Now I have HDMI 1.3, deep color, 1080p-24, and 14 bit color processing.

But where's the programs/software? Anyone know something? Clue me/us in. Thanks.

It is not supported in either BD or HD DVD specs...so nothing very soon.

This is one of the sad things.

A huge amount of energy has gone into the format war, but both formats have unnecessary built-in restrictions that will preclude further growth of quality.

One of my reasons for supporting Blu-ray has been that it had higher capacity and bandwidth. But, even so, the color limitations on both formats are frustrating.

Whichever one wins, we are getting a suboptimal solution.

It is possible that more color information can be added in an upward-compatible manner (similar to HDCD) but I have not heard anyone talking about this yet.

We may never see it.

Home videos and possibly some games down the road. That's about it until the next format comes. Of course by then.........

This is one of the sad things.

A huge amount of energy has gone into the format war, but both formats have unnecessary built-in restrictions that will preclude further growth of quality.

One of my reasons for supporting Blu-ray has been that it had higher capacity and bandwidth. But, even so, the color limitations on both formats are frustrating.

Whichever one wins, we are getting a suboptimal solution.

It is possible that more color information can be added in an upward-compatible manner (similar to HDCD) but I have not heard anyone talking about this yet.

We may never see it.

Not including >8bits is bad but somewhat understandable.

Not including support for xvYCC is just plain stupid. In case anyone doesn't know current colorspace is not even true 8bit. Colors can only use values of 16-235 the values 0-15 and 236-255 are wasted. This means content can only contain information about 62% of the 8bit colorspace. It would cost 0% more bandwidth and storage space to implement yvYCC. The reduced colorspace is an anachronism from analog TV days and the "next gen" formats are still hobbled by it. Support (even optional) was not included in the specs for either format.

The sad thing is you can buy a home video camera today (using xvYCC) that has a larger colorspace than HD media.

Now I have HDMI 1.3, deep color, 1080p-24, and 14 bit color processing.

But where's the programs/software? Anyone know something? Clue me/us in. Thanks.

What you have is 1.3, 1080x24P and 14 bit processing. There is no Deep Color. There are no sources for the signal and there are no displays that can process it. DC takes Color from 8 bit to 16 bit. So we go from 24 to 48 (3 colors each the same value) This allows the Color Gamut to go from millions to billions of colors.

xvYCC is a half measure, It will only allow an increase of 1.7 times the number of colors.

The benefit to all this is to get rid of color banding. We are almost there. 24P eliminites both 3:2 Pull Down and Interlace. Deep Color eliminates Color Banding,

These three items are the main source for problems in the images we see.

A link on Deep Color and xvYCC and how they differ from HDMI.org

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

A link explaining and showing images of color banding

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

Here is an issue that plays into to equation. Both the display and the HDM player have to be HDMI 1.3. HDTV's with HDMI 1.3 - this is a new feature started in the last 12 months so very few have it.

So we may see xvYCC next year. Sony wants to do it. Chances are we may never see Deep Color as all the HDTV's would have to be replaced. By then we will go to a higher format like 2500x2000 or 4000x2000 and DC will be used in those specs.

My new Samsung TV has an option xvYCC on/off, turning it on seems to make green more blueish. Is it supposed to be on or off for blu-ray disc playback on the PS3?

My new Samsung TV has an option xvYCC on/off, turning it on seems to make green more blueish. Is it supposed to be on or off for blu-ray disc playback on the PS3?

Better leave it off because again - no BD movie has the signal embedded in it. Once a movie comes out with the signal - then you would turn it on. And only for that movie. All others would be off.

Here is an issue that plays into to equation. Both the display and the HDM player have to be HDMI 1.3. HDTV's with HDMI 1.3 - this is a new feature started in the last 12 months so very few have it.

So we may see xvYCC next year. Sony wants to do it. Chances are we may never see Deep Color as all the HDTV's would have to be replaced. By then we will go to a higher format like 2500x2000 or 4000x2000 and DC will be used in those specs.

Sawright...except for this. My PS3 boasts the 1.3 spec as does my new display. 14 bit processing engines are now found on the new Toshiba and Philips LCD displays. They tout "trillions", not billions, of colors. 1.7x sounds like childs play.

