isn't 1080i/60 better than 1080p/24 ?

Yup! But that doesn't mean it will look better.

Live TV is recorded at 30 frames per second.

But movie film is recorded at 24 frames per second.

To show a movie on a normal TV you have to raise 24fps to 30fps. That is done by duplicating interlaced half frames (called "fields") in a particular "cadence" -- usually 2,3,3,2. That means some fields are on screen a little longer than other fields. This leads to a slight jerkiness of motion termed "cadence judder".

You've been seeing cadence judder since you first ever watched a movie on a TV. You likely haven't noticed it because the brain is very good at "not seeing" it, even if it is pointed out to you. If you really want to look for it, look at how the credits roll up the screen at the end of any movie.

But if you had a "judder free" setup side by side with a normal TV you would EASILY see the cadence judder in the normal TV version of the movie.

What's a "judder free" setup? Well that's when you send the movie to the TV at its original 24 fps rate *AND* when the TV adjusts its "screen refresh rate" to be a multiple of 24 times per second (usually 48 fps or 72 fps) instead of a multiple of 30 fps (usually 60 fps).

Pretty soon there will be TVs out there which use a fixed 120 fps refresh rate which is cool because 120 is an even multiple of both 24 and of 30. [Such TVs still need to detect whether the incoming content is video-based or film-based as that determines how often they need to duplicate each incoming frame to raise the rate to 120fps correctly. But they don't need to change refresh rate, which means the switch between types of source content is faster and cleaner. NOTE: Some LCD's out there that boast 120fps refresh rate aren't really doing this.]

---------------------------------------------

But cadence judder is not the only issue.

Modern TVs put up a progressive (i.e., a "de-interlaced") image. But de-interlacing is not as straightforward as you might think.

First of all, frames of film are recorded "all at once". That is, if you break them down into two, interlaced fields, both fields have been recorded at the same instant of time.

But live video is actually recorded "interlaced". That is, the two fields that make up a frame are not actually portions of the same image. The second field is recorded slightly later in time than the first field. And thus the two fields actually represent a slightly motion-blurred "double exposure" of the image.

Second, the "resolution" of the imagery is not REALLY 480p or 720p or 1080p. A trick is done to cut down the amount of data that makes up each frame. That trick depends on the fact that the human eye can't resolve color detail as well as it can resolve gray scale detail.

And so the frame is ACTUALLY made up of gray scale data at the full, specified resolution, "colored" by color detail that is only recorded at half that resolution in the horizontal direction.

And that means that de-interlacing has to account both for the difference between film and video style recording AND for the lack of information as to what color should be applied to a given pixel.

When you feed a 1080i/60 video stream to your TV, it has no way of knowing whether that is a "real" live TV program or whether it is a movie which has been restructured for viewing on a TV screen (that repeat cadence I mentioned above). So the TV has to figure this out "on the fly". That's doable (HINT: The TV looks for the "repeat cadence"), but it can be screwed up by edits in the video stream, etc., and such screw ups result in the brief appearance of artifacts on the screen.

However, if you take a 1080p/24 movie off a Blu-Ray disc, and send it as 1080p/24 to the display, there is no de-interlacing that needs to be done! And of course the TV knows that the content coming in is from a movie source since it is NEVER a good idea to send a "live video" style stream at 24fps.

So even if the TV does *NOT* adjust its refresh rate to a multiple of 24, it can, itself (i.e., inside the TV) convert that 1080p/24 video input to 1080p/60 or 720p/60 to light up the screen, and it can do that "perfectly". There will still be "cadence judder" since 24 is not an even multiple of 30 or 60, but there won't be any de-interlace related artifacting. At all.

[NOTE: If you think that horizontal color resolution loss is a bit of a downer, it may depress you even more to learn that the data actually stored on the disc, be it a standard DVD or a Blu-Ray disc, is actually even worse. The data stored on the disc ALSO cuts the VERTICAL color resolution in half. Your TV never sees that because every player "reconstitutes" the vertical color resolution on the fly as part of decoding the data off the disc. Think of this as another type of "scaling". Players that screw up this essential step have the now-infamous "Color Upsampling Error" or CUE.]

The bottom line is that if your AVR or TV will accept 1080p/24 as a valid input resolution and frame rate, then that's what you want to send to it while watching Blu-Ray movies.

When watching "live concerts" and such on Blu-Ray discs -- content that is recorded at 1080i/60 on the disc -- then you want to send 1080i/60 or 1080p/60 to your AVR or TV. Whether you send 1080p/60 depends on how much faith you have in the de-interlacing in your player compared to the de-interlacing in your AVR or TV.
--Bob

Best example of cadence judder is watching a car moving where the wheels appear to be going backwards.

