Television reinvented: rgb +cmyk!!!!! (1 Viewer)

[Kevin Farley]

here's the link


They just succeded in adding cmyk to tv, making color much, much more film like.

But will it ever be implemented?

The company, Genoa Color Technologies

 

[MarkHastings]

That's what I don't get. CMYK is reflective color, while RGB is projected color. How do you get reflective color our of a tv set?

It almost sounds more like a gimmick than anything else.

 

[Max Leung]

I don't see the point of it, when all the material that is being broadcast or encoded on DVD are mastered to SMPTE standards, which set the RGB primaries to be used. You would need to make new DVDs or broadcast materials that can take advantage of the added colour gamut.

Unless this new technology is used to address the inability of TVs to accurately display the current RGB standard...that would be nice!

 

[Seth Paxton]

Well Max, in the article Ms Stone touches on your point precisely



She also touches on the other issue which is that even as it stands the conversion work to go from the 3 color to 5 color system on the fly has to be difficult in real time. They have shown demos, but that's not the same as practical neutral party testing. They could have pre-processed data or selected data that would process optimally, or even designed the processor/software to work optimally with the test data such that it would behave worse than average on other data.


Mark, I think you'll find that they really are just using the terminology of reflective color to represent the idea of a display system using more than 3 primaries. After all, sunlight isn't made up of 3 primaries and that's not reflective color.

 

[ChristopherDAC]

The curious thing about this is that I have read the original NTSC documentation and Zworykin's book on Television, which give a great amount of information on colorimetry and describe in detail how three emitting phosphors of the proper colours can together reproduce all practical colours. Two points were raised in the NTSC documents which seem relevant:

1) The dyes, &c. used for normal coloration [reflective] have a far more restricted range than the phosphors used for television, and so the use of a wide gamut of dyes makes much more sense than the use of a large number of phosphor primaries.
2) Since the triaxial diagramme fully describes the colour space of the human eye, the NTSC signal [and all later colour systems have been built on this approach] contains all the information necessary to reproduce full colour; in fact, the NTSC documents specifically state that "if it were later decided to use a greater number of primaries, the tricomponent colour signal supplies all the necessary information".

So what is the big deal? Met any colours which couldn't be replicated on a well made TV lately?

 

[Dave Corbitt]

Well, there are some fairly common real world colors that Rec 709 primaries cannot reproduce with any fidelity. Cyan is the worst offender. Deep Ruby Red is another. Emerald Green another. Real Cyan is a brilliant turqoise color and the closest that modern displays can come is a dullish pale sky blue. Typical brilliant Cyans in real world images would be the brilliant color of tropical waters, some toothpastes, and clear skies. To resolve this, a tri-color display with wide gamut RGB would get us there, particularly a Green that is further up on the CIE chart (closer to Emerald Green rather than the yellowish green of Rec 709) . In fact, original 1953 NTSC had wide gamut RGB phosphors but the technology back then could not produce bright enough images to be marketable so we eventually wound up with Rec 709, a smaller triangle of color gamut, but a workable one that does a decent job. Newer non-CRT displays are coming on line with wider gamut RGB primaries that approach the original NTSC ideal. You'll start seeing them in the marketplace this year. The problem now is convincing program producers and content providers to ante up and remaster film transfers and buy new cameras that are incormporating the new wider gamut colorimetry, not an easy task nor an economically practical model at all. Using a wide gamut display with a Rec 709 signal can produce some bizarre color distortions unless the drive electronics have a correction matrix inside to bend the color space back to the narrower Rec 709 characteristics.

The physiology of the eye is essentially that of a tri-color sensor. Stimulating those RGB sensitive cones with correct amounts of three properly chosen primaries is proven to be indistinguishable from an equivalent mixture of any broad or narrow spectrum real world "color". This is called a Metameric Approximation and is the basis of all color science used in film, television, and print. Using more primaries can effectively widen the gamut if the choice of primaries was not ideal originally. In print this a practical solution to solve color reproduction problems due to the limited reflective spectrum of real world dyes and inks. In an additive system (television) it is possible to choose wide gamut primaries that capture all real world colors and that was conclusion of the science of the original research done when NTSC was first being developed. The science still holds but the paractical implementation was never economically implemented until very recently.

