Whilst we're on about HD picture quality, at the moment BBC HD and Sky Sports HD 1 are showing the same tennis match. This might be a good opportunity to do some comparison tests.
1920 vs 1440 the test
3 of 6
Originally Posted by jamiec21:
“Whilst we're on about HD picture quality, at the moment BBC HD and Sky Sports HD 1 are showing the same tennis match. This might be a good opportunity to do some comparison tests.”
“Whilst we're on about HD picture quality, at the moment BBC HD and Sky Sports HD 1 are showing the same tennis match. This might be a good opportunity to do some comparison tests.”
A little late look at posts 37/38 in this thread, but thanks for the reminder.
Colleagues
I thought I would have another try given the kind words that were written about my earlier comments. In case it helps then this is just my opinion of what I can see here to the best of my limited knowledge. I hope that some of this makes sense.
I looked at the second picture pair posted by jzee. I must say a big thanks to you for going to all this work to collect the images.
There are a lot of differences between these two images. It is a complex equation. The Granada image is full 1920 pixels. One give-away is the ITV logo because we can assume that this is added at 1920 pixels. The vertical edges of the logo are sharp except for the compression artefacts. The London version is filtered. The edges are soft. You can also see the darkening around the vertical edges – an effect of the filters. If you look closely you can in addition see the darkening above and below the horizontal edges. This shows the logo was filtered vertically too. More of that later.
The Granada image is interlaced. Look at the man’s head which was clearly moving; look toward the top and on the right side. You can see that alternate lines show the horizontal detail (like edges) in two different places. By contrast look at the London picture in the same area. This has lost the signs of interlace. The highlights on the man’s glasses are worth a look. These look frankly awful on the Granada version and a lot more real on the London one. What is happening here then? I think we are seeing two things. One is that the compression is failing on the Granada version. There is a lot of detail here which is just false. The black/white striation is just not real. The other is the interlace. It is really hard to see but around this area the image looks a little sharper on the London picture in the vertical direction. So; the London picture is sharper in the vertical direction) then? No. This is a good illustration of the problems of looking at stills of interlaced video. In places there are two very slightly sharper images from the two fields superimposed (because this is a still) and looking, as a compound image, more blurred than the two fields. Also then looking more blurred (falsely) than the deinterlaced London version. Actually then the Granada image if it were properly viewed as two parts of an interlaced sequence would appear to be a bit better than London image. This also shows some of the weakness of deinterlacers. The deinterlacer has produced a convincing still here but it has lost information. It looks better than the combined two fields frozen in time but not better than the two fields actually look when shown sequentially. Sorry that is probably a very bad explanation.
The London image has been deinterlaced. It has then been filtered in both the horizontal and vertical axis. The horizontal filtering has lost detail in the horizontal direction. The deinterlacing has lost detail in the vertical direction. This has reduced the vertical resolution by something over half in the moving parts of the picture and a fair amount in the still parts. The compression on the London image is much better than the Granada one though. While this is good because some parts of the picture are clearly more real in that image the Granada one the two issues (lack of resolution versus compression artefacts) do not trade in this simple way. However there is another issue with this. The compression artefacts when present in small amounts like this tend to create artificial detail and sharpness. So they aggravate the London image’s lack of detail. Look at the man’s stubble and you see some detail which is actually compression-generated artefact but the result is in-line with what you would expect to see random detail in this area.
Why process the London signal in this way then? You cannot just apply a horizontal filter on an interlaced signal because it creates weird aliasing effects between the lines so this is necessary. So to get it to 1440 by 1080 pixels it had to be deinterlaced.
What is the net effect? The deinterlacing lost a significant vertical resolution – as mentioned in some areas of the picture this amounts to over half. The result depends on movement then but will reach below 1440 x 540 pixels – a lot less than you would like, and to put it another way, this amounts to about a third of the pixel count you were hoping for. Even on a completely still image you will never get the original 1080 lines worth of detail because as I said earlier a vertical filter is being used. We are looking at a picture with a maximum total resolution of somewhere between 50% and 35% of the original maximum resolution depending on the level of movement. So instead of being fives times the resolution of SD it is closer to double the resolution. Note that if production had been in 1080p then the 1440 pixel signal would improve significantly in these stills.
One contributor (on this subject but not necessarily in this thread) mentioned that the 1440 images look less colourful. I have noticed often that this is a perceived effect of softer images. Much of the time untrained viewers notice this effect first long before they notice lack of resolution for example. I think it is an important consequence. A lot of artefacts, that are not in any significant way changing the colour mass, show up as less colourful images. My best guess at why this is relates to the edge-detecting nature of the brains perception system. As it cannot see colour mass it uses the changes in colour to indicate what colour fills the mass. If the edges are softer it is tricked into seeing a less bold colour. The edges can either be less sharp (say filtered) or affected by ringing or noise some other affect that blurs them. So, yes, you loose apparent colour as well as detail.
So, I hope that long ramble is of some use. I think that because some people have more faith in deinterlacing than me they will take issue with parts of what I am saying here. We may have to agree to disagree for now. In time I may be able to show you examples of why I think interlacing is a lot less successful at retaining vertical resolution than the people who sell them would suggest.
I thought I would have another try given the kind words that were written about my earlier comments. In case it helps then this is just my opinion of what I can see here to the best of my limited knowledge. I hope that some of this makes sense.
I looked at the second picture pair posted by jzee. I must say a big thanks to you for going to all this work to collect the images.
