-HD thread that his 1080p movie order was blocked because his display wasn't designed to accept 24p. So, it looks like you need the right DBS set-top box and, ideally it seems, a display that accepts direct 24p inputs and converts 24p into even multiples, such as 48, 72, and up frames per second to prevent display-created judder.

AIUI, many displays still aren't creating this even-24p multiple, but instead performing 2-3 pulldown on the inverse-pulldown 24p frames (from 1080/60i delivery--or perhaps from 1080/24p DBS delivery, or Blu-ray) and creating judder, apparently not noticed by some. -- John

EDIT: Here's a thread, BTW, that's apparently tracking displays that accept direct 24p and display at even multiples:
http://www.avsforum.com/avs-vb/showthread.php?t=942145

That's what I was saying. Maybe you should switch to D* if Comcast is so bad, because D*'s 1080i channels are MPEG-4 and do not have blocking.Are you watching OTA? OTA uses MPEG-2, and even at the theoretical top bandwidth, it can have plenty of blocking.Yes it is. You can take any telecined program and, errors aside, precisely reconstruct the original 24p source using a very simple algorithm. Most TVs have this built in, or have a switch to turn it on or off.

Now, telecined material may not compress identically to (i.e., as cleanly as) a 1080p/24 broadcast, because interlacing throws a little wrench in the compression, but compression issues aside, the detail and the image is the same.

I said in an earlier post that I am OTA only (at the moment, probably will get 1 of the sat services soon). Remember, I said (in an earlier post), that 720p is better within the confines of a 19Mbps MPEG-2 stream, I did NOT say it was ALWAYS better. If a channel did nothing but broadcast low-action programs, ie a lot of documentaries then 1080i may be the correct choice. For every other channel going out on MPEG-2, 720p is clearly the correct choice IMO.

I'm a HTPC user, and yes it does do proper IVTC, I can see a clear 3:2 pattern.

Interlaced broadcasts (including 24p content pulled down to 60i) are usually filtered vertically to prevent artifacts caused by fine detail when watching on interlaced displays. It is not possible to recover the detail lost because of this filtering.How much detail are we talking about here? It's not 55% of the detail (what you lose by going to 720p/24), is it?Also, pulldown removal is far more complex than blindly matching fields since editing is often done after pulldown is applied.Generally this is only a problem if you have CGI or other graphics (e.g., text or credits) added at the native broadcast rate instead of the film rate. Some older science fiction shows do this and it is extremely annoying because those frames aren't telecined or interlaced but some obnoxious hybrid. But I haven't noticed it on more recent shows, and on shows that don't use CGI it's not a problem. Though I have seen shows with telecined film in some scenes, interlaced video in others. Each cut has the potential to throw off the pattern, and the device need to be able to detect this and either find a better match or simply deinterlace the frame to prevent combing from appearing in the output.Yes, poor editing can cause momentary lapses, but your TV should pick up on it again within three frames. Poor editing can mess up the output in many ways.

You clearly did not see Hollywood Green on Planet Green the other night.I've never even heard of that channel. :D

But the blocking could be from the source feed, not D*'s fault, unless D* is being stingy with the bit rate for that particular channel. Typically there is no blocking even on high motion scenes.

Generally this is only a problem if you have CGI or other graphics (e.g., text or credits) added at the native broadcast rate instead of the film rate. Some older science fiction shows do this and it is extremely annoying because those frames aren't telecined or interlaced but some obnoxious hybrid.


It is common this side of the pond. Scrolling 50i native video captions are often used over 25p native film or video content. Some shows mix 25p with 50i - Top Gear (a very popular UK car based show) uses 25p for location films, but 50i for their links and celebrity track races. Quite a few shows over here mix 25p and 50i. It is common in sports coverage - where the live stuff is 50i, but the set-up packages and archive are 25p treated to give them a film-look.

