1366 Display Vs 1920x1080 Display

[Anton-P.]

When you display a 1080P image on a 768P panel you LOSE DETAIL.

Independent of what Owen has said about bob deinterlacing in older plasma models, I submit that the DETAIL LOSS you cited is essentially immaterial.

In shaping 1080 down to 768 the processor doesn't just drop chunks out of the picture, but mathematically rescale the entire picture.

Any adverse effect is due only to rounding (to an integer number of pixels), for example a detail worth

1 pixel on 1080 should be 0.71 on 768 but is shown as 1 pixel,

5 pixels on 1080 should be 3.55 on 768 but is shown as 4 pixels,

10 pixels on 1080 should be 7.1 on 768 but is shown as 7 pixels,

30 pixels on 1080 should be 21.3 on 768 but is shown as 21 pixels, etc.

Note each error is no more than half a pixel (which is a very small 0.4mm on a 50" 768) and with larger "objects" the half-pixel error becomes negligible.

Does it really matter that a hair supposed to be 0.7 pixel thick is shown as 1 pixel thick on a 768, or a skin pore supposed to be 3.55 pixels wide is shown as 4 pixels wide? Not that we set out to watch a movie intently for these fine details anyway, so overall it is immaterial.

PS: I acknowledge that a small error if repeated a hundred times across the picture (like a row of thin lines or something) does become a big & annoying error, but that occurs only once in a while.

[PanaSung]

Yes, Yes and Yes

Actually, I was thinking of another LCD....that said, I still wouldn't buy a 27 for $2k, when I can get a 40 for $3k.

[PanaSung]

The Samsung's were not around when i bought my Dell 2 years ago. Anyway its an excellent PC monitor, not a TV, and thats how I now use it (although i started watching HD TV on it -- a cheap solution at the time).

I wasn't having a go at you btw....I just think that the 27 is stupidly priced.

[PanaSung]

PanaSung, was that a trolling comment? Or seriously intended?

In my opinion, not only was the comment factually incorrect, it was insulting because of its reference to 'right mind'.

* * * * * * * * * * * * * * * * * * * * * * * * * * * *

A number of people routinely watch 1920x1080 displays at an angle of view of view corresponding to a viewing distance of 2.2m or less from a 50" screen diagonal 16:9 display.

I just measured 2.2 from my 68cm CRT{which I usually watch from 4mtrs}.......and I can't believe anyone could watch a 50 from 2.2.

[PanaSung]

I recently played the PS3 and some Blu ray discs like MI3 and XMEN 3 on my panasonic and to be honest it hardly looked better than normal DVD.

In comparison I also have a Sony 40X 1080P panel and on this the Blu Rays looked brilliiant. Lots of detail and very nice, although only 40inch.

Yep....and the PQ of 1080 42in Sharp LCD was brilliant{bray disc}.

Btw, doesn't bray drop to 480p without the right equipement?....you're prolly watching a 480 upscaled to 720

[geese]

Yep....and the PQ of 1080 42in Sharp LCD was brilliant{bray disc}.

Btw, doesn't bray drop to 480p without the right equipement?....you're prolly watching a 480 upscaled to 720

No the PS3 will output at 1080P, 1080i and 480P and 480i for blu ray.

The 50 panas will accept 1080i (well my older unit anyway) and downconvert to 1366X768.

[Anthonyw]

So consider these things. If you want to buy a lot of blu ray of HDDVD discs get a 1080P panel. If you want to watch free to air HD TV and normal TV then get a 50 inch panel. I have my 50 and Im pretty happy with it and will keep it for a few more years until something way bigger and cheaper is available with full 1080P support. The 40X is used for gaming.

Agree with Owen - resoultion isnt everything. And I think this sums it up beautifully, unless your using a 1080p source (PS3, HDDVD or BR) most of the time then you will be far better off with a good 720P panel rather than a cheap 1080P panel that possibly shows less colour, poorer scaling and deinterlacing etc.

Just my 2c

[Roderick]

I wasn't having a go at you btw....I just think that the 27 is stupidly priced.

I tend to agree. The 24" Dell will do me for many years to come. The 27" is for intensive CAD and image work -- not for the run-of-the-mill PC user.