 

[Max Leung]

NEC is about to release a 26" LCD monitor with 92% NTSC color gamut. Samsung has one coming too I believe. But these are "pro" monitors and quite expensive. I can't remember if the NEC uses an LED backlight or improved CCFL tubes though.

I guess if you want to exceed NTSC you can't do that yet. At least not for less than a few thousands.

I wonder what the color gamut of standard 35 mm film is? I assume it exceeds NTSC?

 

[Dave Corbitt]

There is a good tutorial on line posted by ARRI. ARRI makes motion picture cameras, motion picture film recorders, and has an extensive network of film labs and production equipment rental houses around the world. Here's the link:

http://www.arri.com/prod/digital/dig...nion/ch06.html

They have a discussion of color gamuts including film versus digital displays. Some of this relates to DI work (Digital Intermediate) which is how most motion pictures are now post produced. The camera original film is scanned in to a high resolution digital server (usually 4K resolution with 16 bit depth). The data is then projected digitally on a large screen in a color correction theater where a colorist adjusts the images according to the director's wishes. The colorist is viewing this image through a 3D LUT (look up table) to emulate what the data will look like once it is transferred back to film after the post production process is completed. CGI effects and editing are done while the film is in this data mode, thus the term Digital Intermediate.

Film color gamut is not a smooth triangle like video displays. The absorption characteristics of the dyes make for a strangely shaped gamut. In a nutshell, film color gamut is wider in darker areas particulary in Cyan areas, and good digital displays are wider in bright areas for Red and Blue. Yellows in video tend to not be as saturated as on Film. The ARRI website explains all this nicely.

Dave

 

[Dave Corbitt]

More. Regarding NTSC. NTSC gamut does not really exist in practical displays. It can be reproduced on expensive pro gear used in DI theaters and on the few old color TVs from 1954-59 that still had the original NTSC phosphors but those are rare indeed and hard to keep in opperating condition. Besides, who on this list would put up with watching low resolution composite video on a small screen just to see a wide gamut of color? The Original NTSC color as specified in 1953 was a very wide gamut system. It was implemented crudely back then with Image Orthicon cameras and complicated and unreliable tube tv sets with wide gamut phoshors that had all kinds of problems like image lag and smear, and very dim images. So when a monitor manufacturer publishes a spec saying they can reproduce 93% of NTSC, they are referring to the origninal 1953 wide gamut spec, not the current updated narrower spec of Rec 709.

Dave

 

[MarkHastings]

NTSC = Not The Same Color

 

[Dave Corbitt]

Mark, the acronym I remember was Never Twice the Same Color, referring to NTSC's tendency to have hue shifts caused by the analog phase modulation of the color subcarrier being distorted during transmission. Thank goodness it's going away finally.

Dave

 

[ChristopherDAC]

Of course, not only does the present DTV system not call for the proper NTSC color gamut, it makes no attempt to implement Constant Luminance. That was also the subject of intense discussion by the second NTSC, and it's worth noting that the Japanese MUSE analog HDTV system uses a quasi-constant luminance system, with significantly lower nonlinearity.

Anyway, NTSC works quite well for its purpose. It's not a perfect TV system, but this new one has its own defects.

 

[Max Leung]

Whoa thanks for the fountain of information Dave!

It seems most computer LCD monitors these days are tailored for sRGB colors. How close to Rec 709 is sRGB anyways? Well, I guess I'll have to dig around for that. I was very surprised how virtually none of the LCD monitors today are even close to (the original) NTSC color reproduction. Now I'm wondering if all these digital projectors and flat panel TVs are a big step backwards, in terms of color reproduction, compared to the CRT sets from 10 years ago!

 

[JeremyErwin]

according to this pdf, sRGB and Rec709 use the same color primaries, but different gammas.

It's really difficult to explain these differences, when most of us have sRGB monitors. The reds are redder, the greens are greener, and the blues, bluer. Or so they say.