There are a lot of differences between these two images. It is a complex equation. The Granada image is full 1920 pixels. One give-away is the ITV logo because we can assume that this is added at 1920 pixels. The vertical edges of the logo are sharp except for the compression artefacts. The London version is filtered. The edges are soft. You can also see the darkening around the vertical edges – an effect of the filters. If you look closely you can in addition see the darkening above and below the horizontal edges. This shows the logo was filtered vertically too. More of that later.
The Granada image is interlaced. Look at the man’s head which was clearly moving; look toward the top and on the right side. You can see that alternate lines show the horizontal detail (like edges) in two different places. By contrast look at the London picture in the same area. This has lost the signs of interlace. The highlights on the man’s glasses are worth a look. These look frankly awful on the Granada version and a lot more real on the London one. What is happening here then? I think we are seeing two things. One is that the compression is failing on the Granada version. There is a lot of detail here which is just false. The black/white striation is just not real. The other is the interlace. It is really hard to see but around this area the image looks a little sharper on the London picture in the vertical direction. So; the London picture is sharper in the vertical direction) then? No. This is a good illustration of the problems of looking at stills of interlaced video. In places there are two very slightly sharper images from the two fields superimposed (because this is a still) and looking, as a compound image, more blurred than the two fields. Also then looking more blurred (falsely) than the deinterlaced London version. Actually then the Granada image if it were properly viewed as two parts of an interlaced sequence would appear to be a bit better than London image. This also shows some of the weakness of deinterlacers. The deinterlacer has produced a convincing still here but it has lost information. It looks better than the combined two fields frozen in time but not better than the two fields actually look when shown sequentially. Sorry that is probably a very bad explanation.
The London image has been deinterlaced. It has then been filtered in both the horizontal and vertical axis. The horizontal filtering has lost detail in the horizontal direction. The deinterlacing has lost detail in the vertical direction. This has reduced the vertical resolution by something over half in the moving parts of the picture and a fair amount in the still parts. The compression on the London image is much better than the Granada one though. While this is good because some parts of the picture are clearly more real in that image the Granada one the two issues (lack of resolution versus compression artefacts) do not trade in this simple way. However there is another issue with this. The compression artefacts when present in small amounts like this tend to create artificial detail and sharpness. So they aggravate the London image’s lack of detail. Look at the man’s stubble and you see some detail which is actually compression-generated artefact but the result is in-line with what you would expect to see random detail in this area.
Why process the London signal in this way then? You cannot just apply a horizontal filter on an interlaced signal because it creates weird aliasing effects between the lines so this is necessary. So to get it to 1440 by 1080 pixels it had to be deinterlaced.
What is the net effect? The deinterlacing lost a significant vertical resolution – as mentioned in some areas of the picture this amounts to over half. The result depends on movement then but will reach below 1440 x 540 pixels – a lot less than you would like, and to put it another way, this amounts to about a third of the pixel count you were hoping for. Even on a completely still image you will never get the original 1080 lines worth of detail because as I said earlier a vertical filter is being used. We are looking at a picture with a maximum total resolution of somewhere between 50% and 35% of the original maximum resolution depending on the level of movement. So instead of being fives times the resolution of SD it is closer to double the resolution. Note that if production had been in 1080p then the 1440 pixel signal would improve significantly in these stills.
One contributor (on this subject but not necessarily in this thread) mentioned that the 1440 images look less colourful. I have noticed often that this is a perceived effect of softer images. Much of the time untrained viewers notice this effect first long before they notice lack of resolution for example. I think it is an important consequence. A lot of artefacts, that are not in any significant way changing the colour mass, show up as less colourful images. My best guess at why this is relates to the edge-detecting nature of the brains perception system. As it cannot see colour mass it uses the changes in colour to indicate what colour fills the mass. If the edges are softer it is tricked into seeing a less bold colour. The edges can either be less sharp (say filtered) or affected by ringing or noise some other affect that blurs them. So, yes, you loose apparent colour as well as detail.
So, I hope that long ramble is of some use. I think that because some people have more faith in deinterlacing than me they will take issue with parts of what I am saying here. We may have to agree to disagree for now. In time I may be able to show you examples of why I think interlacing is a lot less successful at retaining vertical resolution than the people who sell them would suggest.
What an entertaining read this thread is, but where's the punchline? I assume this really is all just a wind-up by jzee 
Images of, admittedly, the same content, from two different decoders in two different formats - what sort of comparison is that? Even if comparing static images could ever be considered a worthwhile test this particular comparison is worthless.
InMyHo - unfortunately you seem to have fallen into jzee's trap and provided a very interesting and detailed analysis of the differences between the two decoders rather than the two transmissions.
Images of, admittedly, the same content, from two different decoders in two different formats - what sort of comparison is that? Even if comparing static images could ever be considered a worthwhile test this particular comparison is worthless.
InMyHo - unfortunately you seem to have fallen into jzee's trap and provided a very interesting and detailed analysis of the differences between the two decoders rather than the two transmissions.
Originally Posted by InMyHO:
“Why process the London signal in this way then? You cannot just apply a horizontal filter on an interlaced signal because it creates weird aliasing effects between the lines so this is necessary. So to get it to 1440 by 1080 pixels it had to be deinterlaced.”
“Why process the London signal in this way then? You cannot just apply a horizontal filter on an interlaced signal because it creates weird aliasing effects between the lines so this is necessary. So to get it to 1440 by 1080 pixels it had to be deinterlaced.”
I think you're mistaken there. It's very common to perform image manipulation in one dimension without worrying about the other dimension at all. e.g. for horizontal scaling the resize algorithm works on individual rows of pixels without worrying about adjacent rows.