On the matter of 1080p, one factor that has not yet been mentioned is the possibility of much improved encoding for broadcast TV. For example, can MPEG4 handle 1080p 60fps within the current broadcast TV bandwith constraints?

It might be a long way away, but eventually better encoding standards will be introduced, possibly via satellite TV, as in the UK. 5 years, at least. You will need a new tuner, of course!

Rod

[MLXXX]

Again this is incorrect. 1080i 60 from an interlaced video camera has 60 motion updates per second and should always be deinterlaced to 1080p 60.

You don’t weave two fields to make a frame with true interlaced video as is done with 1080i containing 24fps progressive content.

Field data needs to be calculated from several odd and even fields to reconstruct a progressive frame that maintains high vertical resolution and 60fps motion.

If you deinterlace true interlaced 1080i video to 25 or 30 frames per second, motion smoothness is destroyed and is then no better then 24fps video or film.

One of the great advantages on true 1080i is that its motion portrayal is much better then 1080p 24. Deinerlace it poorly and that advantage it negated.

The same applies to 576i, deinterlace it to 25fps and try watching sport, the motion performance just plain sucks compared to proper 50 field per second display.

Thanks Owen for providing this more detailed explanation. You obviously know a lot about this subject. My explanation was oversimplified.

However the point I was trying to make, and which I stand by, is that the 'underlying frame rate' as I termed it, is still 25fps or 29.97fps, even if each frame is refreshed at double that rate, the refresh rate corresponding to the field rate. This is because only half of the lines are updated at each refresh.

I provide the following explanation for people unfamiliar with interlacing.

Assume a tennis ball is moving left to right across the field of view of the video camera and is captured by that camera at points A, B, C and D of its travel.

With a traditional analogue interlaced camera there is a time differential to the fields. Such a camera will generate the following information:

FRAME 1:

field 1, odd lines, tennis ball at position A

field 2, even lines, tennis ball at position B

FRAME 2:

field 1, odd lines, tennis ball at position C

field 2, even lines, tennis ball at position D

In Australia, the analogue video camera frames are at 25Hz, and the fields are are at 50Hz. We usually call that '50 frames per second interlaced', though it is sometimes described as '25 frames per second interlaced'. In particular, technical journals will describe such a picture as having 25 frames and 50 fields, per second.

There is a degree of justification for loosely calling it 50 frames per second because each field captures the tennis ball at a different position in its travel. On the other hand, each field captures only half of the lines.

To be very precise, the top of the tennis ball was scanned before the bottom of the tennis ball in the analogue video cameras of a few decades ago, so that even within a field there was a potential time differential, depending on the exposure duration per field and when the field was read in relation to when it was exposed.

Comb effect

There is a well known artefact of the interlaced format of a video camera and that is that when the fields are stitched together to create a frame for a digital display, alternate lines are of the moving source at different positions, i.e. different moments in time. For example, using the information shown above, a de-interlacer in a home digital TV could generate the following four frames:

Frame 1a: Odd lines, tennis ball at position A. [Generate dummy even lines for this frame as being the same as the odd lines, no actual even line information being available during the first field of transmission of an interlaced format scene.]

Frame 1b: Odd lines, tennis ball at position A (repeat of earlier field). Even lines, tennis ball at position B.

Frame 2a: Odd lines, tennis ball at position C. Even lines, tennis ball at position B (repeat of earlier field).

Frame 2b: Odd lines, tennis ball at position C (repeat of earlier field). Even lines, tennis ball at position D.

This system does give smoother motion than 25fps progressive (which will give the artefact of jerkiness), but can produce a pronounced combing artefact as between adjacent lines for an object moving across the screen.

The de-interlacing method above is 'weave' which is preferred for tranquil scenes. An alternative method 'bob' uses only every second line. This eliminates the 'comb' artefact but at the cost of losing half of the vertical resolution. A more sophisticated method attempts to restrict the 'bob' method to parts of the picture that are rapidly changing.

With 50fps progressive, we would get:

Frame 1a: Odd lines and even lines of the tennis ball at position A.

Frame 1b: Odd lines and even lines of the tennis ball at position B.

Frame 2a: Odd lines and even lines of the tennis ball at position C.