Sawright...except for this. My PS3 boasts the 1.3 spec as does my new display. 14 bit processing engines are now found on the new Toshiba and Philips LCD displays. They tout "trillions", not billions, of colors. 1.7x sounds like childs play.

But your display cannot do 48bit color processing - NO DISPLAY CAN. You need more than just HDMI 1.3. That is the "pipe" needed to carry the non-existant Deep Color signal.

It can do xvYCC but again you need an embedded signal in a movie - none exist yet. That is only 1.7 times what we alread have. NOT 100X.

can the PS3 "up-sample" standard blu-ray discs to xvYCC and would there be any point?

But your display cannot do 48bit color processing - NO DISPLAY CAN. You need more than just HDMI 1.3. That is the "pipe" needed to carry the non-existant Deep Color signal.

It can do xvYCC but again you need an embedded signal in a movie - none exist yet. That is only 1.7 times what we alread have. NOT 100X.

Granted, but I don't see the connection. Why does anyone need 48 bit (16x3) anyway? Your links confirm that is way beyond the color palette the human eye sees.
24 bit color is often expressed as 16.7 million colors in the specs. 1.7x comes out to 28.4 million colors. Not exactly earth shaking in lieu of what the displays can handle.

Granted, but I don't see the connection. Why does anyone need 48 bit (16x3) anyway? Your links confirm that is way beyond the color palette the human eye sees.
24 bit color is often expressed as 16.7 million colors in the specs. 1.7x comes out to 28.4 million colors. Not exactly earth shaking in lieu of what the displays can handle.

Yes it really does not make much of a difference does it. Sounds like another marketing tool to me like Full HD. It will make a difference but not enough to get rid of Color Banding. It may help in the number of shades of Gray for the Gray Scale. This would be important.

Deep Color gets rid of all this:

A link explaining and showing images of color banding

http://www.popularmechanics.com/tec...de/2826881.html

can the PS3 "up-sample" standard blu-ray discs to xvYCC and would there be any point?

IMO - no. As I understand it. The color signal is increased a small bit in bandwidth to get the increase. It would be like Having a DD 5.1 only receiver. Then DTS 5.1 comes out. Nothing you can do to get the "new" audio standard.

Again you need a HD player that has a HDMI 1.3 output, A HDTV that has a HDMI 1.3 input and a color processing engine to make use of the increase in the color bandwidth present on the HD xvYCC encoded disc.

The PS3 has the HDMI 1.3 output already. There is half of the equation. Just need the display to be ready when xvYCC is introduced.

IMO - no. As I understand it. The color signal is increased a small bit in bandwidth to get the increase. It would be like Having a DD 5.1 only receiver. Then DTS 5.1 comes out. Nothing you can do to get the "new" audio standard.

Again you need a HD player that has a HDMI 1.3 output, A HDTV that has a HDMI 1.3 input and a color processing engine to make use of the increase in the color bandwidth present on the HD xvYCC encoded disc.

The PS3 has the HDMI 1.3 output already. There is half of the equation. Just need the display to be ready when xvYCC is introduced.

I think you are missing the point. Neither BD or HD DVD has the ability to include xvYCC data in their specs. Sure the PS3 can output it IF and only IF it is given a source that has xvYCC. A BD (and HD DVD) disc will never be xvYCC because it would not work on existing players and TV.

Now if either side had any long term goal they could have made xvYCC optional on the media and on the payer. All players would be required to output normal colorspace. If they had done that likely the players would have an option Colorspace: normal, expanded, auto. Cheaper players would simply always output to normal colorspace.

So you would have 4 options:
expanded disc, player on expanded mode (output 0-256RGB)
expanded disc, player on normal mode (dither down to 16-235RGB)
normal disc, player on expanded mode (output 16-235RGB - no improvement)
normal disc, player on normal mode (output 16-235RGB - what we have now)

They for whatever reason did not so neither HD DVD or BD will ever output expanded color data. The sheer irony would be that the PS3 could be the first to output expanded color data once Sony starts selling movies online. There is no technical reason preventing online content from using the expanded colorspace (xvYCC). I am sure political reasons would prevent that. It would not look good if a 720p downloaded encode can do more than a 1080p disc. Might make consumers question why they need discs?

I think you are missing the point. Neither BD or HD DVD has the ability to include xvYCC data in their specs. Sure the PS3 can output it IF and only IF it is given a source that has xvYCC. A BD (and HD DVD) disc will never be xvYCC because it would not work on existing players and TV.

