There has been some discussion over the years of the strengths and weaknesses of the HD format and/or their codecs when it comes to banding (aka posterization).

The adoption of 8bit (as opposed to 10 or 12) video formats has been blamed. There has been some discussion of how the various codecs allocate bits (especially in dark scenes).

One thing that has come up from several industry insiders (including compressionists), as well those with some significant understanding of the 8 bit video format and it's gamma characteristics, is that it may often be the DISPLAY that introduces the banding.

While there are undoubtedly cases where poor care in encoding content in to the 8 bit video space has caused banding, or there is room for codec improvement, the post below from Alan Gouger is interesting.

Alan is one of the first few folks to get the new Sony VPL-W200 "Diamond" projector. It's Sony's current high-end offering. Notice what he says about posterization:

I want to mention something I have noticed using the VW200. This projector handles banding or posterization very well. Better then any other projector I have tried.
Titles with fade ins/outs and animation that are known to contain banding just do not show up using this machine. I do not know if they have tweaked their gamma on this machine to perfection or if its in the processing possibly using higher bit rate or something. All I know from what I am seeing banding is almost all but a none issue. Using my VW100 does show banding issues so something has improved.

It would be interesting to test some material material that's been reported to display banding to see in which cases it was the encode, or the display.

On some material I sometimes have no banding at all on my CRT TV, but the banding on the same source is obvious on my LCD TV and my LCD projector.

However, I think it may be a bit of both - the display and the source.

It's most definitely NOT just the display, as it's obvious that some sources just aren't encoded as cleanly as they should be.

P.S. To go a bit off topic... I'm constantly surprised just how freaked out people get over stuff like not having TrueHD on a particular disc, whereas I almost never hear people complaining about the lack of 10-bit video in these formats.

Bugs:

That is because for $380 - you can have Lossless Audio (the price of an Onkyo 605 at Amazon)

So how much do you have to spend to get rid of Banding?

What is the issue OP? Making the best of 8bit color? Or the desire to go to 10bit or 12 bit color? 16bit really seems to be overkill.

P.S. To go a bit off topic... I'm constantly surprised just how freaked out people get over stuff like not having TrueHD on a particular disc, whereas I almost never hear people complaining about the lack of 10-bit video in these formats.

Check out the thread where the question was asked whether you want HD video or lossless audio. A fair number of people said they'd take the audio.

I think a reason why you don't see as much complaining is because 8-bit video is set in the specs, so nothing can really be done about it. Lossless audio, on the other hand, is a viable option that could be there if the studio chooses to include it.

Bugs:

That is because for $380 - you can have Lossless Audio (the price of an Onkyo 605 at Amazon)

So how much do you have to spend to get rid of Banding?

What is the issue OP? Making the best of 8bit color? Or the desire to go to 10bit or 12 bit color? 16bit really seems to be overkill.

The OP (me) is making this point: careful handling of 8bit video (particualrly with gamma) and diligent encodng can provide posterization-free content.

Many of the criticisms laid at the feet of the format and/or codecs might actually be blamed on the displays.

The OP (me) is making this point: careful handling of 8bit video (particualrly with gamma) and diligent encodng can provide posterization-free content.

Many of the criticisms laid at the feet of the format and/or codecs might actually be blamed on the displays.

IMO - based on working with tons of people here at HDF . . .

Almost all people have the Color Saturation level set way too high. In almost every instance where I showed them how to properly set the Color Sat - they were much happier with the new result and saw much less banding because of it.

It is not the display per say (though some are gulity) but people not understanding the Color Sat. level and how you introduce banding when it is not set properly.

And yes - it also has to do with the transfer/encode. If it is there to start with - it will be there unless you have some really expensive video processor to deal with it.

Good to hear 8 bit color source can deliver no banding, depending on display device processing and/or source encode quality.

...so much for DeepColor ;)

Good to hear 8 bit color source can deliver no banding, depending on display device processing and/or source encode quality.

...so much for DeepColor ;)


No DC is VERY important. Not only does it improve the number of colors and increase the "color resolution" - it also increases the number of shades of gray - which adds detail to dark scenes.

IMO - based on working with tons of people here at HDF . . .

Almost all people have the Color Saturation level set way too high. In almost every instance where I showed them how to properly set the Color Sat - they were much happier with the new result and saw much less banding because of it.

It is not the display per say (though some are guilty) but people not understanding the Color Sat. level and how you introduce banding when it is not set properly.

And yes - it also has to do with the transfer/encode. If it is there to start with - it will be there unless you have some really expensive video processor to deal with it.


