Jump to content


Photo

Red One's Color Space/Gamut?


  • Please log in to reply
10 replies to this topic

#1 Peter Moretti

Peter Moretti
  • Sustaining Members
  • 306 posts
  • Other
  • Sherman Oaks, CA

Posted 23 September 2008 - 03:08 AM

HD cameras record to the Rec 709 colorspace (which uses the sRGB color gamut). But what colorspace(s) and gamut(s) can the Red One use?

Thanks much ;).
  • 0

#2 Matthew Rogers

Matthew Rogers
  • Basic Members
  • PipPipPip
  • 174 posts
  • Cinematographer
  • Knoxville, TN

Posted 23 September 2008 - 07:48 PM

Red doesen't really shoot in a color space like REC709. It shoots in RAW. However, you can monitor in a space like rec709. Does that make sense.

Matthew
  • 0

#3 Peter Moretti

Peter Moretti
  • Sustaining Members
  • 306 posts
  • Other
  • Sherman Oaks, CA

Posted 23 September 2008 - 08:50 PM

I don't see why Red's RAW wouldn't have a specific color gamut. All mediums have a breadth of sensitivity to color. Film of course has a color gamut, which is why a film LUT is used to mimick or stay within that gamut.

I would imagine that there are colors the Red does not see (probably in the blue spectrum) or chooses not to see (probably in the infrared spectrum).

Edited by Peter Moretti, 23 September 2008 - 08:52 PM.

  • 0

#4 Brian Drysdale

Brian Drysdale
  • Basic Members
  • PipPipPipPip
  • 3914 posts
  • Cinematographer

Posted 24 September 2008 - 04:54 AM

There's a discussion on CML about the RED's spectral sensitivity at the moment.
  • 0

#5 Peter Moretti

Peter Moretti
  • Sustaining Members
  • 306 posts
  • Other
  • Sherman Oaks, CA

Posted 24 September 2008 - 05:29 AM

CML?

Please pardon my ignorance, LOL.
  • 0

#6 Brian Drysdale

Brian Drysdale
  • Basic Members
  • PipPipPipPip
  • 3914 posts
  • Cinematographer

Posted 24 September 2008 - 06:18 AM

CML?

Please pardon my ignorance, LOL.


Cinematography Mailing List

http://www.cinematography.net/
  • 0

#7 Keith Walters

Keith Walters
  • Sustaining Members
  • 2024 posts
  • Other
  • Sydney Australia

Posted 24 September 2008 - 06:25 AM

HD cameras record to the Rec 709 colorspace (which uses the sRGB color gamut). But what colorspace(s) and gamut(s) can the Red One use?

Thanks much ;).

More correctly, the HD camera takes the available signals from the Red, Green and Blue pickup devices, (whether that means separate CMOS or CCD sensors or camera tubes or a single sensor marked off into Red, Green and Blue squares), and does its best to fit them into the 709 colorspace (or whatever you are trying to intercut with).

Essentially that means that with any particular wavelength of light, with a Rec 709 compliant device you should get a particular ratio of Red, Green and Blue values being recorded onto the record medium.

In theory at least, any camera can have a conversion table generated that will allow its output to be converted to any colourspace you like.

Unfortunately the are many factors complicating this. One is that silicon does not have an even response to all wavelengths of light. Its response is quite poor in the blue region, and strongest in the infra-red region. This is a major problem with single-chip sensors because it is very difficult to filter out incident infra-red without cutting into the visible red response.

Another problem is that the dyes used for Bayer Mask cameras not only have to produce the correct spectral response, they will be expected to continue to do so after years of exposure to light, with no significant fading or discoloration. Dyes which combine both properties are very difficult to produce and so they will always be a compromise.

Still, adjusting a camera's response to match a particular colorspace in theory is just a matter of automatically tweaking the Red, Green and Blue levels to produce the correct colour. Perhaps the most obvious approach would be a simple "lookup table" where the camera would would be successively "shown" all the colours possible in the Chromaticity diagram, its output would be recorded for each step of its entire dynamic range, the three necessary correction factors required to convert the actual values to the correct values for each possible colour would be calculated and they would all be stored in a computer memory.

The only problem would be that if you had for example a 10-bit camera, to cover its entire dynamic range, you would need a 1024 x 1024 x 1024 x 3 or 3 gigabyte lookup table. 12 bit ADCs would require 24 gigabytes, 14 bits would need 192 gigabytes!

Not such a big ask, you are still thinking? If you are thinking of Flash RAM, that would be far too slow. Take the RED with active 4096x2304 pixels. At 24 fps you would have to look up 226,492,416 values per second. There is simply no sort of memory available that is both that big and can also can be totally random accessed at anything like that speed.

Also your calibration process would be somewhat problematic. Suppose you had some sort of prism device that could progressively expose the camera to all possible colours in the desired gamut, at the 16,384 different brightness levels possible with 14-bit ADCs. Assuming you can only realistically calibrate the camera at 24 frames per second, (and you can make your colour generator work that fast) that means it would take you 192,000,000,000/24 seconds or about 253 years!

So obviously manufacturers have to take some pretty severe shortcuts and this is one reason why lookup tables can only ever be an approximation at present. They take a much smaller range of gamut samples and interpolate between them.

Another problem is that in marginal exposure situations cranking up the weaker colours simply introduces noise. This is also one reason why you should always use filters for "coarse" colour balancing and only use the camera's auto white balance for "fine tuning". I can't believe that we're onto another generation of chuckleheads who still think you can "fix it in post".