This is highly advantageous when dealing with an interlaced signal since it saves a lot of work - specifically you don't have to deinterlace. That's why DVB has so many horizontally squashed resolutions (i.e. the standard 720 and 704, plus the squashed 544, 528, 480 and 352 × 576 pixels), but no vertically squashed resolutions. Some US Pay-TV broadcasters apparently use lower vertical resolutions, which looks predictably horrendous.
Anyway, it doesn't matter - as Badvok said, this is probably down to jzee's screen capturing methods, and nothing to do with the broadcasts.
Cheers,
David.
Originally Posted by Nigel Goodwin:
“It doesn't matter if you consider it an 'impression' or not, the old BBC HD transmissions were a better picture and contained more detail than all others.
Presumably you are a newcomer to HD?, and never saw them - the 'wow factor' was huge, and there's none at all now.”
“It doesn't matter if you consider it an 'impression' or not, the old BBC HD transmissions were a better picture and contained more detail than all others.
Presumably you are a newcomer to HD?, and never saw them - the 'wow factor' was huge, and there's none at all now.”
That is because everyone has become habituated to HDTV now. When you had never seen it before there was indeed a huge 'wow factor'. Now that people have been watching it for over 4 years it doesn't have the same impact.
Your memories of 3-4 year old transmissions are no evidence that the picture was any better then. Only comparing recordings of the same material is valid.
2Bdecided
Good point well made that you can get away with just horizontal filtering if you change nothing in the vertical axis at all. That was a mistake - it was late at night. Apologies.
Badvok
It could be the receiving equipment that is deinterlacing here. It is certainly deinterlaced after the logos are added. If the box deinterlaces just to create nice still frames that is quite a sophisticated feature for a little-used function; but possible.
One other thing puzzles me a little though Badvok. The compression artefacts are so much more obvious on the Granada feed. The highlights on the glasses and the stubble on the man's face for example. This is consistent with the deinterlacing happening before the transmission on the London feed. I guess that effect could be from just using a different encoder or different bit rate but it looks to me more likely that the material was softer before it was encoded.
I am sure jzee is not trying to trap anyone. That would be strange behaviour indeed.
Good point well made that you can get away with just horizontal filtering if you change nothing in the vertical axis at all. That was a mistake - it was late at night. Apologies.
Badvok
It could be the receiving equipment that is deinterlacing here. It is certainly deinterlaced after the logos are added. If the box deinterlaces just to create nice still frames that is quite a sophisticated feature for a little-used function; but possible.
One other thing puzzles me a little though Badvok. The compression artefacts are so much more obvious on the Granada feed. The highlights on the glasses and the stubble on the man's face for example. This is consistent with the deinterlacing happening before the transmission on the London feed. I guess that effect could be from just using a different encoder or different bit rate but it looks to me more likely that the material was softer before it was encoded.
I am sure jzee is not trying to trap anyone. That would be strange behaviour indeed.
Originally Posted by nigelbb:
“That is because everyone has become habituated to HDTV now. When you had never seen it before there was indeed a huge 'wow factor'. Now that people have been watching it for over 4 years it doesn't have the same impact.
”
“That is because everyone has become habituated to HDTV now. When you had never seen it before there was indeed a huge 'wow factor'. Now that people have been watching it for over 4 years it doesn't have the same impact.
”
Yet other channels do - funny how you 'get used to' only one channel?.
Quote:
“
Your memories of 3-4 year old transmissions are no evidence that the picture was any better then. Only comparing recordings of the same material is valid.”
“
Your memories of 3-4 year old transmissions are no evidence that the picture was any better then. Only comparing recordings of the same material is valid.”
No memories involved - simple direct comparison of recordings from then, and the exact same images following the crippling of bit-rates, side by side. As explained endlessly, perhaps you should try reading the threads?.
The old ones are still 'wow'.
Originally Posted by nigelbb:
“That is because everyone has become habituated to HDTV now. When you had never seen it before there was indeed a huge 'wow factor'. Now that people have been watching it for over 4 years it doesn't have the same impact.
”
“That is because everyone has become habituated to HDTV now. When you had never seen it before there was indeed a huge 'wow factor'. Now that people have been watching it for over 4 years it doesn't have the same impact.
”
I disagree the recent Lost Land of the Tiger Episode 1 blew me away with what I felt was a return to true HD when compared to all of the other HD I'd seen for such a long time.
Unfortunately it was a one off episode at that quality.
However, the fact that the increase in sharpness and detail was so obvious to me tells me straight away that HD can still have a wow factor.
Now maybe if all BBC HD was that perfect all of the time then we'd get used to it and you'd have a valid point Nigel. Unfortunately it isn't which is why when something good comes along it truly stands out.
Originally Posted by InMyHO:
“This is consistent with the deinterlacing happening before the transmission on the London feed.”
“This is consistent with the deinterlacing happening before the transmission on the London feed.”
I think that would be very strange but I guess anything is possible. Both the transmissions are 1080i, i.e. interlaced, so if there was some de-interlacing before transmission the picture would have to be re-interlaced/telecined for transmission. But maybe that is how these encoders work to reduce the bit rate required.
I'd say that rather than obsessing over the (IMHO) marginal differences in particular HD streams we should all be concerned with the quality of the actual content instead.
For me, once the initial excitement of receiving HD transmissions had worn off i stopped tuning into stuff just because it was HD and returned to watching stuff that was well-made, whether it be HD, SD or black and white and grainy (like the rerun of the excellent World at War on BBC2 recently).
For me, once the initial excitement of receiving HD transmissions had worn off i stopped tuning into stuff just because it was HD and returned to watching stuff that was well-made, whether it be HD, SD or black and white and grainy (like the rerun of the excellent World at War on BBC2 recently).