Of course if it's there in the encode you will see it on the display... that would be why I said in my original post: "While there are undoubtedly cases where poor care in encoding content in to the 8 bit video space has caused banding, or there is room for codec improvement,...".

Undoubtedly incorrectly set displays can introduce problems as well.

But it would seem that there may be a significant number of displays where it's the internal processing that introduces the error, and no amount of end-user twiddling can fix it. Particularly YCrCb --> RGB conversion or panel driver issues.

NOTE: - For readers and lurkers of this thread . . .

Color Banding:

http://www.popularmechanics.com/technology/upgrade/2826881.html

Of course if it's there in the encode you will see it on the display... that would be why I said in my original post: "While there are undoubtedly cases where poor care in encoding content in to the 8 bit video space has caused banding, or there is room for codec improvement,...".

Undoubtedly incorrectly set displays can introduce problems as well.

But it would seem that there may be a significant number of displays where it's the internal processing that introduces the error, and no amount of end-user twiddling can fix it. Particularly YCrCb --> RGB conversion or panel driver issues.

Have we touched upon the subject of "dithering" yet? I don't think so.

Something I'd like clarification on - I'm confused.

Would an HDMI 1.3 HD player connected to an HDMI 1.3 display be less likely to show banding in some cases than an earlier HDMI version? I was thinking this allows the necesary full path for upsampling to 10 bit video.

Something I'd like clarification on - I'm confused.

Would an HDMI 1.3 HD player connected to an HDMI 1.3 display be less likely to show banding in some cases than an earlier HDMI version? I was thinking this allows the necesary full path for upsampling to 10 bit video.

Yes - HDMI 1.3 supports 10Bit and up:

http://www.hdmi.org/pdf/HDMI_Insert_FINAL_8-30-06.pdf

But it is only the "pipe."

Yes - HDMI 1.3 supports 10Bit and up:

http://www.hdmi.org/pdf/HDMI_Insert_FINAL_8-30-06.pdf

But it is only the "pipe."

So, is an HDMI 1.3 HD player a theoretical advantage over an HDMI 1.2 player *IF* the display is HDMI 1.3?

So, is an HDMI 1.3 HD player a theoretical advantage over an HDMI 1.2 player *IF* the display is HDMI 1.3?

Yes - if Deep Color ever came to HDM - not in the specs by the way for either format. Doesn't mean the specs can't be changed though.

IMO - based on working with tons of people here at HDF . . .

Almost all people have the Color Saturation level set way too high. In almost every instance where I showed them how to properly set the Color Sat - they were much happier with the new result and saw much less banding because of it.

It is not the display per say (though some are gulity) but people not understanding the Color Sat. level and how you introduce banding when it is not set properly.

And yes - it also has to do with the transfer/encode. If it is there to start with - it will be there unless you have some really expensive video processor to deal with it.

good point about color saturation. I rarely see that brought up in these "banding" discussions

Yes - if Deep Color ever came to HDM - not in the specs by the way for either format. Doesn't mean the specs can't be changed though.

What if a company built an HD player (either brand) that processes the video internally at a higher resolution and can then output that directly to the TV via HDMI 1.3 at 24/32/48-bits? Assuming of course that both devices support "deep color".

This was just occurring to me as I was thinking about my FX1 camera which processes the image internally at 1920x1080i, 4:4:4 and 14-bits (if I recall correctly) and then of course down-samples that to 1440x1080i, 4:2:0 and 8-bits to fit the HDV specs.

My numbers may be off. To google!

Do I have this list correct - as far as color bit depth (low to high):

1. 8bit
2. x.v.YCC
3. 10bit
4. 12bit
5. 16bit

What if a company built an HD player (either brand) that processes the video internally at a higher resolution and can then output that directly to the TV via HDMI 1.3 at 24/32/48-bits? Assuming of course that both devices support "deep color".

This was just occurring to me as I was thinking about my FX1 camera which processes the image internally at 1920x1080i, 4:4:4 and 14-bits (if I recall correctly) and then of course down-samples that to 1440x1080i, 4:2:0 and 8-bits to fit the HDV specs.

My numbers may be off. To google!
I think it might help if you have lots of CGI overlays/PiP compositing by the player.

I think such a player, if playing a movie full screen might be able to 'smooth' over banding in some way. But wouldn't it be 'inventing' colours that weren't necessarily there? I think it would be much better to have the video actually encoded with a 10 bit or whatever colour palette. And I don't think it would always increase the bitrate too much (ie. imagine a 'still' background over moving credits like the end of King Kong).

SquirrelPhister... Yes, you'll likely see this... Marketing will love it. You already see "Deep Color" tags everywhere with near nothing to drive it. The actual benefit of "Deep Color Upconversion" would need to be proven, of course.