Any decent de-interlacer has to be able to cope with mixed content - say a split screen between video and film content, or video and film content overlaid. (During end credit promos it isn't unusual to have shrunk video overlaid over film)

Assuming that a single de-interlacing technique will work across a frame - such as 3:2 detection and removal is antiquated and just won't deliver decent results. Broadcast quality de-interlacers used for SD to HD upconversion are now designed to cope with mixed content - and de-interlace on a pixel-by-pixel basis using phase correlation to track motion AIUI (S&W Quasar for example)


But I haven't noticed it on more recent shows, and on shows that don't use CGI it's not a problem. Though I have seen shows with telecined film in some scenes, interlaced video in others.


It still happens - in HD and SD shows.


Yes, poor editing can cause momentary lapses, but your TV should pick up on it again within three frames. Poor editing can mess up the output in many ways.

It isn't always "poor editing" - if you are making a live show - say Letterman or Leno - then your studio cameras are 60i, but you could be playing in clips of a 24p movie with 3:2. Result - as you cut to the clip the TV has to change modes from video to film, and then back again on the transition back. If the de-interlacer is a basic one - you'll likely get a few frames of nastiness either side of the cut.

But then it can get worse. Say you decide to put a 60i native camera in a box over this movie clip - because the clip is embarassing or funny and you want to see the guest's reaction... Result - 60i native and 24p in 60i with 3:2 on-screen at the same time. You need a better solution than just a crude 3:2 cadence detection and removal to deliver decent results on that kind of content - you can't apply the same de-interlacing algorithm universally across the image...

How much detail are we talking about here? It's not 55% of the detail (what you lose by going to 720p/24), is it?
No, but when combined with a progressive display that can't reverse pulldown and simply bob deinterlace to 60p (pulldown removal on the HDMI input wasn't a common feature until the 2007 model year), it's something on the order of 60% resolution lost before taking into account loss of effective resolution due to Nyquist, additional compression, and all the other variables that John Mason likes to reference. When viewed on these TVs, which comprise a large portion of the installed base of HDTVs, the 1080i format loses any inherent advantage over 720p on 24p content.

I said in an earlier post that I am OTA only (at the moment, probably will get 1 of the sat services soon). Remember, I said (in an earlier post), that 720p is better within the confines of a 19Mbps MPEG-2 stream, I did NOT say it was ALWAYS better.

480p is even better than 720p at 19Mbps. I never saw any blocking from Fox's "Digital Widescreen", not even during the Super Bowl. It was the best SD video I have ever seen.

So when it comes to video compression, it's a matter of what you like or hate the most.

No, but when combined with a progressive display that can't reverse pulldown and simply bob deinterlace to 60p (pulldown removal on the HDMI input wasn't a common feature until the 2007 model year), it's something on the order of 60% resolution lost before taking into account loss of effective resolution due to Nyquist, additional compression, and all the other variables that John Mason likes to reference. When viewed on these TVs, which comprise a large portion of the installed base of HDTVs, the 1080i format loses any inherent advantage over 720p on 24p content.Ok. But a failure of your TV should not be construed as a failure of the format. You could just as well argue that 720p is flawed because most TVs are not 720p native resolution, and some have poor scalers that cause artifacts in the resizing process.

My Sharp TV handles 1080i just fine (IVTC works, and I don't get any wierdness in cuts or transitions), and there is a big difference between 720p and 1080i detail.

In other words, even after all the variables you mentioned, when I tune to a 1080i channel displayed telecined content I see a much sharper, better resolved image than I do on the 720p channels. And blocking is no longer a significant issue with MPEG-4 channels, so what's the benefit?

480p is even better than 720p at 19Mbps. I never saw any blocking from Fox's "Digital Widescreen", not even during the Super Bowl. It was the best SD video I have ever seen.

So when it comes to video compression, it's a matter of what you like or hate the most.

Yep - and that was 480i originated and network distributed, de-interlaced at the station to 480p prior to encoding.