Nope! "Deep Color" also uses the "trick" I mentioned.

A pixel is represented by 3 "component" data values: The gray scale brightness, or luminance, denoted as "Y" and two "color difference" values that tell how to color that gray scale, denoted as "Cb" and "Cr" for blue and red respectively. The Cb and Cr values denote adding or subtracting blue and red from the gray. So for example, if you subtract all the blue and all the red from a pixel of a given, non-black luminance or "Y" value, you get a GREEN pixel of that brightness.

The simplest "data format" that works this way is called YCbCr 4:4:4. The "4:4:4" in that means what you had in mind -- i.e., there is a Y value, a Cb value, and a Cr value for each and every pixel.

What gets transmitted in broadcasts on the other hand is YCbCr 4:2:2. The "4:2:2" means that there are only half as many color samples across the line as luminance samples. Like this: Y,Cb,Y,Cr,Y,Cb,Y,Cr.... Analog color TV broadcasts do a similar trick, and have done so since the dawn of "compatible" color TV broadcasting.

When you want to record stuff on a disc however, even that's not enough! The imagery on the disc needs to be "compressed" WAY more to take up less space on the disc, and even more important to require a less demanding bit transfer rate when reading the disc. That compression is made up of two styles of things. First there is micro-compression which reduces the information content in each frame of imagery and then there is macro-compression which uses the similarities from frame to frame to reduce things even more.

The most important micro-compression is that the YCbCr 4:2:2 data is cut down even further to YCbCr 4:2:0. This is the step where the vertical color resolution is also halved. The people setting up the disc transfer may also apply softening filters that reduce detail in the imagery so that adjacent frames are more similar -- which makes the macro compression step more effective.

The result of that is sent to the macro-compression encoder: MPEG2, MPEG4, AVC, or whatever. It discovers similarities from frame to frame and encodes things by recording blocks of similar stuff plus how they move from frame to frame.

When reading the disc, the player decodes the macro-encoded stuff to get YCbCr 4:2:0 and invents the missing vertical color information to turn that into YCbCr 4:2:2. Depending on the output format you have set for the player, it may also go the extra step and invent the missing horizontal color information to turn THAT into YCbCr 4:4:4. By "invent" here, I mean the player extrapolates the "missing" color information from the "present" color information surrounding it.

------------------------------------

OK, now one thing I haven't mentioned here is the SIZE of each of these pixel component values.

"Normal" data format for home theater stuff uses 8 bits each for Y, for Cb, and for Cr. When jiggered up to YCbCr 4:4:4 data format, that amounts to 24 bits per pixel -- 3 components per pixel of 8 bits each.

But of course the data coming off the disc is not that. It is YCbCr 4:2:0 -- and yes, with only 8 bits each per component. Since the horizontal and vertical color resolutions are both halved the "real" bit count per pixel is quite a bit less than 24 bits.

This works as well as it does because, as stated above, the human eye just can't distinguish color detail all that well, and because the brain does a fantastic job of "seeing" smoothly changing imagery from a series of pictures each of which may not be all that great looking on its own.

------------------------------------------

"Deep Color" on the other hand, means that the Y, Cb, and Cr components are processed and transmitted at MORE than 8 bits per component. Common supported values would be 10 or 12 bits per component. HDMI V1.3 allows even higher but no hardware does that yet.

If you had a YCbCr 4:4:4 data stream at 12 bits per component that would add up to 36 bits per pixel! Pretty cool, right? Until you remember that the data actually coming off the disc (be it standard DVD, HD-DVD, or Blu-Ray) is really only YCbCr 4:2:0 at 8 bits per component.

The extra bits per pixel are created by the magic of mathematics -- built around a series of assumptions as to what the result ought to look like. You could, for example, decree that nearby pixels must always represent a smooth transition of the colors between them. That means you could extrapolate from the colors you have for certain nearby pixels to the colors of the pixels in between because you know it must be a smooth transition. [If your assumptions are wrong, then of course the result will also be wrong.]

You could do this with 8 bits per component math (pretty much obsolete these days) or you could do it at higher precision with 10 or 12 (or higher) bits per component math (pretty much what all modern video processors and TVs do internally for their own, individual video processing steps).

But wait! If you want to transmit that stream to the next device in your "video chain" you are stuck! "Normal" color HDMI only lets you retain 8 bits per component. So even if you send YCbCr 4:4:4 you can only retain 24 bits per pixel!