Sorry to be so long-winded, but the basic answer is that the RED's RAW output can be tweaked to approximate any colorspace that anybody feels like generating a LUT for. However if it doesn't have the dynamic range of the device you want it to emulate, it will only ever be an approximation.
  • 0

#8 Peter Moretti

Peter Moretti
  • Sustaining Members
  • 306 posts
  • Other
  • Sherman Oaks, CA

Posted 24 September 2008 - 07:33 AM

Thank you for your reply, and it totally makes sense.

It seems like with any camera there are two ?'s in terms of color gamut.

1) Does the sensor pickup the color in question at all?

2) If so, does the sensor accurately register the color?

I imagine the answer to 1 is mostly yes, but not always.

I imagine the answer to 2 depends mostly on the color. In the case of blue/violet, probably not, in which case a lot of manipulation takes place to recast what the sensor reads to what the color actually is.

It would be pretty neat to see a camera's recording of an entire visible spectrum chipchart/gradient. For an HDV camera, would the tape be black (except maybe for superwhites and superblacks) for colors outside of Rec 709's gamut?

Would the Red pickup everything?
  • 0

#9 Keith Walters

Keith Walters
  • Sustaining Members
  • 2024 posts
  • Other
  • Sydney Australia

Posted 25 September 2008 - 06:26 AM

Thank you for your reply, and it totally makes sense.

It seems like with any camera there are two ?'s in terms of color gamut.

1) Does the sensor pickup the color in question at all?

2) If so, does the sensor accurately register the color?

I imagine the answer to 1 is mostly yes, but not always.

I imagine the answer to 2 depends mostly on the color. In the case of blue/violet, probably not, in which case a lot of manipulation takes place to recast what the sensor reads to what the color actually is.


You're right; like many things, it depends. If you're trying to image a colour that's right on the edge of what the sensor can pick up (eg deep blue or violet) and there is plenty of ambient light available, you may be able to get a useful response from the blue sensor, that can be massaged into your desired colourspace. But where there is less light available, the blue response may be mostly noise. If you try to "turn up the volume" on that all you will do is make the noise much more visible.

This can happen even when there is more than enough light to image the bulk of the picture. A good processing algorithm will have enough smarts to know when that colour should be imaged, and when it would be better left out of the equation. Many cameras show lower level blue and purple shades as grey for example.


It would be pretty neat to see a camera's recording of an entire visible spectrum chipchart/gradient. For an HDV camera, would the tape be black (except maybe for superwhites and superblacks) for colors outside of Rec 709's gamut?


Not black. More likely to be either monochrome or some colour closer to the centre of the gamut. It's actually very difficult to artificially produce the full gamut.

Would the Red pickup everything?


I doubt it. As far as I am aware, the UV-hardened dye polymers used for single-chip sensors are only produced by a few specialist companies, and the exact composition is a closely guarded trade secret, much the same way the "recipe" for dye couplers is for colour film.

The upshot of this is that the same dyes are used for things like the RED, D-20 or Genesis, as are used for sub-$50 still cameras. The amount of polymer required per sensor is pretty minute, so there is little scope for cost savings by using anything less than the best they can produce.

I have certainly seen stunning colour images from really cheap cameras, the main deficiency being in the cheap lenses they use. So on that basis I seriously doubt the RED's raw data gathering capability will be any better, it would just have better resolution and dynamic range (maybe). Plus the RED can take advantage of vastly better lenses, which will make far more of a difference than many people seem to realize. (Although the extra contrast they produce may also tend to show up any deficiencies in dynamic range).
  • 0

#10 Jay Bauer

Jay Bauer

    New

  • Basic Members
  • Pip
  • 3 posts
  • 1st Assistant Camera
  • los angeles

Posted 02 December 2008 - 02:30 PM

As I can see all these posts are a bit old... But i recently found this upcoming seminar with the RedOne and this company, Gamma & Density?? Anyone heard of this? this link shows the seminar info.. http://3cp.gammadens...dex-4-News.html. From what I can tell I think what they plan to offer will be a strong tool for REDONE color correction/workflow solution.. Let me know if you guys have heard anything on this...thanks!
  • 0

#11 Kevin Pontuti

Kevin Pontuti

    New

  • Basic Members
  • Pip
  • 1 posts
  • Digital Image Technician

Posted 29 December 2008 - 10:33 PM

As I can see all these posts are a bit old... But i recently found this upcoming seminar with the RedOne and this company, Gamma & Density?? Anyone heard of this? this link shows the seminar info.. http://3cp.gammadens...dex-4-News.html. From what I can tell I think what they plan to offer will be a strong tool for REDONE color correction/workflow solution.. Let me know if you guys have heard anything on this...thanks!


Hi Jay,

I'm a color management consultant (and new to this list.) and I saw a demo of their software today which was really exciting. I come more from a stills photography background and have been dealing with profiling scanners, cameras and various output devices. It's pretty exciting to see what Yuri Newman and his partner have put together especially for DP's and everyone in the color pipeline. The people at Gamma and Density really know their color and have some really good ideas for workflow so I think it will be a really good class.

Kevin Pontuti
  • 0




Glidecam

Abel Cine

System Associates

Paralinx LLC

Rig Wheels Passport

Visual Products

K5600 Lighting

Pro 8mm

NIBL

rebotnix Technologies

Aerial Filmworks

Cinelicious

Ritter Battery

Zylight

Cadrage Directors Viewfinder

Lemo Connectors

The Slider

CineLab

CineTape

Robert Starling

System Associates

Aerial Filmworks

K5600 Lighting

Cadrage Directors Viewfinder

NIBL

Glidecam

Lemo Connectors

CineLab

Pro 8mm

Ritter Battery

Abel Cine

Rig Wheels Passport

CineTape

The Slider

rebotnix Technologies

Paralinx LLC

Cinelicious

Robert Starling

Visual Products

Zylight