Thankyou 2Bdecided and InMyHO for your detailed posts. First @ 2Bdecided in reference to this post, I'm glad you agree the difference between ITV1 HD Granada and ITV1 HD London is clear. In reference to your observation that if you rescale the 1920 ITV1 HD Granada to 1440, and then rescale back to 1920, more detail remains that is seen in the ITV1 HD London samples, this does indeed appear to be true, as I have tested. In fact, I have observed if you scale to 1280x720 and rescale back to 1920x1080, the image still has more detail than the ITV1 HD London image! What I have observed is that if you scale the ITV1 HD Granada 1920 image to 1440, and then back 1920, you either end up with rather unsightly jaggies on edges, or softening depending on what photo package you use to do the rescaling. Using Paint.NET, the image from 1920 ITV1 HD Granada>1440 jpeg (reflecting the conversion to lossy H.264), and then back to 1920 png, the end result (png) is not at all far away from the ITV1 HD London results. So I think the problem with 1440 is this, when you scale it down and then back again you either get jaggies or you have to apply some kind of blur filter which then inevitably loses detail.
InMyHO, in reference to your posts above, the first point is I believe talk of deinterlacing artifacts is erroneous, as I believe Monte Carlo and Bust was based on a progressive source, if deinterlacing is completely disabled no interlace lines are visible, which is not the case, for example with I'm a Celebrity GMOOH which appears to be filmed with interlaced cameras. You say the reflection on the man's glasses in Image 2 looks terrible- I think this is just very strong glare that appears as strong glare appears to our own eyes, if you look at the sun on the two men's jackets you can see the sun is indeed very strong. The fact the glare is dampened on the ITV1 HD London image is merely another example of brightness on reflective objects being dulled by the loss of resolution that I have previously pointed out as a problem with 1440 caused by the conversion to lossy 1440 and possible added blur in the process.
Regarding compression artifacts on ITV1 HD Granada, it is inevitable it will have more as it runs at the same bitrate as ITV1 HD London but has 518,000 more pixels to provide information for- I believe the recommended EBU fixed rate for 1920 is at least 14Mbps, so ITV are running substantially under that at 10.2 so you will get more artifacts on complicated scenes and fast moving multiple objects or water.
InMyHO, in reference to your posts above, the first point is I believe talk of deinterlacing artifacts is erroneous, as I believe Monte Carlo and Bust was based on a progressive source, if deinterlacing is completely disabled no interlace lines are visible, which is not the case, for example with I'm a Celebrity GMOOH which appears to be filmed with interlaced cameras. You say the reflection on the man's glasses in Image 2 looks terrible- I think this is just very strong glare that appears as strong glare appears to our own eyes, if you look at the sun on the two men's jackets you can see the sun is indeed very strong. The fact the glare is dampened on the ITV1 HD London image is merely another example of brightness on reflective objects being dulled by the loss of resolution that I have previously pointed out as a problem with 1440 caused by the conversion to lossy 1440 and possible added blur in the process.
Regarding compression artifacts on ITV1 HD Granada, it is inevitable it will have more as it runs at the same bitrate as ITV1 HD London but has 518,000 more pixels to provide information for- I believe the recommended EBU fixed rate for 1920 is at least 14Mbps, so ITV are running substantially under that at 10.2 so you will get more artifacts on complicated scenes and fast moving multiple objects or water.
jzee
Thank you for your reply - and once again well done for obtaining these pictures. This is somewhat like a detective story and all the more entertaining for it. Are you sure you cannot see any interlace on that Granada picture? I am surprised as it seems quite clear to me. I have no idea what the acquisition format was but it really does not look to be progressively acquired. Do broadcasters or production companies give us some feedback on such things nowadays then? That is a good development if it is the case. The effect cannot either be misalignment of adjacent lines because it is absent in most of the frame but pronounced where credibly there is faster movement. This has got to be interlace surely.
When viewing these two pictures I have each pixel enlarged to about 2mm (I think 800x zoom) and the two fields appear to be clearly delineated. Perhaps you would be kind enough to take another look and reassure me I am not going mad. At this zoom do you also clearly see the faint darkening around the ITV logo too in both the horizontal and vertical axis? Demonstrating the filtering.
Again if you look at the glint on the glasses at this magnification you will see dark and light stripes running perfectly vertically on the Granada version. I perhaps should have been clearer here - this does not look bad at normal resolution at all but when enlarged the detail is clearly not from the original scene. The contrast is too great between two adjacent pixels to be credible; given the optical performance visible in the rest of the picture - also the exact perpendicular nature is too much of a coincidence but we really do not need this as proof I think this is clearly a comperssion artefact caused by losing frequency components in this block.
You might have looked this close before I do not know but if not then do please take a look and see if what I am saying makes sense.
Thanks.
Thank you for your reply - and once again well done for obtaining these pictures. This is somewhat like a detective story and all the more entertaining for it. Are you sure you cannot see any interlace on that Granada picture? I am surprised as it seems quite clear to me. I have no idea what the acquisition format was but it really does not look to be progressively acquired. Do broadcasters or production companies give us some feedback on such things nowadays then? That is a good development if it is the case. The effect cannot either be misalignment of adjacent lines because it is absent in most of the frame but pronounced where credibly there is faster movement. This has got to be interlace surely.
When viewing these two pictures I have each pixel enlarged to about 2mm (I think 800x zoom) and the two fields appear to be clearly delineated. Perhaps you would be kind enough to take another look and reassure me I am not going mad. At this zoom do you also clearly see the faint darkening around the ITV logo too in both the horizontal and vertical axis? Demonstrating the filtering.