Now if either side had any long term goal they could have made xvYCC optional on the media and on the payer. All players would be required to output normal colorspace. If they had done that likely the players would have an option Colorspace: normal, expanded, auto. Cheaper players would simply always output to normal colorspace.

So you would have 4 options:
expanded disc, player on expanded mode (output 0-256RGB)
expanded disc, player on normal mode (dither down to 16-235RGB)
normal disc, player on expanded mode (output 16-235RGB - no improvement)
normal disc, player on normal mode (output 16-235RGB - what we have now)

They for whatever reason did not so neither HD DVD or BD will ever output expanded color data. The sheer irony would be that the PS3 could be the first to output expanded color data once Sony starts selling movies online. There is no technical reason preventing online content from using the expanded colorspace (xvYCC). I am sure political reasons would prevent that. It would not look good if a 720p downloaded encode can do more than a 1080p disc. Might make consumers question why they need discs?

Is there a link for this? - just want to know for my own reference - not questioning you - just want verification of some kind.

If this is so, I do know where xvYCC is going to be used - HD Camcorders

Here is the description for the Brand new Sony LCD coming out any day Sony calls xvYCC - x.v.Color:

Sony's BRAVIA W series Full HD 1080p televisions take performance to the next level with advanced HDMI v1.3 features such as x.v.Color which greatly broadens the color space input capabilities to include 1.8 times as many natural colors as existing HDTV signals. In addition, Deep Color input capability works with the 10-bit processor and panel to deliver 64 times the level of color expression versus current 8-bit systems. Wrap all of this up with an elegant a new brushed metal picture frame design and there's nothing like W-Series HDTV's.

52" BRAVIA® W series LCD Flat Panel HDTV

http://www.sonystyle.com/webapp/wcs/stores/servlet/ProductDisplay?catalogId=10551&storeId=10151&langId=-1&productId=8198552921665119487

Sawright...except for this. My PS3 boasts the 1.3 spec as does my new display. 14 bit processing engines are now found on the new Toshiba and Philips LCD displays. They tout "trillions", not billions, of colors. 1.7x sounds like childs play.

Phillips is just playing marketing. The processing engine may have a 14bitdepth however all consumer panels (with exception of a couple $20K+ models) are 8bit panels.

8 bit means 2^8= 256. Each subpixel red, green, blue can have one of 256 options (0-255).

Your display "could" output 256 * 256 * 256 = 16.7 million possible colors

Right now without a xvYCC source your TV displays less than half that.
First colorspace used for TV is limited to the values 16-235 meaning a signal from your DVD player, or a broadcast TV, or even a laser disk will never have a value <16 or >235. So instead of 256 possible values there are 219 for each subpixel

Your display "could" output 219*219*219 = 10.6 million possible colors. That's pretty bad we lost about 6 million possible colors for no reason but it gets worse BT.709-5 the spec that handles content transmission removes some values from the table due to issues with CRT having problems on undershoot and overshoot.

So in reality your brand new HDTV can only display the content that is fed to it and at best it is around 7 million colors possible colors. You are losing nearly 10 million colors due to unneeded limitations that have been around since TV first became color.

All xvYCC does is remove all the restrictions and allows content to be encoded with any value for 0-255. The amazing part is since data is stored or transmitted as an 8 bit value anyways you get about twice the possible colors for free. It doesn't cost one extra bit. It is hard to see what you are missing when every TV since dawn of TV has had this limitation. When you have seen two displays side by side one with expanded color and the without it is night and day. It really would make HD media "pop" to have richer expanded color.

Of course your TV manufacture will never tell you that. Big numbers sell so they say 1.67 million. Which is true IF your TV received a signal with values from 0-255 it would display 1.67 million colors. Without xvYCC your TV simply will never receive a signal with that wide of a color space.

Sadly neither HD DVD nor BD implemented xvYCC. Moving to 10bit, 12bit, 16bit color will take a LOOOOOOOOONG time. All panels are simply 8bit right now and the few plants that can produce 12bit panels are selling them to hospitals for medical imaging for 50K+ making little incentive for them to push the price of the technology down anytime soon. Also higher bitdepth will require 12%, 25%, 50%, or 100% more bandwidth which will be huge limit on even BD since the studios would much rather use that space for extras. So deepcolor will remain a dream for a long time and I can kinda understand why neither HD DVD or BD provided support for it. xvYCC on the other hand is free.