AIUI nothing produced by Fox in anger was ever shot 480/60p (apart from a few trials with some early Philips prototype cameras) - though some episodic drama was shot and posted 480/24p (The X-Files?) and delivered in 480/60i such that the Fox local encoders could de-interlace properly to 480/60p (with 3:2 field-pulldown removal followed by 3:2 frame repetition?) So events like the SuperBowl will have been through a 480/60i process and de-interlaced (albeit by a broadcast rather than consumer grade de-interlacer)

(AIUI The 16:9 episodes of The X-Files were shot on film, telecined in 576/48i with a clean 2:2 cadence, edited in 24p, and then the master tapes were played out at either 576/50i 16:9 for 50Hz audiences and 480/60i with 3:2 pulldown for 60Hz audiences? I may be wrong on this - but at least one US show did it I think. It was a post solution promoted by Quantel in the 90s - as it allowed 24p editing, no resolution loss for 576/50i viewers and no compromises for 480/60i viewers either...)

I've never even heard of that channel. :D

But the blocking could be from the source feed, not D*'s fault, unless D* is being stingy with the bit rate for that particular channel. Typically there is no blocking even on high motion scenes.

Clearly you do not know of the 2 HD non-premium channels D* added on Thursday then.

Regardless, as we are on the same side of this battle, your statement that "D*'s 1080i channels are MPEG-4 and do not have blocking" sounds like the arguement if a tree falls in the woods and no one is around does it make a sound.

Just because you have not seen blocking on D*'s MPEG4 channels =! it does not exists.

Even in the response above, you went for "do not have blocking" to "typically" so even you seem to be backing off your earlier statement.

Just because you have not seen blocking on D*'s MPEG4 channels =! it does not exists.Ok, let me clarify, by "not having blocking" I was responding to the guy who was saying all his 1080i channels were a blocky mess.

I am not suggesting that D*'s MPEG-4 will never have blocking any under circumstances, just that, in normal viewing, there's no significant blocking, i.e., it's not a blocky mess.

You can easily see the difference by flipping between HD Theater MPEG-2 and MPEG-4.

How much detail are we talking about here? [From vertical filtering] It's not 55% of the detail (what you lose by going to 720p/24), is it?

No, but when combined with a progressive display that can't reverse pulldown and simply bob deinterlace to 60p (pulldown removal on the HDMI input wasn't a common feature until the 2007 model year), it's something on the order of 60% resolution lost before taking into account loss of effective resolution due to Nyquist, additional compression, and all the other variables that John Mason likes to reference. When viewed on these TVs, which comprise a large portion of the installed base of HDTVs, the 1080i format loses any inherent advantage over 720p on 24p content.

Engineer Greg Rogers (http://archive.avsforum.com/avs-vb/showthread.php?postid=1097351#post1097351) had a frequently linked article outlining resolution and FP CRT specs that's no longer available, but he pointed out that added anti-twitter filtering can trim effective vertical resolution by 50% when coupled with the Kell factor (~0.7 X line count). Here's an archived excerpt (http://archive.avsforum.com/avs-vb/showthread.php?postid=2293916#post2293916). Vertical resolution is variable with the rate of vertical motion, and earlier above I posted static/dynamic resolution measurements (800, 400 lines for 1080) made during ATSC approval testing.

Since 720p-delivered 24p movies are generally obtained from downconverting 1080/24p master tapes, presumably boosting the 720p vertical resolution typically seen with standard (non-1080) sampling (also outlined earlier), it seems vertical effective resolution of the two HD formats could approximate each other. Video-expert/720p-advocate Joe Kane puts it this way at his site: (http://www.videoessentials.com/D_TheaterQA.php)
The majority of movies that Hollywood is making available in the D-Theater format are mastered in 1080p then converted to 1080i before they are mastered and duplicated. The irony of this is that the majority of 1080p film transfers only have enough resolution to support the horizontal capability of 720p. Horizontal resolution of most film masters in 1080p is in the area of 800 to 1300 lines. The horizontal capability of 720p is 1280 lines, the top end of what’s on the 1080p master. An argument could be made that even with the 30% vertical filtering in the 1080p to 1080i conversion the vertical resolution of 1080i is still slightly better than 720p. It’s 756 lines as opposed to 720 lines. Interlaced artifacts in 1080p/24 frame material converted to 1080i/60 are not nearly as bad as video material initially created in 1080i/60.Kane's 756 lines just above results from factoring in Kell reduction: (1080 X 0.7 = 756), but apparently not vertical anti-twitter filtering. This post (http://www.avsforum.com/avs-vb/showthread.php?p=9314235&&#post9314235) elaborates on Kane's 24p resolution comments.