Well the optional "Deep Color" feature in HDMI V1.3 allows you to break free of that limit. You can send 30, or 36 (or even more, although nobody does it) bits per pixel!

So this is a "good thing". But don't confuse that with seeing something that you've been missing up to now. Because the data on the disc is still YCbCr 4:2:0 at 8 bits per component.

And that's not going to change for standard DVDs or HD-DVD or Blu-Ray discs. Not ever. It is part of the disc spec. If you tried to make a disc with higher bit depth, no player would know what to do with it. At some point (think years from now) there may be a *NEW* disc format for recording Deep Color content, but that won't be Blu-Ray. It will be something else. Time to buy a new player for one thing.

There is a lot of speculation at the moment as to whether passing these extra "rounding bits" between devices will make a real, quantitative improvement in what people see on screen. It is also really REALLY hard to evaluate this stuff in real world products because each new product also comes with OTHER changes that are designed to enhance its video. So ascribing any improvement you might see entirely to Deep Color is likely incorrect.

But we are just now (this month) seeing the first devices that are trying to stake out a marketing position for this -- the Pioneer 51 and 05 Blu-Ray players -- so this forum should be full of opinions pro and con within the next few months.
--Bob

4:2:2 YCbCr has always had 12 bits per component on HDMI, even before "Deep Color" was invented. The purpose of "Deep Color" is to allow 4:4:4 YCbCr and 4:4:4 RGB to have more than 8 bits per component on HDMI.

Change that to "up to" 12 bits. YCbCr 4:2:2 over "normal color" HDMI could also be 8, or 10 bits per component.

But that said, you are correct. Many people don't realize they ALREADY have displays which can receive YCbCr 4:2:2 data format at 12 bits per component (24 bits per pixel). Of course not all sources will offer that for output.
--Bob

If color is being decimated by a factor of 4, what's the point to adding more colors with DeepColor in the future?

Makes more sense to me to first store the video as 4:4:4 before worrying about adding all these micro gradations in color. Heck, it makes more sense to me to first ship TVs that are correctly factory calibrated before worrying about a few extra colors. I can't even balance the color correctly on my relatively new Panny Plasma - green seems to have some issues.

DeepColor Seems silly. Thoughts?

xvYCC seems more useful as it actually extends the color space, unless of course it only serves to add colors we can't see.

Actually, 4:2:2 YCbCr is always transmitted at 12 bits per component, according to the official HDMI specification. However, most source devices just put zeros in the lowest 4 bits, and most display devices just ignore the lowest 4 bits. In other words, most devices only take advantage of the highest 8 bits, even though all 12 bits are always being transmitted.

DC is silly in some ways. For example, if a display is advertised as supporting DC, the only DC format that it is required to support is 36-bit 4:4:4 RGB, which has 12 bits per channel. It is not required to support any other DC format. In addition, the display may ignore the lowest 4 bits, in which case the display only uses the highest 8 bits.

Deep color is a marketing gimmick.

Leaving the 1080p/24 film aside, how does the player or display recognise the difference between 1080p/30 video encoded as 1080i/60 and native 1080i/60 video?

+1.

Until new technology comes out, current HDTV TV technoligies, sans CRT, barely can display all the colors in the 'normal color' signal, not to mention 'deep color'.

The original reason for going i/60 instead of p/30 was a combo of the expense of capturing the p/30 image and the fact that the eye can see some "refresh flicker" of contrasty scenes at /30 that gets concealed at /60 (even though you are only actually seeing half of the interlaced image at once). Refresh flicker is the perception that the screen is getting light and dark over an over again -- i.e., you see it blinking.

For the same reason, movie projectors have a shutter-like mechanism in them that blocks and reveals each frame twice -- thus making the flicker rate be 48 fps instead of 24 fps. Even though the imagery is only "moving" at 24fps, the screen blinks at 48 fps and the brain blurs that out.

Modern fixed pixel displays actually have more persistence of light output so this isn't really necessary, but not all TVs out there are made that way so /60 (or /50 in Europe) remains the norm. But 1080p/60 setups are more costly, so 1080i/60 is the way it is done.

As it turns out 720p/60 and 1080i/60 are almost exactly the same bandwidth. [When you do the math, be sure to remember that a 720p image also has less horizontal resolution.] And that's why 720p/60 and 1080i/60 are paired as the two standard HDTV resolutions. They take up the same amount of channel bandwidth licensed to an HDTV station.
--Bob