Frame 2b: Odd lines and even lines of the tennis ball at position D.

The labelling could be changed to "Frames 1, 2, 3 and 4" to emphasize the fact there would truly be 4 separate and independent frames. There is no need for approximations using motion adaptive de-interlacing. The detail is there, rock solid, fully updated for each frame. [That's before we get to compression artefacts from digital encoding but that's another story!]

Very little video content is available at true 50/60 fps progressive, other than computer game generated video.

I just measured 2.2 from my 68cm CRT{which I usually watch from 4mtrs}.......and I can't believe anyone could watch a 50 from 2.2.

Hi PanaSung. I agree that CRTs (particularly the ones that don't scan at 100Hz) can be hard to watch at close distances.

Many people find modern digital displays a different ball-game (particularly those with 1920x1080 architecture, and fed with good quality source material) and are happy sitting at much closer distances than for traditional TV viewing.

[Owen]

I get the difference between HD sourced from 24/25fps film, and 1080i50/60 video. And I think I get the fact that motion adaptive deinterlacing is the bees knees for handling 1080i50/60. Some questions though:

If a video processor is incapable of performing motion adaptive deinterlacing on 1080i50/60, would it be fair to say that weave deinterlacing would still offer some amount of benefit over and above bob deinterlacing? ie. Do 50/60 updates per second, but only update the odd or even lines in each update? Or does this lead to ridiculous combing/stepping/watever-you-want-to-call-it, that would mean even bob would be preferable?

Weave is the correct method to deinterlace progressive video contained in interlaced video, but it cannot be used for true interlaced source.

BOB is intended for true interlaced source as it maintains the frame rate, and works of for progressive source as well, but it lower vertical resolution substantially.

Pixel or motion adaptive systems are for true interlaced source only.

How does a video processor detect whether a 1080i signal is sourced from 24/25fps film (and therefore should do weave) or from 1080i50/60 video (and therefore should perform motion adaptive)?

The simplest method is to read flags in the video, but this is unreliable as sometimes the flags inaccurate.

Without flags the system must analyze several fields to work out if the video stream is realty progressive or true interlaced and if true interlaced in what direction and at what speed motion is occurring in various parts of the picture. An appropriate deinterlacing technique then needs to be applied to each part of the image.

This is a very complex operation requiring high power processing and smart algorithms, that’s why most displays don’t have very impressive deinterlacing, it simple costs too much.

I know you pretty much ignore FTA, but did you happen to catch any of the AFL tonight? AFAIK, the vast majority of HD on Australian TV is sourced from 24fps film, and this AFL broadcast was an example of the rarer native 1080i50 video. In fact, probably one of (if not the most) motion intensive 1080i50 video programs ever seen on Australian TV. Besides the very evident MPEG blocking, I think I could also see an additional fuzziness that I thought was perhaps related to the interlacing/deinterlacing. Perhaps it was just more MPEG blocking, but I think it looked more like combing/stepping. You mentioned on the SXRD thread way back in November that the SXRD "seemed" to be capable of both weave and motion adaptive. Did you ever confirm this? Is there any chance that our magical SXRDs are performing weave deinterlace on 1080i50 video?

No I don’t follow sport at all, so I did not see the footy.

There are three main factors affecting detail in fast motion

1. Video compression

2. Deinterlacing

3. Shutter speed limitations in the camera due to low light.

Shutter speed limitations can definitely be an issue for night sports events under lights.

I have seen some very server motion blur due to shutter speed and posted about it on this forum a long while back.

The bandwidth used for local 1080i is woefully inadequate to handle anything more then slow pans, so fast motion will be a total mess, and is.

With the above two limitations, deinterlacing is probably the least of your worries.

The deinterlacing on the SXRD is good, but definitely not great, and if you want something better, a high end PC or a good stand alone video processor will be required.

I have only ever looked at the deinterlacing performance on free to air HD before I purchased my SXRD.

Since owning it I have never used the STB or any input other then 1080p form my HTPC, so my experience on the SXRD’s deinterlacing is limited.

[Owen]

Thanks Owen for providing this more detailed explanation. You obviously know a lot about this subject. My explanation was oversimplified.