Again if you look at the glint on the glasses at this magnification you will see dark and light stripes running perfectly vertically on the Granada version. I perhaps should have been clearer here - this does not look bad at normal resolution at all but when enlarged the detail is clearly not from the original scene. The contrast is too great between two adjacent pixels to be credible; given the optical performance visible in the rest of the picture - also the exact perpendicular nature is too much of a coincidence but we really do not need this as proof I think this is clearly a comperssion artefact caused by losing frequency components in this block.
You might have looked this close before I do not know but if not then do please take a look and see if what I am saying makes sense.
Thanks.
Originally Posted by InMyHO:
“jzee
Thank you for your reply - and once again well done for obtaining these pictures. This is somewhat like a detective story and all the more entertaining for it. Are you sure you cannot see any interlace on that Granada picture? I am surprised as it seems quite clear to me. I have no idea what the acquisition format was but it really does not look to be progressively acquired. Do broadcasters or production companies give us some feedback on such things nowadays then? That is a good development if it is the case. The effect cannot either be misalignment of adjacent lines because it is absent in most of the frame but pronounced where credibly there is faster movement. This has got to be interlace surely.
When viewing these two pictures I have each pixel enlarged to about 2mm (I think 800x zoom) and the two fields appear to be clearly delineated. Perhaps you would be kind enough to take another look and reassure me I am not going mad. At this zoom do you also clearly see the faint darkening around the ITV logo too in both the horizontal and vertical axis? Demonstrating the filtering.
Again if you look at the glint on the glasses at this magnification you will see dark and light stripes running perfectly vertically on the Granada version. I perhaps should have been clearer here - this does not look bad at normal resolution at all but when enlarged the detail is clearly not from the original scene. The contrast is too great between two adjacent pixels to be credible; given the optical performance visible in the rest of the picture - also the exact perpendicular nature is too much of a coincidence but we really do not need this as proof I think this is clearly a comperssion artefact caused by losing frequency components in this block.
You might have looked this close before I do not know but if not then do please take a look and see if what I am saying makes sense.
Thanks.”
“jzee
Thank you for your reply - and once again well done for obtaining these pictures. This is somewhat like a detective story and all the more entertaining for it. Are you sure you cannot see any interlace on that Granada picture? I am surprised as it seems quite clear to me. I have no idea what the acquisition format was but it really does not look to be progressively acquired. Do broadcasters or production companies give us some feedback on such things nowadays then? That is a good development if it is the case. The effect cannot either be misalignment of adjacent lines because it is absent in most of the frame but pronounced where credibly there is faster movement. This has got to be interlace surely.
When viewing these two pictures I have each pixel enlarged to about 2mm (I think 800x zoom) and the two fields appear to be clearly delineated. Perhaps you would be kind enough to take another look and reassure me I am not going mad. At this zoom do you also clearly see the faint darkening around the ITV logo too in both the horizontal and vertical axis? Demonstrating the filtering.
Again if you look at the glint on the glasses at this magnification you will see dark and light stripes running perfectly vertically on the Granada version. I perhaps should have been clearer here - this does not look bad at normal resolution at all but when enlarged the detail is clearly not from the original scene. The contrast is too great between two adjacent pixels to be credible; given the optical performance visible in the rest of the picture - also the exact perpendicular nature is too much of a coincidence but we really do not need this as proof I think this is clearly a comperssion artefact caused by losing frequency components in this block.
You might have looked this close before I do not know but if not then do please take a look and see if what I am saying makes sense.
Thanks.”
Hi InMyHO, I have looked at it looked at Image 2 closely again, and I can see what looks like interlacing artifacts to the right of the man's glasses and in the hair of both the man & woman. This is very strange though, as they cannot be artifacts as a result of deinterlacing as even when I completely disable deinterlacing on my system the screenshot is exactly the same, whereas I would expect many more and prominent horizontal lines as you can see in this non deinterlaced screenshot from Celebrity Juice if deinterlacing is disabled. Can it be that it was filmed in interlaced, then converted to progressive for transmission, but the progressive conversion was imperfect? I have uploaded both videos here, so people can have a look for themselves- ITV1 HD Granada (22MB), ITV1 HD London (29MB). Here is also Image 2 ITV1 HD Granada rescaled to 1440 and then with Gaussian blur level 1 added (png), note the deinterlace artifacts are gone, so it could be ITV1 HD London 1440 does add blur as I suggested which may account for the missing detail.
Originally Posted by Nigel Goodwin:
“No memories involved - simple direct comparison of recordings from then, and the exact same images following the crippling of bit-rates, side by side. As explained endlessly, perhaps you should try reading the threads?.”
“No memories involved - simple direct comparison of recordings from then, and the exact same images following the crippling of bit-rates, side by side. As explained endlessly, perhaps you should try reading the threads?.”
I have been reading the threads which is why I know that others think that there is little or no difference between recordings made with the old & new encoders e.g.
Originally Posted by jzee:
“I did a comparison on a pre encoder change recording with the 9.6 encoder transmission of Joanna Lumley's Northern Lights- to be quite honest there was hardly any difference at all, this old vs new encoder is really a red herring.”
“I did a comparison on a pre encoder change recording with the 9.6 encoder transmission of Joanna Lumley's Northern Lights- to be quite honest there was hardly any difference at all, this old vs new encoder is really a red herring.”
jzee
Thanks again, an excellent picture to compare with - well not an excellent picture in many ways but fit for purpose here.