On my 1080i CRT RPTV screen, despite the filtering reducing both effective horizontal and vertical resolution of movies, they usually appear slightly crisper from 1080/60i delivery compared to 720p. Programming other than 24p frequently appears more detailed, likely from the extended horizontal resolution--even though it's also restricted along the delivery chain. Compression undoubtedly factors into the reductions, but as mentioned before it doesn't result in significant blocking artifacts from my cable-only HD viewing (NYC's TWC). Significant means it might only appear for a few seconds when cameras are switched during a live broadcast--or other insignificant momentary production glitches. And no, doubt that I'm not seeing constant MPEG blocking because my 9"-CRT RPTV can't resolve it. :-) -- John

You guys are getting too technical for me, but the Olympics look pretty good over the NBC cable channels over Dish, but are pretty much unwatchable via OTA via KKCO. They have the CW going in SD on 11.2, but something very wrong is going on here. Anybody can see how bad it is. Even slow movement causes massive macroblocking.

I've speculated that the Kell factor has to do with the overlap of the scan lines between fields which is necessary in interlace for smooth diagonals and minimize vertical aliasing. With such overlap I don't think anti-twitter filters are required. If anti-twitter filters are used for progressive to interlace conversion (I'm not sure if they normally are) then they would be doing something similar to the line overlap. Therefore I don't think the two factors need to be combined.

Older fly spot telecines had real twitter issues because as CRTs got sharper there could be a gap between the scan lines (especially in 525 cinemasope). One could reduce that by misadjusting the astigmatism a bit. The line height on CCD telecines is defined by the integration time so the rows nearly touch each other.

Anybody can see how bad it is. Even slow movement causes massive macroblocking.Anybody except the sales oriented management: Problem? No problem here! Our signal is all digital so it can't have any problems. Must be your TV.

Here's more on the combined resolution-reducing effects of the Kell factor and vertical filtering (anti-twitter) discussed just above. There's a good diagram (figure 3) illustrating reductions from the format resolution in this recent pdf paper (http://www.ebu.ch/en/technical/trev/trev_308-hdtv.pdf) by European Broadcasting Union engineer Hans Hoffmann. The figure captions and text about the diagrams (appendix A) outline the Kell factor (0.7 X line count) and anti-twitter filtering for interlaced delivery (Hoffmann's 'interlace factor'). In addition, besides the horizontal reductions of resolution from Nyquist reconstruction filtering of sampled HDTV images, which I outlined with references earlier above , (http://www.avsforum.com/avs-vb/showpost.php?p=14450656&postcount=42) there's the resolution loss from standard lossy-MPEG bit rate reductions (compression)--or image artifacts from excess reductions (OP topic). -- John

You guys are getting too technical for me, but the Olympics look pretty good over the NBC cable channels over Dish, but are pretty much unwatchable via OTA via KKCO. They have the CW going in SD on 11.2, but something very wrong is going on here. Anybody can see how bad it is. Even slow movement causes massive macroblocking.

I don't see massive macroblocking, but I see normal macroblocking like I notice on all the HD channels from FIOS, COmcast and OTA.(and when I had DirecTV HD from MPEG2)

While my 5 year old RCA Scenium DLP was in service I never saw blocking on HD programming. About 10 days ago, though I put a new 60 inch Pioneer Kuro 6020 into service and I now sometimes see it. Last night I watched Transformers, which I had recorded on my TiVo S3 from HBO and there was noticeable blocking in every scene with fast movement. It didn't take long for BD at 1080p to spoil me.

I now assume that the artifacts were there all the time but I just didn't see them on my old RCA.