However the point I was trying to make, and which I stand by, is that the 'underlying frame rate' as I termed it, is still 25fps or 29.97fps, even if each frame is refreshed at double that rate, the refresh rate corresponding to the field rate. This is because only half of the lines are updated at each refresh.

I provide the following explanation for people unfamiliar with interlacing.

Assume a tennis ball is moving left to right across the field of view of the video camera and is captured by that camera at points A, B, C and D of its travel.

With a traditional analogue interlaced camera there is a time differential to the fields. Such a camera will generate the following information:

FRAME 1:

field 1, odd lines, tennis ball at position A

field 2, even lines, tennis ball at position B

FRAME 2:

field 1, odd lines, tennis ball at position C

field 2, even lines, tennis ball at position D

In Australia, the analogue video camera frames are at 25Hz, and the fields are are at 50Hz. We usually call that '50 frames per second interlaced', though it is sometimes described as '25 frames per second interlaced'. In particular, technical journals will describe such a picture as having 25 frames and 50 fields, per second.

There is a degree of justification for loosely calling it 50 frames per second because each field captures the tennis ball at a different position in its travel. On the other hand, each field captures only half of the lines.

To be very precise, the top of the tennis ball was scanned before the bottom of the tennis ball in the analogue video cameras of a few decades ago, so that even within a field there was a time differential. This assisted apparent fluidity of motion when the picture was dispayed on a CRT screen scanning in a synchronised maner. (On the other hand, modern digital displays do not lend themselves to line by line refreshing.)

Comb effect

There is a well known artefact of the interlaced format of a video camera and that is that when the fields are stitched together to create a frame for a digital display, alternate lines are of the moving source at different positions, i.e. different moments in time. For example, using the information shown above, a de-interlacer in a home digital TV could generate the following four frames:

Frame 1a: Odd lines, tennis ball at position A. [Generate dummy even lines for this frame as being the same as the odd lines, no actual even line information being available during the first field of transmission of an interlaced format scene.]

Frame 1b: Odd lines, tennis ball at position A (repeat of earlier field). Even lines, tennis ball at position B.

Frame 2a: Odd lines, tennis ball at position C. Even lines, tennis ball at position B (repeat of earlier field).

Frame 2b: Odd lines, tennis ball at position C (repeat of earlier field). Even lines, tennis ball at position D.

This system does give smoother motion than 25fps progressive (which will give the artefact of jerkiness), but can produce a pronounced combing artefact as between adjacent lines for an object moving across the screen.

With 50fps progressive, we would get:

Frame 1a: Odd lines and even lines of the tennis ball at position A.

Frame 1b: Odd lines and even lines of the tennis ball at position B.

Frame 2a: Odd lines and even lines of the tennis ball at position C.

Frame 2b: Odd lines and even lines of the tennis ball at position D.

Interlaced video camera’s, be they analogue or digital do not capture frames, they capture fields 50 times per second for Oz. These fields where never intended to be converted to frames at 25fps.

Each field should be treated as a frame, and that is what BOB deinterlacing does.

BOB scales fields to full frames and displays them at field rate (50 times per second).

With a clever deinterlacer, we can use information from adjoining fields to fill in the missing lines to get back vertical resolution. During still or low motion scenes this is easy to do as there is little or no displacement of information between fields, but with faster motion we need to use motion adaption or compensation to remove combing artifacts.

Interlaced video is a compromise, it has the same resolution as progressive video for still or low motion scenes, but trades vertical resolution for motion smoothness with faster motion.

In my view this is a good compromise as resolution in fast motion is already heavily compromised by video compression and in a lot of cases shutter speed related blur as well.

The problem is that digital displays just have not had, and in the large part still don’t have good deinterlacing abilities especially for 1080i. That’s not a limitation of interlaced video it’s a limitation of digital displays ability to deal with interlacing effectively.

CRT TV’s work great with interlaced video as they where designed to do, and maintain clean images with smooth motion.

We have all been watching 576i on CRT TV’s all our lives and sporting events have always looked great. In fact I recon things used to look better in the days before digital TV camera’s and production systems, as transmissions where not hampered by data rate limits and we never got the dreadful pixilation and artifacts that afflict digital TV.

[MLXXX]

Interlaced video camera’s, be they analogue or digital do not capture frames, they capture fields 50 times per second for Oz. These fields where never intended to be converted to frames at 25fps. ...