Anyone (insider?) know why xvYCC was not included in the spec for either BD or HD DVD?

The xvYCC option is quite hidden within the depths of the menu system of my Samsung TV. Does this mean I would have to turn it on and off depending on whether the media is xvYCC encoded or not? How would I tell?

namechamps:

some calculation errors:

256x256x256 is 16.7 million

219x219x219 is 10.5 million

so we lost 6.2 million colors

and the 780k figure should thus be around 7 million, so that the user's display has "lost" 10 million colors.

Slight correction. xvYCC provides 1.8X - not 1.7X

Here is the blurb from the top of the line Sony HD Camcorder:

x.v. Color™ Technology

x.v.Color can capture or display nearly twice as many (1.8x) viewable colors than currently possible. This provides a more accurate and vivid color range, for a more natural, lifelike overall experience.

The xvYCC option is quite hidden within the depths of the menu system of my Samsung TV. Does this mean I would have to turn it on and off depending on whether the media is xvYCC encoded or not? How would I tell?

What is the model #?

What is the model #?

LE37M86BDX, attached is a photo of the option in the menu, and the TV's instruction manual (which I think you'll agree doesn't help much!).

Nothing - I searched all over Samsung U.K. and there is nothing on it. Only in the manual as you showed.

The only one that discusses this is Sony.

So according to the poster above xvYCC is not part of HD DVD specs (which I did find via the DVD Forum - 78 page PDF) yet in your manual it mentions DVD player. Huh? (not questioning poster - questioning Samsung) I know DVD doesn't have it.

But again leave it off and if you are in the market for a HD Camcorder - get one with x.v.Color. :)

Here is the link for the HD DVD Specifications PDF:

http://www.dvdforum.org/images/Requirements%20Specification%20for%20HD%20DVD%20Video%20Appl ication-July2005.pdf

The banding you are being critical of is highly likely to be caused primarily by the display rather than the 8bit nature of the video.

So we may see xvYCC next year. Sony wants to do it. Chances are we may never see Deep Color as all the HDTV's would have to be replaced. By then we will go to a higher format like 2500x2000 or 4000x2000 and DC will be used in those specs.

I would be incredible surprised to see these resolutions @ 48 bit in my lifetime or my kidds lifetime. The entire broadcast sytem in the United States would have to be overhauled and gutted. I can tell you that it will never go through a cable coax @ 48 bit depths. With Fiber yes with challenges. Even with these newer encoding standards you are still looking at gigbits per second. I read an article that the University of Washington sent an 8k signal at 12 bits and it still maxed out the server and had a vbr of 90 gbps. You can't cram that type of bandwidth over any installed network in the United States.

I would be incredible surprised to see these resolutions @ 48 bit in my lifetime or my kidds lifetime. The entire broadcast sytem in the United States would have to be overhauled and gutted. I can tell you that it will never go through a cable coax @ 48 bit depths. With Fiber yes with challenges. Even with these newer encoding standards you are still looking at gigbits per second. I read an article that the University of Washington sent an 8k signal at 12 bits and it still maxed out the server and had a vbr of 90 gbps. You can't cram that type of bandwidth over any installed network in the United States.

Very interesting. Then why the development of Deep Color? Why make a "fuss" if it something we may never see? I do agree with you. I cannot see people going out and replacing all the brand new HDTV's they bought which would be necessary.

I never thought the application would be for anything other than HDM. If at all.

Do the new 4K RED cameras have this kind of colorspace?

Not including >8bits is bad but somewhat understandable.

Not including support for xvYCC is just plain stupid. In case anyone doesn't know current colorspace is not even true 8bit. Colors can only use values of 16-235 the values 0-15 and 236-255 are wasted. This means content can only contain information about 62% of the 8bit colorspace. It would cost 0% more bandwidth and storage space to implement yvYCC. The reduced colorspace is an anachronism from analog TV days and the "next gen" formats are still hobbled by it. Support (even optional) was not included in the specs for either format.

The sad thing is you can buy a home video camera today (using xvYCC) that has a larger colorspace than HD media.

Yes, I was including these issues as well in my lamentation, I think we agree.

I would like to see 4320 resolution, 120 FPS, and deep color as a production standard. This could be downconverted to just about everything and would provide a true archive of the original. It would also support continuous upgrades in both exhibition and home use.