That [being there the whole time] is usually the case. Even if you go back to check out your RCA, you might actually start seeing them there, as well. :eek:

It's like the first time you see bigfoot...then you start seeing bigfoot everywhere you go! ('kay, that's a weird statement)

I hope I don't see a bunch more blocking with my new 60" Kuro than I did with the old 50 inch RCA. I am hopeful that I will be all right. I watched Monday Night Football on ESPN last night and my recordings of The Closer and Weeds earlier today. Neither showed any signs of the kind of artifacting I saw in HBO's Transformers telecast. In fact, Transformers was the first instance of that sort of thing I had noticed in the 10 days I have had the Kuro up and running.

Another problem with HBO's Transformers was their execrable practice of trimming 2.35:1 movies, which Transformers is, to 16:9. It often made me claustrophobic. The closeups looked particularly cramped.

Another problem with HBO's Transformers was their execrable practice of trimming 2.35:1 movies, which Transformers is, to 16:9. It often made me claustrophobic. The closeups looked particularly cramped.

You're too close to the screen! :)

You're too close to the screen! :)
Of course! Why didn't I think of that or, better yet, just have got another 50 inch HDTV. :)

Here's more on the combined resolution-reducing effects of the Kell factor and vertical filtering (anti-twitter) discussed just above. There's a good diagram (figure 3) illustrating reductions from the format resolution in this recent pdf paper (http://www.ebu.ch/en/technical/trev/trev_308-hdtv.pdf) by European Broadcasting Union engineer Hans Hoffmann. The figure captions and text about the diagrams (appendix A) outline the Kell factor (0.7 X line count) and anti-twitter filtering for interlaced delivery (Hoffmann's 'interlace factor'). In addition, besides the horizontal reductions of resolution from Nyquist reconstruction filtering of sampled HDTV images, which I outlined with references earlier above , (http://www.avsforum.com/avs-vb/showpost.php?p=14450656&postcount=42) there's the resolution loss from standard lossy-MPEG bit rate reductions (compression)--or image artifacts from excess reductions (OP topic). -- JohnI don't know if I agree with the charts as it shows interlace vertical resolution as being half the line count. This is true with motion and also true on a given field, but static vertical resolution should be higher albeit reduced by the larger vertical size of the capture pixel. I also wonder if they used pixel paring when they converted from 50p to 50i. Also the charts show a linear rolloff while I would expect to see something more resembling a sinc curve. However the point is made that with interlace the vertical response rolloff is faster and limited to field resolution with motion.

There were a couple of statements that stood out to me. One was:

"In the presentation of uncompressed sequences, the delegates reported difficulties in seeing difference between the three formats – even at a viewing distance of 3h. But when the compressed images were shown, the viewers did notice differences in the visibility of compression artefacts. Depending on the viewing distance and scene content, the artefacts became visible to a greater or lesser extent and, with few exceptions, the following were reported:

The 720p/50 format showed better image quality than the 1080i/25 format for all sequences and for all bitrates;

With decreasing bitrate in the compressed domain, the difference between the 720p/50 and 1080i/25 format became more marked;

The 1080p/50 format was rated equal or better than 720p/50 for the higher bitrates – the extent depending on the test sequence. However, 720p/50 was rated better than 1080p/50 at the lower bitrates."

Later on this difference was explained:

"Although different interlacing techniques are possible, roughly half the vertical-temporal information compared to 1080p/50 is removed. Consequently, an encoder has less information available to make intelligent decisions for compression and it is necessary to make more approximations which become visible as artefacts."

While the pixel count in a given time frame for 720p is about 11% less than 1080i, I think this paper supports the observations that progressive formats are more efficiently encoded compared to interlaced and that at a given bitrate 720p can have less artifacts compared to 1080i. At least in this test case the source material was much more controlled than just comparing different networks and channels.

The 720p/50 format showed better image quality than the 1080i/25 format for all sequences and for all bitrates;
Were deinterlacing methods mentioned or was it viewed on a CRT?