As the source HDM standards are not Deep Color and won't be in this generation, the advantage would need to be solely in the processing and conversion. What can actually be 'improved' is up to the engineer. This also assumes modern Deep Color capable TVs can honestly resolve it. :) Wait, marketing says that doesn't matter. Nevermind. :D

Do I have this list correct - as far as color bit depth (low to high):

1. 8bit
2. x.v.YCC
3. 10bit
4. 12bit
5. 16bit
It would be better to separate out xvYCC, since it can be used at any bit depth. BTW, xvYCC is possible on HDM with a small tweak to the specifications. I'm not exactly 100% sure, but I believe it would also be backward compatible (wide gamut colors would just be clipped to the REC 709 gamut on legacy displays).

Anything above 8-bit is a huge problem because the encoders and decoders would have to change.

Ron

It would be better to separate out xvYCC, since it can be used at any bit depth. BTW, xvYCC is possible on HDM with a small tweak to the specifications. I'm not exactly 100% sure, but I believe it would also be backward compatible (wide gamut colors would just be clipped to the REC 709 gamut on legacy displays).

Anything above 8-bit is a huge problem because the encoders and decoders would have to change.

Ron

Here is my understanding of the term "x.v.YCC" as it is being used today to sell better HD-CAM's . . .

x.v.YCC is 1.8X the number of colors that HD has. It increases the number of colors from 16 million to 30 million giving a more "life like" image.

Correct?

I thought HD only used around 11 million colours - about the same as DVD but a a different palette?

I thought HD only used around 11 million colours - about the same as DVD but a a different palette?

http://en.wikipedia.org/wiki/Color_space

Color space density

The RGB color model is implemented in different ways, depending on the capabilities of the system used. By far the most common general-use incarnation as of 2006 is the 24-bit implementation, with 8 bits, or 256 discrete levels of color per channel. Any color space based on such a 24-bit RGB model is thus limited to a gamut of 256×256×256 ≈ 16.7 million colors. Some implementations use 16 bits per component, resulting in the same range with a greater density of distinct colors. This is especially important when working with wide-gamut color spaces (where most of the more common colors are located relatively close together), or when a large number of digital filtering algorithms are used consecutively. The same principle applies for any color spaces based on the same color model, but implemented in different bit depths.

Well I was close!:o

Also

"xvYCC uses the full range of values (1 to 254 in an 8-bit space) to represent colors. In BT.601 and BT.709, RGB colors are represented only by 8-bit values from 16 to 235. This limited range was established to allow for undershoot and overshoot, attributes of analog TV signaling. With digital TV signaling, there is no undershoot or overshoot, and the values from 1-15 and 235-254 can be used to represent real colors. In order to maintain backward-compatibility with earlier standards, the red (R), green (G), blue (B) and white standard colors are still calculated at the same indices in the color space. The wider ranges of digital values allow representation of deeper greens, deeper reds, and deeper blues - and of course intermediate colors previously beyond the boundary limit in the CCIR 601 color space.

Even the high-definition television color space defined by BT.709-5 encompasses only 45 per cent of the possible values with RGB indices of 1 to 254, since those indices are limited to 16 to 234 by the BT spec. The xvYCC standard uses 1 to 255, and can encode more than twice the number of color values. (The sRGB system is in between the BT.706-1 and xvYCC systems, allowing use of all combinations of R, G and B values from 16 to 235 - more combinations, but not so many as in xvYCC.)"

from http://en.wikipedia.org/wiki/XvYCC

Do I have this list correct - as far as color bit depth (low to high):

1. 8bit
2. x.v.YCC
3. 10bit
4. 12bit
5. 16bit

Color bit depth and colorspace are two different parameters.

xvYCC, as described in the links and citations, is a colorspace. Each of the popular colorspaces can be used at the various bit resolutions you listed.

Wiki is our friend:D

Do I have this list correct - as far as color bit depth (low to high):

1. 8bit
2. x.v.YCC
3. 10bit
4. 12bit
5. 16bit

Bit depth (low to high):

1. 8-bit
2. 10-bit
3. 12-bit
4. etc.


Color range within a given bit depth (low to high):

BT.601/BT.709/RGB16-235
xv.YCC
RGB0-255


Colorimetry (low to high):

BT.601 (SD)
BT.709 (HD)
HDTVs (each one different, almost none meet the specs)


The three above parameters are not dependent on each other.

OK - then please make a simple list like I did to show us this.

Would it look like this?