In an indirect sense they were intended to blend together, displaced vertically by one line position. The two fields gave an effective weave to the extent the CRT phosphors and the human eye led to persistence of vision sufficient for the immediate prior field to still be visible as the current field was being drawn on the traditional CRT display screen.

It was a remarkable system as although it allowed fuller detail to be present by way of persistence of vision, the current update to the screen was always brighter than the fading preceding update, encouraging the eye to follow the updated parts of a moving object, rather than dwelling on the comb effect [comparing the new lines with the fading lines in between the new lines].

...

CRT TV’s work great with interlaced video as they where designed to do, and maintain clean images with smooth motion.

We have all been watching 576i on CRT TV’s all our lives and sporting events have always looked great. In fact I recon things used to look better in the days before digital TV camera’s and production systems, as transmissions where not hampered by data rate limits and we never got the dreadful pixilation and artifacts that afflict digital TV.

An increase in video data rates would be very welcome.

Just last night I was watching a free to air nature program and when the camera zoomed out to show a vast number of small fish swimming in turgid water, the picture .... failed. All that could be seen was a mass of small colourless blocks. There were no recognisable fish. Analogue TV would have enabled the fish to be seen despite the high measure of chaos [entropy] of the scene. Now here's an impractical idea - a digital TV that monitors the analogue simulcast and extracts useful info to supplement the digital picture!

It's interesting too the effect of compression on audio. So much of the music available on the internet is in mp3 format (or in a compressed streaming format), but when compared to the same uncompressed material, there is often a detectable loss in quality (particularly with orchestral music).

[Owen]

Unfortunately the “analogue” broadcast is not truly analogue, it’s a digital to analogue conversion of the digital broadcast with most of the compression artifacts inherent in the digital version, but with analogue issues added.

The old full analogue camera to TV system was much better if you lived in a good reception area.

[PanaSung]

Hi PanaSung. I agree that CRTs (particularly the ones that don't scan at 100Hz) can be hard to watch at close distances.

Many people find modern digital displays a different ball-game (particularly those with 1920x1080 architecture, and fed with good quality source material) and are happy sitting at much closer distances than for traditional TV viewing.

If the pixel grid is tight, then it's all good,....I had to get very close to the 1080 Sharp 42 LCD before it became a problem, but a 50in screen is HUGE, and at 2.2, I'd feel dominated by FTA madness.

I got a shock at how big a 40in LCD looked in a lounge room as compared to the showrooms.....but anyway, as long as one has the room, then you're better off with as big a TV as you can afford.

[PanaSung]

Yes, Yes and Yes

Show me a pic of the HDMI port on any 27 in DELL

[merovingian]

Weave is the correct method to deinterlace progressive video contained in interlaced video, but it cannot be used for true interlaced source.

BOB is intended for true interlaced source as it maintains the frame rate, and works of for progressive source as well, but it lower vertical resolution substantially.

Now I'm really curious with what's going on with the SXRD. Until such time as get my HTPC (hopefully in a few weeks), I am making do with the HD-STB that Sony packaged with the SXRD. The static scenes last night were truly something to behold, so I would be suprised if the SXRD was performing bob deinterlace. However, given that Sony have not gone out of their way to highlight motion adaptive delinterlacing abilities, I doubt that it was doing that either. Hence my guess that it was just always weaving. It seems that guess was wrong, so I'm none the wiser.

The bandwidth used for local 1080i is woefully inadequate to handle anything more then slow pans, so fast motion will be a total mess, and is.

The fast pans were indeed woeful. Especially the close-ups of players on the run, with the very out-of-focus crowd in the background.

With the above two limitations, deinterlacing is probably the least of your worries.

The deinterlacing on the SXRD is good, but definitely not great, and if you want something better, a high end PC or a good stand alone video processor will be required.

I have only ever looked at the deinterlacing performance on free to air HD before I purchased my SXRD.

Since owning it I have never used the STB or any input other then 1080p form my HTPC, so my experience on the SXRD’s deinterlacing is limited.