I too have looked again. I think I am going mad and I am inclined to say now that the material you posted originally was not interlaced. You are right. The rather nasty alternate line stuff simply amounts to artefacts caused by the compression I assume brought about during encoding. The Granada picture is suffering quite heavily from these artefacts while the London one does not. The London one is filtered in both axis - again look at the logo. It is clearly a reduced resolution version. Either these two encoders are tuned significantly differently or one of these is filtered by your receiving equipment. I think you have ruled out the latter and anyway as I mentioned in a previous post it would seem strange to apply vertical filtering on reception.
Thanks again, an excellent picture to compare with - well not an excellent picture in many ways but fit for purpose here.
I too have looked again. I think I am going mad and I am inclined to say now that the material you posted originally was not interlaced. You are right. The rather nasty alternate line stuff simply amounts to artefacts caused by the compression I assume brought about during encoding. The Granada picture is suffering quite heavily from these artefacts while the London one does not. The London one is filtered in both axis - again look at the logo. It is clearly a reduced resolution version. Either these two encoders are tuned significantly differently or one of these is filtered by your receiving equipment. I think you have ruled out the latter and anyway as I mentioned in a previous post it would seem strange to apply vertical filtering on reception.
Originally Posted by nigelbb:
“I have been reading the threads which is why I know that others think that there is little or no difference between recordings made with the old & new encoders e.g.”
“I have been reading the threads which is why I know that others think that there is little or no difference between recordings made with the old & new encoders e.g.”
Going back over a year now, those who claimed no difference were found to be viewing from much too far away to see HD in the first place - that's why they couldn't see any difference.
Originally Posted by tgabber:
“I'd say that rather than obsessing over the (IMHO) marginal differences in particular HD streams...”
“I'd say that rather than obsessing over the (IMHO) marginal differences in particular HD streams...”
If you've looked at those screen shots and think the difference is marginal, then I'm sorry, but HD is wasted on you!
Quote:
“...we should all be concerned with the quality of the actual content instead.”
“...we should all be concerned with the quality of the actual content instead.”
I think there's at least one other forum section for discussing whether the programmes themselves are any good.
I don't think any one here is looking for pretty pictures irrespective of content (there are plenty of demo BluRay discs, and Luxury Television, for that!). Most people enjoy watching at least something on BBC HD, and we'd like it to be pin-sharp, and artefact-free. Most of the time it would be nice if it was in focus, well lit, properly exposed, and moving images didn't stutter.
Quote:
“For me, once the initial excitement of receiving HD transmissions had worn off i stopped tuning into stuff just because it was HD and returned to watching stuff that was well-made, whether it be HD, SD or black and white and grainy (like the rerun of the excellent World at War on BBC2 recently).”
“For me, once the initial excitement of receiving HD transmissions had worn off i stopped tuning into stuff just because it was HD and returned to watching stuff that was well-made, whether it be HD, SD or black and white and grainy (like the rerun of the excellent World at War on BBC2 recently).”
I find it easy to get lost in a good drama, irrespective of technical quality - but across all genres I'd like the technical quality to be good enough that I never even notice it.
Cheers,
David.
It's easy enough to re-check this - jzee, can you post a pair of raw transport streams on mediafire? Then everyone can download them and take a look for themselves.
I'd grab some myself, but it means dropping my box into non-Freesat mode, and it always seems to be recording something when I'm around to do this. I'll get shot if I upset the recording of The Apprentice!
Cheers,
David.
I'd grab some myself, but it means dropping my box into non-Freesat mode, and it always seems to be recording something when I'm around to do this. I'll get shot if I upset the recording of The Apprentice!
Cheers,
David.
To concur with 2Bdecided you have to understand that everyone's viewing experience is different tgabber. In fact I am sure that you already do. Some people will more easily see particular artefacts and effects more easily than others. That is without the vast differences that choice of viewing equipment makes.
Surely everyone has a right to object to a particular quality level or process. It is also true that this can obscure more important issues in the manner you mention of course. How do we get the balance right. I cannot agree with you that quality is good enough at the moment. For me it is below where it should be. However even if it were at the right level we need to take care to maintain it. In truth the broadcasters - all of them - know they are pushing it with HD quality.
The following may be of interst. I did some research a few years ago about broadcast spend and GDP. Ignoring countries like North Korea and Equatorial Guinea the spend for each country on broadcasting is fairly predictable but not , as I initially thought, proportional to GDP per head. Instead the richer a country is the more of its GDP goes on broadcasting. The money comes from taxes, license fees, advertising, subscription, and so on. That may not be a big surprise when you think about it as poorer countries will spend proportionatly more on basics. What might be surprising though is this; the spend on equipment and technology is roughly proportional with GDP - that is the a country with twice the GDP per head will only spend twice as much on equipment in its broadcasting infrastructure. So there is a gap; richer countries spend more on broadcasting but not on the technical quality. The UK is a rich country. It spends a lot per head on broadcasting. Where does the extra spend go? The answer was content. The money ends up in the hands of the talent and the rights owners. These include footballers, actors, presenters and so on. You can literally count on the fingers of one hand the countries who spend more on content than the UK does per head in proportion to GDP. The content may or may not be very good but it is certainly well funded. Do not take my word for it; the European Union has analysed spend within the EU that accidentally shows this trend, and if I remember rightly, questioned the high cost of television in Germany. The talent has their agents fighting for their share of the loot which they do very effectively. Rights owners have their lawyers and capable negotiators. Who though stands-up for the right funding to go to the technical quality? Extra money spent on content arguably tends to just line the pockets of those producing it so someone has to take care to generate pressure to get the balance right.
Surely everyone has a right to object to a particular quality level or process. It is also true that this can obscure more important issues in the manner you mention of course. How do we get the balance right. I cannot agree with you that quality is good enough at the moment. For me it is below where it should be. However even if it were at the right level we need to take care to maintain it. In truth the broadcasters - all of them - know they are pushing it with HD quality.