Also note that the gamut of visible colors is a curved object-- the shape of the gamut depends on the number of primaries. sRGB, NTSC, SECAM, etc, are all triangles. cmyk systems are quadrilaterals. A six primary system would produce a hexagonal gamut.

 

[MarkHastings]

That's good too. Actually, your rendition reminded me that I actually heard it as "Never The Same Color" instead of "Not The Same Color", but I like "Never Twice...."

 

[JeremyErwin]

I guess atsc is "always the same color"

 

[Dave Corbitt]

Jeremy,

Thanks for that PDF. It clearly shows a lot of very useful info. To all, keep in mind that the large NTSC triangle on these diagrams is not what we now have. Rec 709 is essentially the same as the PAL/SECAM triangle and that is what we now have for both SD and HD displays in the USA.

The specified gamma for Rec 709 is 1.93 according to the Hoffman paper and 2.2 for sRGB but in actuality, most broadcast monitors are measured at around a gamma of 2.4. So in practice, a gamma closer to sRGB is what mastering is done on in post facilities and sRGB primaries are very close to Rec 709. For most video content that is carefully mastered, a good Rec 709 display but with a gamma of approximately 2.4 would reproduce the original mastering conditions well.

The way gamma works, the higher the number, the more compressed the lowlight brighness will be so a gamma of 2.4 will create darker lowlights and more saturated looking colors than a gamma of 1.9 which will look a little washed out and pale in the lowlights with less saturated color.

Got to go. Need to do some food shopping.

Dave

 

[MarkHastings]

That's precisely what I've talked about in the Mac vs. PC threads. A Mac's gamma is native to 1.8 whereas PC's are native to 2.2

2.2 is way too dark for trying to color correct anything. Starting off at 1.8 and then adjusting contrast levels (to make the darks a bit darker) is more effective than starting at 2.2 and trying to brighten.

 

[Dave Corbitt]

Mark,

1.8 gamma is really the odd man out in color space systems. 2.22 has been industry spec in broadcast for decades. All imaging devices (cameras, telecines, etc.) are designed to have the inverse of 2.22 or approximately 0.4545 gamma to compensate for the 2.22 gamma inherent in CRT displays. The net result is supposed to approach linear light output (2.22 x 0.4545 = 1.00899) to reproduce the original image as closely as possible. Modern broadcast monitors generally have a bit more gamma (around 2.4) these days and mastering colorists have adjusted their color correction work to make images look good on these contrasty monitors. Gamma is totally adjustable on color correction hardware. A color correction suite in a post production facility will usually have a DaVinci or Pogle color corrector with lots of handles to control image quality in real time. Films or shows on tape or on a server are run scene by scene and each sgment is carefully adjusted manually by a skilled colorist who enters the values in the system controller. Scenes are tagged by their frame numbers or time code. Once all the color values are set and agreed upon by the DP or director working with the colorist, the program is run back to the head and played back in real time with the color values coming in exactly on their marks. The resulting color corrected program is usually recorded on D5 tape or back to another server for editorial finishing (adding titles, credits, slate, color bars, etc.).

At no point in the professional chain of processing for video delivery is a 1.8 gamma used to view reference quality images. On an Apple, at least on mine, when setting gamma, the menus specifically suggest using 2.2 for video work, and 1.8 for other work (graphics, I assume, such as for print work?) I wonder why Apple uses 1.8, have you got some insight in to that?

Anyway, this has been an interesting thread.

BTW, for more info on color space, go here:

http://www.fho-emden.de/~hoffmann/

then click on Documents Comp. Vision

Dave

 

[MarkHastings]

I'm not 100% positive, but I believe Adobe (Photoshop) invented the 1.8 color space for Apple. This was to better insure color correcting capabilities from their monitors to the printer.

Yeah, it's definitely a print thing and not a video thing. But even though I do both, I prefer starting at the 1.8 and using the controls to darken my monitor. I just find that starting off at 2.2 is way too dark.

I even lightened up my PC laptop because when I set it to 2.2, I was having odd issues where I would adjust images in Photoshop and then when they went to DVD, they revealed stuff I never saw in Photoshop. But that too could be a misscommunication between the Photoshop color space and the desktop. I'm not 100% familiar with all of this to know the real cause.