It is too bad that the true visionaries like Cooper and Todd are no longer with us. They would have understood that we have painted ourselves into a digital corner.

namechamps:

some calculation errors:

256x256x256 is 16.7 million

219x219x219 is 10.5 million

so we lost 6.2 million colors

and the 780k figure should thus be around 7 million, so that the user's display has "lost" 10 million colors.

oops. damn decimal place. corrected.
thanks.

Has anyone ever seen a demo of Deep Color? I know there have been a few. I would love to hear about someone's real life experience seeing it compared to today's colorspace.

You don't need more than 10bits to encode a video intensity range beneath the threshold of visible banding.

The actual visual differences between 8bit and 10bit video are subtle at best on broadcast equipment let alone domestic displays very few of which can adequately resolve an 8bit range let alone anything higher ( I think the JVC HD1/RS1 projector is actually transparent with 8bit video at least in terms of banding).

With perceptual encoding you can fit a film negative density range into 10bit with very high levels of transparency ( you can actually even get away with 8bit for 90% of imagery).

To make use of more bits you'd have to move into a higher dynamic range format and step away from a video intensity range. Something like a print dynamic or a dynamically lut'd negative format

The red camera is 12bit linear although how this transfers to something that supposedly mimics a film negative density range is somewhat mysterious. Even 16bit linear is not regarded as enough to transparently "flatten" ( I hesitate to say degamma) a 10bit log encoded film scan at least perceptually. I'm suspecting noise in the blacks "hides" banding.

The actual visual differences between 8bit and 10bit video are subtle at best on broadcast equipment let alone domestic displays very few of which can adequately resolve an 8bit range let alone anything higher ( I think the JVC HD1/RS1 projector is actually transparent with 8bit video at least in terms of banding).

.

By definition, this also implies a limitation to human vision. But, 16.7 million colors hardly sounds like a billion to me. Can we see a billion colors?

By definition, this also implies a limitation to human vision. But, 16.7 million colors hardly sounds like a billion to me. Can we see a billion colors?

Colour is different to banding perception threshold. The amount of colours we can see is a purely subjective thing to gauge. Large groups of people sat in rooms being shown different images to ascertain some sort of overall pattern of colour perception kind of thing.

I wouldn't necessarily want to get into whether we can see millions or billions of colours with regard to a topic about bit depth and banding.

Your display "could" output 219*219*219 = 10.6 million possible colors. That's pretty bad we lost about 6 million possible colors for no reason but it gets worse BT.709-5 the spec that handles content transmission removes some values from the table due to issues with CRT having problems on undershoot and overshoot.


As a technical aside, from information theory the number of bits is roughly the base 2 log of the number of (equally likely) possible values. So if the setups cause each color plane to have only 219 possible values then it is only about 7.775 bit color (2^7.775 = 219.03). That may seem very close to 8 bit but these things increase exponentially which it is why we go from 16 million to a billion colors just by going from 8 to 10 bits per color.

- Tom

So what actually causes banding?

I always assumed it was a result of a non-linear colour spectrum response of displays, much like how flourescent lights have peaks and troughs in their colour outputs.

Did you know humans are twice as sensitive to variances in the green spectrum range than any other colour?

My mind boggles at the fact that we can perceive more than 16.7million colours, yet most of us can see the difference between 16bit and 32bit on our computer displays, are those bit resolutions not in the same scale as those used on video?

So what actually causes banding?

I always assumed it was a result of a non-linear colour spectrum response of displays, much like how flourescent lights have peaks and troughs in their colour outputs.


Intensities being mapped together because of insufficient bit depth to represent a transition precisely enough to fall below the threshold of human intensity step perception.


Did you know humans are twice as sensitive to variances in the green spectrum range than any other colour?

Why do people use red light at night and in dark enviroments? So as not to knacker your night vision ( least sensitive colour having less impact if you have to suddenly turn your light off). We're not so hot on blue either. Component colour encoding also exploits this.


My mind boggles at the fact that we can perceive more than 16.7million colours, yet most of us can see the difference between 16bit and 32bit on our computer displays, are those bit resolutions not in the same scale as those used on video?