The following may be of interst. I did some research a few years ago about broadcast spend and GDP. Ignoring countries like North Korea and Equatorial Guinea the spend for each country on broadcasting is fairly predictable but not , as I initially thought, proportional to GDP per head. Instead the richer a country is the more of its GDP goes on broadcasting. The money comes from taxes, license fees, advertising, subscription, and so on. That may not be a big surprise when you think about it as poorer countries will spend proportionatly more on basics. What might be surprising though is this; the spend on equipment and technology is roughly proportional with GDP - that is the a country with twice the GDP per head will only spend twice as much on equipment in its broadcasting infrastructure. So there is a gap; richer countries spend more on broadcasting but not on the technical quality. The UK is a rich country. It spends a lot per head on broadcasting. Where does the extra spend go? The answer was content. The money ends up in the hands of the talent and the rights owners. These include footballers, actors, presenters and so on. You can literally count on the fingers of one hand the countries who spend more on content than the UK does per head in proportion to GDP. The content may or may not be very good but it is certainly well funded. Do not take my word for it; the European Union has analysed spend within the EU that accidentally shows this trend, and if I remember rightly, questioned the high cost of television in Germany. The talent has their agents fighting for their share of the loot which they do very effectively. Rights owners have their lawyers and capable negotiators. Who though stands-up for the right funding to go to the technical quality? Extra money spent on content arguably tends to just line the pockets of those producing it so someone has to take care to generate pressure to get the balance right.
Originally Posted by InMyHO:
“jzee
Thanks again, an excellent picture to compare with - well not an excellent picture in many ways but fit for purpose here.
I too have looked again. I think I am going mad and I am inclined to say now that the material you posted originally was not interlaced. You are right. The rather nasty alternate line stuff simply amounts to artefacts caused by the compression I assume brought about during encoding. The Granada picture is suffering quite heavily from these artefacts while the London one does not. The London one is filtered in both axis - again look at the logo. It is clearly a reduced resolution version. Either these two encoders are tuned significantly differently or one of these is filtered by your receiving equipment. I think you have ruled out the latter and anyway as I mentioned in a previous post it would seem strange to apply vertical filtering on reception.”
“jzee
Thanks again, an excellent picture to compare with - well not an excellent picture in many ways but fit for purpose here.
I too have looked again. I think I am going mad and I am inclined to say now that the material you posted originally was not interlaced. You are right. The rather nasty alternate line stuff simply amounts to artefacts caused by the compression I assume brought about during encoding. The Granada picture is suffering quite heavily from these artefacts while the London one does not. The London one is filtered in both axis - again look at the logo. It is clearly a reduced resolution version. Either these two encoders are tuned significantly differently or one of these is filtered by your receiving equipment. I think you have ruled out the latter and anyway as I mentioned in a previous post it would seem strange to apply vertical filtering on reception.”
Thanks InMyHO, but do we know why would (seemingly) alternate line artifacts would occur on progressively sourced material? I am not saying it doesn't happen, I just don't have enough knowledge of broadcasting to say whether it would or would not. Re: why they do not appear on the ITV1 HD London sample, I think it may be that the artifacts are simply 'disappeared' by the reduced resolution and blurring filter added to the 1440 image at the encoding stage (not by the decoder/renderer on a STB or TV), although that may not be the case and merely down to a peculiarity or faulty setting in the encoder used on ITV1 HD Granada- I am not sure it can be down to lack of bitrate as that scene doesn't look like one that would be particularly bitrate hungry.
Originally Posted by 2Bdecided:
“It's easy enough to re-check this - jzee, can you post a pair of raw transport streams on mediafire? Then everyone can download them and take a look for themselves.”
“It's easy enough to re-check this - jzee, can you post a pair of raw transport streams on mediafire? Then everyone can download them and take a look for themselves.”
Sure, I linked to these above- ITV1 HD Granada (22MB), ITV1 HD London (29MB).
Originally Posted by InMyHO:
“I cannot agree with you that quality is good enough at the moment. For me it is below where it should be. However even if it were at the right level we need to take care to maintain it. In truth the broadcasters - all of them - know they are pushing it with HD quality.
Where does the extra spend go? The answer was content. The money ends up in the hands of the talent and the rights owners. These include footballers, actors, presenters and so on. You can literally count on the fingers of one hand the countries who spend more on content than the UK does per head in proportion to GDP. The content may or may not be very good but it is certainly well funded.”
“I cannot agree with you that quality is good enough at the moment. For me it is below where it should be. However even if it were at the right level we need to take care to maintain it. In truth the broadcasters - all of them - know they are pushing it with HD quality.
Where does the extra spend go? The answer was content. The money ends up in the hands of the talent and the rights owners. These include footballers, actors, presenters and so on. You can literally count on the fingers of one hand the countries who spend more on content than the UK does per head in proportion to GDP. The content may or may not be very good but it is certainly well funded.”
I totally agree, when you look at the salaries of the management in the BBC and ITV, surely a tiny part of that would pay for more transponder space and greater resolution and bitrate. The problem is unless they have an incentive to do so, i.e. people are informed about the issues and lobby for an improvement, it will never happen. I do think some blame must be placed with the government for this current situation though as the lack of space allocated to Freeview almost gives the broadcasters an excuse into not aiming for better on satellite or cable.
Originally Posted by jzee:
“Sure, I linked to these above- ITV1 HD Granada (22MB), ITV1 HD London (29MB).”
“Sure, I linked to these above- ITV1 HD Granada (22MB), ITV1 HD London (29MB).”
Thanks.