Fair enough, so it remains a mystery for now. Part of my reason for wanting to get to the bottom of it is that I am trying to settle on my HTPC video card. Your strong endorsement of the 8800 is something to consider, particularly if the SXRD isn't a great deinterlacer in its own right.

Anyway, sports fan or not, surely a HD/video enthusiast would be interested in taking a quick gander at the HD AFL just from the technical point of view? It's all happening again next Saturday night on Ten.

[merovingian]

On the matter of 1080p, one factor that has not yet been mentioned is the possibility of much improved encoding for broadcast TV. For example, can MPEG4 handle 1080p 60fps within the current broadcast TV bandwith constraints?

After watching HD AFL last night, it seems we don't have enough bandwidth to do a great job with 1080i50 carried on MPEG2. Maybe MPEG4 solves 1080i50, but I suspect that 1080p60 would still be too demanding without more bandwidth. Just my hunch.

[Roderick]

After watching HD AFL last night, it seems we don't have enough bandwidth to do a great job with 1080i50 carried on MPEG2. Maybe MPEG4 solves 1080i50, but I suspect that 1080p60 would still be too demanding without more bandwidth. Just my hunch.

I seem to remember reading somewhere that MPEG4 can handle the same information in about half the bandwith of MPEG2, so the existing channels may have enough bandwidth. Trouble is, everyone is now locked into MPEG2, so a change to a better encoding standard would require a major upheval. Unlikely to occurr for a long time to come, but a new satellite service could do it easily with new decoders.

With increasing computing power, better compression algorithms are likely to permit even more information to squeezed into the existing channels. So 1080p at 60fps may not be that that far off.

My previous point about interlaced signals being "historical baggage" (not "garbage", by the way, Owen!) is that there are likely to be more efficient ways of squeezing better moving images into less bandwidth. With digital signal processing there is no pressing reason to choose every second row -- why not every second column, or even a random 50% of all pixels. There may even be some holographic method of ringing more resolution out of a given bandwith. No doubt some of the best mathematical brains have been hard at work for a long time now, and that future encoding methods will be more efficient even than MPEG4. Will need considerable computing power, no doubt.

So, 1080p screens will be needed eventually, but I'm not in any hurry, personally. My 1366 x 768 screen will do just fine for the time being, and I recommend that resolution to all economy-minded buyers at present. .

Rod

[drsmith]

And I recommend 1366 x 768 screens to all who sit too far away from the TV to get any meaningful visual benefit from a 1920x1080 screen.

[MLXXX]

Here is a review published in 2000 that explains interlaced vs progressive very well, and includes animations.

... With increasing computing power, better compression algorithms are likely to permit even more information to squeezed into the existing channels. So 1080p at 60fps may not be that that far off. ...

I tend to agree with a new source such as Blu-ray, which is just starting out. However as you point out, Rod, the existing MPEG2 format for Australian DTV may be an impediment for FTA in Australia changing to true 1080 50p. I'm not sure a new generation of set-top boxes would go down well, or the need to simulcast the old '50i' along with the new 50p.

________________________________________________________

As a broad statement, algorithms for encoding true 50p although they may entail high peak bit-rates for highly chaotic individual frames, can be expected to entail less than double the average encoded bit-rate, compared with encoding of interlaced frames at the same field rate, i.e. true 25i (also known as 50i); all other things being equal. This is because:

• many parts of a picture have a similar intensity and colour, particularly parts of a picture that only differ in location by a few pixels left and right or up and down. Hence the additional detail available by updating all lines in every frame, compared with updating alternate lines as is done with an interlaced method, may to a large extent consist of duplicate or near duplicate information.
  
• it is computationally less complex to encode a picture that is in a progressive format. For example, the colour information (which is at a coarser resolution than the intensity information) is easier to manipulate.
  

[Owen]

After watching HD AFL last night, it seems we don't have enough bandwidth to do a great job with 1080i50 carried on MPEG2. Maybe MPEG4 solves 1080i50, but I suspect that 1080p60 would still be too demanding without more bandwidth. Just my hunch.

At the 12Mbps that is allocated to HD in Oz, there is no chance in hell that Mpeg4 will cope with 1080p 50 with good quality results.

Even using the full 18Mbps (approx) channel allocation Mpg4 would struggle at 50 frames per second, appart from the fact that 1080p 50 source does not exist anywhere.