Not the same scale. 32bit in computer land is actually 8bit colour : RGBA with 8bits in each channel (A is the alpha channel so in truth its only reallyu 24bit colour if you want to express it that way). For purposes of this discussion 8bit video is 8bits per RGB channel.

My mind boggles at the fact that we can perceive more than 16.7million colours, yet most of us can see the difference between 16bit and 32bit on our computer displays, are those bit resolutions not in the same scale as those used on video?

16bit color in this case is 2^16 = 65,536 colors
32bit color afaik is 24bit + 8 bit alpha, so in essense it is 2^24 = 16.7 million colors + and 8bit (256 color) alpha channel.


EDIT:
these descriptions appear to be reasonable:
http://en.wikipedia.org/wiki/Highcolour 15/16 bit
http://en.wikipedia.org/wiki/Truecolor 24/32 bit

So what actually causes banding?

I always assumed it was a result of a non-linear colour spectrum response of displays, much like how flourescent lights have peaks and troughs in their colour outputs.

Did you know humans are twice as sensitive to variances in the green spectrum range than any other colour?


Banding is a result of digital steps in the color spectrum.

I don't think we should be caught up in the words millions or billions. The color gamut is 3-dimensional. We need 3 numbers to specify it. The fact that in YUV color space, there are only 220 values for each dimension, gives us a problem. This is not a perceptual colorspace either, but a gamma-corrected one.

In a perceptual colorspace on regular home TVs, our eyes can make out between 300 and 800 different grays, depending on the brightness of the display and the black level. This means that the green channel has to have at least this many levels, and the other colors a bit less. 220 on a gamma colorspace doesn't cut it.

Having said that, 220 bits-per-plane is acceptable if

1) All subsequent processing (i.e. the TV, or communication between processor and tv if the processor does some real work like contrast and color adjustments) is done and kept in a high-bit-depth form.
2) Quantization in the compression process hasn't reduced the number of displayed colors. This usually shows up as blocking rather than banding, though.
3) The mastering process hasn't worked in a low-bit-depth colorspace, further reducing the colors.

Would high bit depth movies look more "real"? Perhaps. But film grain tends to mediate the problems with low bit-depth. Until the film grain is eliminated, I think it will always have that film look, and be a bit different from reality.

Bad banding is usually a result of bad film mastering, poor processing in the display, or incorrect settings like too high a contrast on the display.

It's not so much the "16.7 mil colors" thing, rather the number of possible shades of a given color that can yield visible banding. For a 24-bit color system, that comes out to 256 shades for a single basic color. For a "complex" color that is a mixture of the basic colors, the effective possible shades of that color may be even less than 256. Add in lossy data compression, and it even becomes a subset of that 256 shades of a needed color. At that point, it is not impossible at all to not have enough colors/shades to render a smooth gradient.

Maybe for masters, the extra data might be worthwhile. But let's remember that all we'll ever see at home is compressed video. starting out with a higher data rate, then having to compress even more to fit into the available channel may be counterproductive. The end result may improve by reducing the original data by lowering resolution, color res, or color depth, then having to do LESS compression to fit the channel available.

Dave

It's not so much the "16.7 mil colors" thing, rather the number of possible shades of a given color that can yield visible banding. For a 24-bit color system, that comes out to 256 shades for a single basic color. For a "complex" color that is a mixture of the basic colors, the effective possible shades of that color may be even less than 256. Add in lossy data compression, and it even becomes a subset of that 256 shades of a needed color. At that point, it is not impossible at all to not have enough colors/shades to render a smooth gradient.

Bingo.

This is why 10bit and 12bit panels are already being used in black & white medical imaging. It is an extreme example but a grayscale lcd panel simply has one "color" channel. 8 bit provides a mere 256 shades. 10bit moves it to 1024 and 12bit to 4096.

Movies that either for artistic reasons or cinematography limitations have a limited color palette will get the largest "wow" effect. Most movies should get a boost in scenes with limited colors like low light shots.

Well, I didn't mean to make a direct proposition for higher bit depth. ;) Imo, the bit depth strategy is a rather wasteful way to achieve finer shades of a needed color. As was noted earlier by DaveKennett, more data in the master seems beneficial, but that only worsens the compression issue on the consumer end (which seems to be the more pressing bottleneck, at this point). I agree with him that the consumer end needs to facilitate free breathing bitrates and lesser compression, rather than more compression, where possible, if we are to enjoy the finer points of these masters, in the first place.