Quote:
“I do think some blame must be placed with the government for this current situation though as the lack of space allocated to Freeview almost gives the broadcasters an excuse into not aiming for better on satellite or cable.”
“I do think some blame must be placed with the government for this current situation though as the lack of space allocated to Freeview almost gives the broadcasters an excuse into not aiming for better on satellite or cable.”
I'm not sure that's entirely fair. One old analogue channel can deliver 5-10 digital SD channels or 2-5 digital HD channels. Each legacy broadcaster swapped one analogue channel for half a digital mux, and most transmitters changed from four or five analogue channels to six digital muxes. So individual broadcasters, and broadcasting in general, has far far more useful capacity than in the analogue days.
If it's still not enough, then they need to be more creative. I'm sure they will as the pressure increases.
As you say though, they use it as an excuse to cripple satellite and cable, which is unfair. Especially so when you consider that more houses have satellite or cable than don't - and more houses potentially have access to satellite than Freeview.
Cheers,
David.
I've re-created jzee's screen caps using AVIsynth, directshowsource2, haali media splitter. Same processing for both transport streams.
The results match jzee's. Sorry for doubting it jzee!
Your doubts were misplaced Badvok - turns out it's a fair comparison. What you see is what was broadcast - no more, no less.
Obviously jzee's software is stretching the 1440 pixels to 1920 pixels, but that's no issue - it looks like bog standard bicubic resize (hard to tell, but it's not causing any problems - "better" methods wouldn't help that much on the soft ITV1 HD London source).
1440 ITV1 HD London really is that soft.
To me, it looks like the detail you'd get from decent 1280x1080 (though I doubt that's how it was generated!).
Cheers,
David.
EDIT: no interlacing or interlaced artefacts that I can see, except where my decoder crashes near the end, so that doesn't count!
It's always possible for 25p-in-50i to acquire subtle interlace-like artefacts due to the final encoder mistakenly encoding some frames as pairs of fields instead of frames and adding different artefacts to each field (but even in 50i mode, MPEG encoders can encode whole frames rather than separate fields if/when they choose), but it doesn't mean the picture content itself is really interlaced, or has been badly deinterlaced.
The results match jzee's. Sorry for doubting it jzee!
Your doubts were misplaced Badvok - turns out it's a fair comparison. What you see is what was broadcast - no more, no less.
Obviously jzee's software is stretching the 1440 pixels to 1920 pixels, but that's no issue - it looks like bog standard bicubic resize (hard to tell, but it's not causing any problems - "better" methods wouldn't help that much on the soft ITV1 HD London source).
1440 ITV1 HD London really is that soft.
To me, it looks like the detail you'd get from decent 1280x1080 (though I doubt that's how it was generated!).
Cheers,
David.
EDIT: no interlacing or interlaced artefacts that I can see, except where my decoder crashes near the end, so that doesn't count!
It's always possible for 25p-in-50i to acquire subtle interlace-like artefacts due to the final encoder mistakenly encoding some frames as pairs of fields instead of frames and adding different artefacts to each field (but even in 50i mode, MPEG encoders can encode whole frames rather than separate fields if/when they choose), but it doesn't mean the picture content itself is really interlaced, or has been badly deinterlaced.
Originally Posted by jzee:
“No, I have had HD for about 3 years now, I saw the old BBC HD encoder at 16Mbps, and still have recordings using that encoder, as I explained in my addition to my post, I did a comparison of Joanna Lumley's Northern Lights on the old 16Mbps encoder and compared to the new encoder when it was fixed at 9.6Mbps, there was no difference at all, there was absolutely not more detail in one or the other. A drop in perceived quality was due to increased artifacts when there was fast movement or complicated imagery, this was largely fixed when the channel switched to VBR, although as I explained above there are still situations where the encoder should increase the bitrate but it doesn't which will lead to artifacts and an unrealistic appearance. The simple fact is 1440 x 1080 has 518,400 LESS pixels of information than 1920x1080, 1440 cannot magic this extra information back, it is gone, as many of my examples show, where vertical patterning completely disappears.”
“No, I have had HD for about 3 years now, I saw the old BBC HD encoder at 16Mbps, and still have recordings using that encoder, as I explained in my addition to my post, I did a comparison of Joanna Lumley's Northern Lights on the old 16Mbps encoder and compared to the new encoder when it was fixed at 9.6Mbps, there was no difference at all, there was absolutely not more detail in one or the other. A drop in perceived quality was due to increased artifacts when there was fast movement or complicated imagery, this was largely fixed when the channel switched to VBR, although as I explained above there are still situations where the encoder should increase the bitrate but it doesn't which will lead to artifacts and an unrealistic appearance. The simple fact is 1440 x 1080 has 518,400 LESS pixels of information than 1920x1080, 1440 cannot magic this extra information back, it is gone, as many of my examples show, where vertical patterning completely disappears.”
May I add that Nigel may well be talking about the BBC HD images when they were at 22 Mbs using MPEG2 - which I missed but did see at BBC HD 20 Mbs - before the drop to 18Mbs etc
Cheers, daveac
Originally Posted by daveac:
“May I add that Nigel may well be talking about the BBC HD images when they were at 22 Mbs using MPEG2 - which I missed but did see at BBC HD 20 Mbs - before the drop to 18Mbs etc”
“May I add that Nigel may well be talking about the BBC HD images when they were at 22 Mbs using MPEG2 - which I missed but did see at BBC HD 20 Mbs - before the drop to 18Mbs etc”
No, the drop just over a year ago, when the new encoders were fitted and the quality plummeted.
Before that the BBC HD pictures were the best of any HD channel, the new encoders made it one of the poorest.
3 of 6