BluRay and HDDVD run at around 28Mbps peak for 1080p 24 frames per second, so we will need over double that for 1080p 50

Satellite systems don’t have anywhere near 18Mbps to through at a single channel so 1080p 50 is just not viable.

[Roderick]

At the 12Mbps that is allocated to HD in Oz, there is no chance in hell that Mpeg4 will cope with 1080p 50 with good quality results.

Of course it can! According to my sources, the H264 part of the MPEG-4 standard can deliver similar video quality with a third of the number of bits used by the MPEG-2 standard.

That means that the current HD channels easily have sufficient bandwith to transmit 1080p at 50 frames a second using H264. At worst, that is only twice the bitrate required for 1080i. Remember that 1080p can be delivered somewhat more efficiently than 1080i, as per MLXXX's post.

Rod

[MikeAusDTV]

With digital signal processing there is no pressing reason to choose every second row -- why not every second column, or even a random 50% of all pixels.

That's the way MPEG compression works - it looks at the way pixels change from frame to frame and and only send information about the parts of the picture that are changing.

Remember that if you send broadcast quality video with NO compression you need 140MBPS !

The reason Interlace is still supported in existing digital TV standards is because a lot of CRT TVs were still in use when the current standards were formulated. Interlace was a very easy way to halve the bandwidth to the display - because CRT displays can handle Interlace directly.

With LCD/Plasma/LCoS now dominating, interlacing is no longer the most effective way to reduce bandwidth in a DIGITAL transmission/storage environment. You may as well do ALL the compression in MPEG, because the display just needs to reprocess the Interlaced signal that has come out of the MPEG decoder, before it can send a de-interlaced signal to the screen.

[Owen]

Of course it can! According to my sources, the H264 part of the MPEG-4 standard can deliver similar video quality with a third of the number of bits used by the MPEG-2 standard.

That means that the current HD channels easily have sufficient bandwith to transmit 1080p at 50 frames a second using H264. At worst, that is only twice the bitrate required for 1080i. Remember that 1080p can be delivered somewhat more efficiently than 1080i, as per MLXXX's post.

Rod

What you read was inaccurate.

h.254 is more efficient them Mpeg2, but if you attempt to push it too far it looks very bad.

I have the same 1080i content broadcast in both 12-14mbps Mpeg2 and 6mbps h.264 and can tell you that the Mpeg2 version is MUCH better. Any movement and the low bit rate h.264 version disintegrates into an unwatchable mess.

1080i at 12mbps h.264 is about the same as 18mps Mpg2, which is very good but not perfect.

For really good quality we need HDDVD or BluRay data rates of 28mbps peak or even greater. BluRay allows rates up to 40mbps.

The suggestion that 1080p 50 is viable at Oz 12-14mbps data rates is completely ridiculous. That sort of data rate gives good performance with 1080p 24 or 1080i 50 but that’s all.

There is also the issue that 1080p 50 video does not exist. Nothing is shot in that format.

[Owen]

Remember that if you send broadcast quality video with NO compression you need 140MBPS !

You left off a 0 mate, uncompressed 1080p 24 or 1080i 25 is closer to 1,400mbps (1.4Terabits)

The reason Interlace is still supported in existing digital TV standards is because a lot of CRT TVs were still in use when the current standards were formulated. Interlace was a very easy way to halve the bandwidth to the display - because CRT displays can handle Interlace directly.

With LCD/Plasma/LCoS now dominating, interlacing is no longer the most effective way to reduce bandwidth in a DIGITAL transmission/storage environment. You may as well do ALL the compression in MPEG, because the display just needs to reprocess the Interlaced signal that has come out of the MPEG decoder, before it can send a de-interlaced signal to the screen.

There is a lot more to it then that.

1080p 24/25 cannot replace 1080i 50, because the frame rate is too low for sport, news or other fast moving content often broadcast on TV.

1080p 24/25 is only suitable for film or 1080p 24 video camera source.

1080p 50 is not practical due to data rate limits so 1080i 50 is therefore a very good compromise between 1080p 24 and 1080p 50. It can carry 1080p 24/25, as well as true interlaced 1080i 50 which has 50fps motion in a single container, making it a very flexible and effective format.