The deinterlacer in my new HDTV makes the bob deinterlacer in my cable box and my four year old HDTV look awful. What a huge difference! It's difficult to find any interlacing artifacts on my new set although I can still find them in rare cases. For example the bottom line of a tiny China flag in the scoreboard during a volleyball game kept disappearing and reappearing.

I don't know if I agree with the charts as it shows interlace vertical resolution as being half the line count. This is true with motion and also true on a given field, but static vertical resolution should be higher albeit reduced by the larger vertical size of the capture pixel. I also wonder if they used pixel paring when they converted from 50p to 50i. Also the charts show a linear rolloff while I would expect to see something more resembling a sinc curve. However the point is made that with interlace the vertical response rolloff is faster and limited to field resolution with motion.
Yes, when Hoffmann's article (http://www.ebu.ch/en/technical/trev/trev_308-hdtv.pdf)appeared a while back in SMPTE's Imaging Journal I recall writing a few posts here about his appendix A diagrams, too, but mostly puzzling about the EBU's conclusions that 720p was preferable to 1080i.

Noticed that his appendix diagrams give the vertical axis in cycles per pixel height (horiz. rez in fig. 3 is c/pw), so his fig2 ~260 reading, presumably, would be 260X2=520 lines effective vertical resolution including a 0.7 Kell factor (1080 X 0.7)= 756. That 756 is further reduced by his 'interlace factor' or vertical filtering. If you assume a 30% anti-twitter filtering (30% from 756) that reduces effective vertical resolution to ~530, or about 520 lines. Yes, that differs from ATSC approval testing which measured 800 lines for 1080 static and 400 lines dynamic with a 5-rpm pattern ( table 2.3 (http://www.atsc.org/news_information/papers/1995_acats/tsreport.pdf) and this simplified table (http://archive.avsforum.com/avs-vb/showthread.php?postid=326565#post326565)). Also, Hoffmann's appendix text has this comment about Kell factor: "This factor was based on CRT measurements, and ideally would be measured in a non-CRT environment." His paper was peer reviewed, presumably for SMPTE publication, but Kell-factor data based on current CCD/CMOS cameras and fixed-pixel displays would be interesting to see.

I've no idea whether 30% anti-twitter filtering is accurate for most interlaced video delivery--although I've read overall filtering (h and v) for DVDs at least varies with individual producers, so assume 30% isn't absolute. Here's an extract (http://www.avsforum.com/avs-vb/showthread.php?p=9314235&&#post9314235)from Joe Kane comparing 720p and 1080i vertical resolution: Joe Kane: "An argument could be made that even with the 30% vertical filtering in the 1080p to 1080i conversion the vertical resolution of 1080i is still slightly better than 720p. It’s 756 lines as opposed to 720 lines." But as mentioned earlier the figures are still unclear since 1080 X 0.7 Kell factor only = 756 lines. And 720 lines given for 720p seems to be assuming 1080 oversampling and downconversion fully boosting typical ~550-line static (ATSC measured) 720p vertical resolution resulting from standard sampling with the Kell factor applied.
There were a couple of statements that stood out to me. One was:

"In the presentation of uncompressed sequences, the delegates reported difficulties in seeing difference between the three formats – even at a viewing distance of 3h. But when the compressed images were shown, the viewers did notice differences in the visibility of compression artefacts. Depending on the viewing distance and scene content, the artefacts became visible to a greater or lesser extent and, with few exceptions, the following were reported:

The 720p/50 format showed better image quality than the 1080i/25 format for all sequences and for all bitrates;

With decreasing bitrate in the compressed domain, the difference between the 720p/50 and 1080i/25 format became more marked;

The 1080p/50 format was rated equal or better than 720p/50 for the higher bitrates – the extent depending on the test sequence. However, 720p/50 was rated better than 1080p/50 at the lower bitrates."

Later on this difference was explained:

"Although different interlacing techniques are possible, roughly half the vertical-temporal information compared to 1080p/50 is removed. Consequently, an encoder has less information available to make intelligent decisions for compression and it is necessary to make more approximations which become visible as artefacts."