1. 8bit/ x.v.YCC
2. 10bit/ x.v.YCC
3. 12bit/ x.v.YCC
4. 16bit/ x.v.YCC

Looks like kjack beat me to it ;)

For the record, I've always been a full-range kind of RGB :D

...kinda fun having those initials since birth :)

Looks like kjack beat me to it ;)

For the record, I've always been a full-range kind of RGB :D

...kinda fun having those initials since birth :)

FINALLY! I can't believe it! After ALL these years!

Hello; Roy G. Biv

It's great to finally meet you!:D

What if a company built an HD player (either brand) that processes the video internally at a higher resolution and can then output that directly to the TV via HDMI 1.3 at 24/32/48-bits? Assuming of course that both devices support "deep color".

When I play an HD DVD at 1080p on my HD-A35, my display says "1080p 36 bit" -- so I think some players are already doing something like this. My guess is that the player is just shifting the bits left 4 positions without any interpolation or fancy processing, though.

I have an Optoma HD70 720p DLP front projector, and the HDMI input only has an 8-bit colour depth while the component input has a 10-bit colour depth.

I absolutely notice some banding issues when going from my HDA2 to my HD70 when using HDMI. Specificly on facial colouring (rosy cheeks), fog and smoke (King Kong).

But when I switch over to component all of that banding is gone. The colours easily and cleanly blend with eachother.

I have an Optoma HD70 720p DLP front projector, and the HDMI input only has an 8-bit colour depth while the component input has a 10-bit colour depth.

I absolutely notice some banding issues when going from my HDA2 to my HD70 when using HDMI. Specificly on facial colouring (rosy cheeks), fog and smoke (King Kong).

But when I switch over to component all of that banding is gone. The colours easily and cleanly blend with eachother.

It's refreshing to hear good things about component video. Yet another reason you don't want HDMI, in addition to HDCP issues, etc.

Is it just me, or is there a growing contrarian trend on avs? ;)

I have an Optoma HD70 720p DLP front projector, and the HDMI input only has an 8-bit colour depth while the component input has a 10-bit colour depth.

I absolutely notice some banding issues when going from my HDA2 to my HD70 when using HDMI. Specificly on facial colouring (rosy cheeks), fog and smoke (King Kong).

But when I switch over to component all of that banding is gone. The colours easily and cleanly blend with eachother.


This is just 10bit or greater dither in the DAC coupled with a light sprinkling of noise : it can hide banding but its nothing fancy , think of it as smearing. I actually quite like analogue component myself.

As for 8bit video having banding. Yes it does , almost any type of imagery will have quantisation artifacts somewhere with 8bit imagery, the trick is whether its obvious or not.

However most of the really visible banding people complain about is usually down to the display device. Given access to 8bit and 10bit video versions of the same material it would be very hard to differentiate between the two on most imagery even if you were aware of what to look for.

So 8bit video is jut fine for consumer formats. To justify a higher bitdepth you'd reallyneed to move to a higher dynamic range than just "video" . Something like Dcinema or a print film dynamic.

Yes NOW - is a very good time to discuss "Dithering"

http://en.wikipedia.org/wiki/Dither

http://www.colorcube.com/illusions/dither.htm

It's refreshing to hear good things about component video. Yet another reason you don't want HDMI, in addition to HDCP issues, etc.

Is it just me, or is there a growing contrarian trend on avs? ;)
I think component is great too, but I find the lack of 1080p24 support over component annoying. 1080p24 support over component is possible but usually not allowed.

So 8bit video is jut fine for consumer formats. To justify a higher bitdepth you'd reallyneed to move to a higher dynamic range than just "video" . Something like Dcinema or a print film dynamic.I apologize in advance but I wanted to quote this post. It validates a conjecture of mine regarding bit depth and dynamic range. Nominal peak dynamic range for "Print Film" is ~200:1 and peak negative film latitude is ~1000:1, also keep in mind film tracks color logarithmically like the human eye.

edit/

http://www.quantel.com/resource.nsf/Files/Digital_Film/$FILE/Digital_Film.pdf

*note copy and paste url*

"A modern camera negative stock will have almost a 10-stop exposure range representing a 500:1 brightness range. However, a typical scene will be lit to give a contrast nearer to 100:1 (approximately 6.7 stops) or, for higher contrast, 300:1. But print film is different and will only show a portion of the scene contrast stored on the negative. Generally, black to white on the print corresponds to a scene contrast of 200:1 so a 300:1 scene contrast means the Director of Photography (DoP) needs to make a decision on how to print the material (see Tutorial: Film Basics). Does he roll off the highlights to show the deep shadow detail or crush the shadows to reveal the detail in the highlights?"
www.quantel.com