On the issue of color depth, I think a better system would be some sort of adaptive color coding. Right now, it is a linear system that represents as many colors as can be encoded in a 24-bit number system. It's 16.7 mil colors, which is a whole lot of color choices, but realistically, we only need a fraction of those colors to represent any one scene. So we have a burden of always having to represent this huge number of colors, when we only need to use some subset of these colors. This is a waste of data, which could have been more useful if it went to encoding finer shades of the colors that are used in a scene. Employing a deeper color depth does get us finer shades, but also at the expense of exponentially more data waste (representing billions more of colors that we end up having no use for, at all).

The current system offers 16.7 mil "slots", which translates thereabouts of 65,000 unique colors, with 256 shades for each color. An "adaptive" system would realize that a scene or group of scenes only needs 10,000 of those unique colors, and then use the slots for the unused colors to enable 1670 shades for each of those unique colors (vs. only 256 shades). That would ruthlessly clobber any conceivable visible gradient, in my estimation. The adaptive system would intelligently optimize the working pallete for the scene(s) to provide the necessary colors and any possible shade of those colors.

Naturally, that is a whole lot more complex. It would also require a whole new generation of equipment to support it. It's not even a new idea (variations of the theme have been used to represent useable colors in videogames using ridiculously small data footprint to achieve it). It would utterly nuke the number of shades of a color issue, w/o increasing the data needs beyond the handling of 24-bit color we already do now. It may even allow us to move below the needs of 24-bit color (which would in-turn help the compression issue).

Well, that's my 2 bits. :)

Thanks for all the great explanations guys, the lights are slowly starting to turn on in my head now :D

Mr Hanky, so AVC-1 and VC-1 don't have adaptive colour palettes? Is this due to the lack of processing power to decode them if they did? Haven't computer picture formats had selective colour palettes for a long time now?

On the issue of color depth, I think a better system would be some sort of adaptive color coding. Right now, it is a linear system that represents as many colors as can be encoded in a 24-bit number system. It's 16.7 mil colors, which is a whole lot of color choices, but realistically, we only need a fraction of those colors to represent any one scene. So we have a burden of always having to represent this huge number of colors, when we only need to use some subset of these colors. This is a waste of data, which could have been more useful if it went to encoding finer shades of the colors that are used in a scene. Employing a deeper color depth does get us finer shades, but also at the expense of exponentially more data waste (representing billions more of colors that we end up having no use for, at all).

The current system offers 16.7 mil "slots", which translates thereabouts of 65,000 unique colors, with 256 shades for each color. An "adaptive" system would realize that a scene or group of scenes only needs 10,000 of those unique colors, and then use the slots for the unused colors to enable 1670 shades for each of those unique colors (vs. only 256 shades). That would ruthlessly clobber any conceivable visible gradient, in my estimation. The adaptive system would intelligently optimize the working pallete for the scene(s) to provide the necessary colors and any possible shade of those colors.

Naturally, that is a whole lot more complex. It would also require a whole new generation of equipment to support it. It's not even a new idea (variations of the theme have been used to represent useable colors in videogames using ridiculously small data footprint to achieve it). It would utterly nuke the number of shades of a color issue, w/o increasing the data needs beyond the handling of 24-bit color we already do now. It may even allow us to move below the needs of 24-bit color (which would in-turn help the compression issue).

Well, that's my 2 bits. :)

I would think something like this could be addressed, if not already, in a video processor. It does make alot of sense. Then again, a good VP costs more than your TV.

Mr Hanky, so AVC-1 and VC-1 don't have adaptive colour palettes?

I don't know if they do or not. It does not appear to me that they are using them if they do, if we are discussing banding as even a remote possibility in a "modern" encoding.

Is this due to the lack of processing power to decode them if they did? Haven't computer picture formats had selective colour palettes for a long time now?

I don't think it is really a processing power problem, so much as it needs to be supported on both ends of the chain, or it won't work (or won't be particularly useful). Presumably, the determination of pallettes would be something that is performed offline, at the beginning of the process for a known piece of material. Obviously, encoding would not work (or it could be extremely fallible) in "realtime" or in a "live" format, since you can't know what colors are about to occur until they have actually happened and that event has been recorded. Naturally, initial capture and studio equipment should just stick with deeper color depths (so as to be prepared for "any scenario", by default).