While the pixel count in a given time frame for 720p is about 11% less than 1080i, I think this paper supports the observations that progressive formats are more efficiently encoded compared to interlaced and that at a given bitrate 720p can have less artifacts compared to 1080i. At least in this test case the source material was much more controlled than just comparing different networks and channels.A number of the European studies seem to conclude 720p is better than 1080i, which contrasts with many observations here from AVSers viewing 1080i vs 720p for years. -- John

I hope I don't see a bunch more blocking with my new 60" Kuro than I did with the old 50 inch RCA. I am hopeful that I will be all right. I watched Monday Night Football on ESPN last night and my recordings of The Closer and Weeds earlier today. Neither showed any signs of the kind of artifacting I saw in HBO's Transformers telecast. In fact, Transformers was the first instance of that sort of thing I had noticed in the 10 days I have had the Kuro up and running.

Follow the Kuro and other threads in the plasma forum but haven't spotted mention of such MPEG-originated artifacts. Noticed a "crawling ants" thread by Kuro owners there complaining about visible spatial dithering (plasma pulse width modulation) seen near their screens, but that seems different from MPEG compression artifacts spoiling HD viewing . Concerned reading about such artifacts here since I'm awaiting the next-generation so-called 10-lumen plasmas (http://www.avsforum.com/avs-vb/showthread.php?t=890780) before phasing out my CRT RPTV.

HBO and a few other sources used to use MPEG repeat flags to conserve bit rate space, so perhaps that's tied in with Kuro's video processing. Borrowing an external video processor might show different results. Setting sharpness controls too high results in ringing and perhaps enhances higher-resolution blocking artifacts, too. -- John

Follow the Kuro and other threads in the plasma forum but haven't spotted mention of such MPEG-originated artifacts. Noticed a "crawling ants" thread by Kuro owners there complaining about visible spatial dithering (plasma pulse width modulation) seen near their screens, but that seems different from MPEG compression artifacts spoiling HD viewing . Concerned reading about such artifacts here since I'm awaiting the next-generation so-called 10-lumen plasmas (http://www.avsforum.com/avs-vb/showthread.php?t=890780) before phasing out my CRT RPTV.

HBO and a few other sources used to use MPEG repeat flags to conserve bit rate space, so perhaps that's tied in with Kuro's video processing. Borrowing an external video processor might show different results. Setting sharpness controls too high results in ringing and perhaps enhances higher-resolution blocking artifacts, too. -- John
I don’t think that my settings were the problem. I have been using Movie mode and, mostly, the settings recommended by D-Nice in the Kuro threads. In fact, I have been following very closely the 9G Kuro threads.

For whatever reason, I have not seen any of the kind of artifacting I saw while watching Transformers on any other program but Transformers, even other HBO programs, such as Generation Kill. Most of us who have bought the 9G Kuros have been thrilled by their PQ. The sheer resolution of CRT may never be matched on a pixel by pixel basis but the huge displays now available in the Kuro line are simply wonderful. Try one, you’ll like it. :)

A number of the European studies seem to conclude 720p is better than 1080i, which contrasts with many observations here from AVSers viewing 1080i vs 720p for years. -- JohnAnd yet 1080/50i seems the dominant format. Overall I think 1080i looks better, especially when looking at material with no or little compression. Video operators I've talked to prefer 1080i. In my past direct conversations with Joe Kane he was very (and that's putting it lightly) opposed to interlace. His 720p demos looked pretty good.

Were deinterlacing methods mentioned or was it viewed on a CRT?

The deinterlacer in my new HDTV makes the bob deinterlacer in my cable box and my four year old HDTV look awful. What a huge difference! It's difficult to find any interlacing artifacts on my new set although I can still find them in rare cases. For example the bottom line of a tiny China flag in the scoreboard during a volleyball game kept disappearing and reappearing.Which brings up an interesting question: How much more beneficial would 1080/60p (or 50p) be? I still think it will happen for production as 3Gbs HD SDI becomes more practical.

Which brings up an interesting question: How much more beneficial would 1080/60p (or 50p) be?
I guess I wouldn't see the bottom line of that little Chinese flag disappearing and reappearing.