Jump to content




Photo

What is the parameter to quantize color Metamerism?


  • Please log in to reply
11 replies to this topic

#1 Kelvin Xu

Kelvin Xu
  • Basic Members
  • PipPip
  • 17 posts
  • Colorist
  • Singapore

Posted 25 August 2015 - 12:25 AM

I am aware of the metamerism which means two light sources of different spectrum may cause a subject to exhibit the same color. What is the parameter to differentiate the two light sources? 

 

For an LED and a florescent lamp to have different spectrum but of same 6500K CCT, how can one declare which is closer/accurate to CIE D65? 

 

What are the typical materials used to make filter gels? Can one mix or stack layers of such materials in layers of plastic films to convert a 6500K LED closer to CIE D65? 

 

Thank you. 

 

 


  • 0




#2 Mark Dunn

Mark Dunn
  • Basic Members
  • PipPipPipPip
  • 2182 posts
  • Other
  • London

Posted 25 August 2015 - 04:20 AM

I don't know the answer to your question, but you were given a good answer yesterday.

LEDs, in common with all non-incandescent light sources, are deficient in certain wavelengths, and no filter can put them back. Only incandescent lamps have a continuous spectrum which can be fully corrected by filters.


Edited by Mark Dunn, 25 August 2015 - 04:21 AM.

  • 0

#3 Stuart Allman

Stuart Allman
  • Basic Members
  • PipPipPip
  • 149 posts
  • Cinematographer
  • San Diego, CA

Posted 25 August 2015 - 09:40 AM

Kelvin,

 

I'm not a color scientist, but my (paying) job has me dealing with color science issues.  Standard illuminant D65 is defined by CIE xyz parameters.  Google "Wikipedia D65 standard illuminant" and you'll find exactly what you're looking for, including information about metamerism.

 

Keep in mind that D65 is not necessarily the same as 6500K.  In fact D65 is closest to 6504K.  6500K is defined with a specific power spectrum emanating from a black body radiator, whereas D65 can be emulated with any source that produces a similar spectral power distribution.  So it's possible for an LED or fluorescent lamp to produce D65, but they would be hard pressed to produce true 6500K in the visible spectrum.  Most people don't understand the difference so vendors just use 3200K or 5600K to describe their light output, even though the lamp doesn't technically doesn't match the black body radiator spectrum.  That's where the CCT - correlated color temperature - comes in; because to our eyes (not necessarily a camera) the lamp and the black body radiator appear similar...but still has problems with metamerism.

 

Stuart


  • 0

#4 John E Clark

John E Clark
  • Basic Members
  • PipPipPipPip
  • 780 posts
  • Other
  • San Diego

Posted 25 August 2015 - 11:58 AM

I am aware of the metamerism which means two light sources of different spectrum may cause a subject to exhibit the same color. What is the parameter to differentiate the two light sources? 

 

For an LED and a florescent lamp to have different spectrum but of same 6500K CCT, how can one declare which is closer/accurate to CIE D65? 

 

What are the typical materials used to make filter gels? Can one mix or stack layers of such materials in layers of plastic films to convert a 6500K LED closer to CIE D65? 

 

Thank you. 

 

 

 

For me 'metamerism' has been a problem with printing and subsequent viewing of prints under different lights. It relates to the mapping of a continuous spectrum to the 3 (for some 4) color bands of the human vision system.

 

Here's a article about color meters, and the writer does mention metamerism... very briefly... The wiki on the subject indicates there is a 'metamerism' index, but I don't know that any 'ordinary' photo meter has any indication for this parameter.

 

Lightmeter comparisons on measuring the 'color' of light.

 

http://www.dvinfo.ne...nt-thereof.html

 

Wiki on the topic of Metamerism

 

https://en.wikipedia...amerism_(color)


  • 0

#5 Phil Rhodes

Phil Rhodes
  • Sustaining Members
  • 11223 posts
  • Other

Posted 25 August 2015 - 02:20 PM

A brief treatise on the subject can be found here.

 

I didn't write the headline.

 

P


  • 0

#6 Guy Holt

Guy Holt
  • Basic Members
  • PipPipPipPip
  • 511 posts
  • Gaffer
  • Boston

Posted 25 August 2015 - 03:27 PM

Can one mix or stack layers of such materials in layers of plastic films to convert a 6500K LED closer to CIE D65? 

 

 

Wouldn't the question be rather can one mix or stack layers of gels to convert CIE D65 LED  to 6500K? The short answer is NO.  The long answer can be found  here.

 

Guy Holt, Gaffer

ScreenLight & Grip

Lighting Rental and Sales in Boston


Edited by Guy Holt, 25 August 2015 - 03:27 PM.

  • 0

#7 Kelvin Xu

Kelvin Xu
  • Basic Members
  • PipPip
  • 17 posts
  • Colorist
  • Singapore

Posted 25 August 2015 - 09:23 PM

A brief treatise on the subject can be found here.

 

I didn't write the headline.

 

P

 

Thank you Phil. That explains metamerism clearly.

 

Just to check whether my observation/monitor is correct, in these two images, the right samples appear slightly more saturated than the left? (This question has nothing to do with the explanation)

http://cdn.redsharkn...merism_fig1.png

http://cdn.redsharkn...merism_fig2.png


  • 0

#8 Kelvin Xu

Kelvin Xu
  • Basic Members
  • PipPip
  • 17 posts
  • Colorist
  • Singapore

Posted 25 August 2015 - 09:29 PM

 

Wouldn't the question be rather can one mix or stack layers of gels to convert CIE D65 LED  to 6500K? The short answer is NO.  The long answer can be found  here.

 

Guy Holt, Gaffer

ScreenLight & Grip

Lighting Rental and Sales in Boston

 

In sound synthesis we generate sound by combining multiple amplitude , frequency and phase. 

If we carefully select a number of single color LEDs, each with controllable brightness, then put through a light pipe to bounce/mix the light and a diffuser at the exit. Would that work? 

 

Are there linear light guide which could bounce and distribute light from a single point source to a line? 


Edited by Kelvin Xu, 25 August 2015 - 09:30 PM.

  • 0

#9 John E Clark

John E Clark
  • Basic Members
  • PipPipPipPip
  • 780 posts
  • Other
  • San Diego

Posted 25 August 2015 - 10:59 PM

 

In sound synthesis we generate sound by combining multiple amplitude , frequency and phase. 

If we carefully select a number of single color LEDs, each with controllable brightness, then put through a light pipe to bounce/mix the light and a diffuser at the exit. Would that work? 

 

Are there linear light guide which could bounce and distribute light from a single point source to a line? 

 

Unfortunately, LEDs and other 'spiky' sources would continue to give rise to spiky spectra, which is the problem when compared to continuous spectra such as Sunlight and Tungsten.

 

Optical filters, analogous to audio filters, such as 'band pass' or 'notch' are produced, but they are not available in the same way that one deals with audio filters, or audio DSP processing in the digital domain to give effects. If one thinks about how one 'tunes' a room for speakers, one hears of 'bass traps' or other physical elements, that are used, and 'custom' configured for the particular setup one has. Where most analog audio filtering happens, is in the electronic domain once the sound pressure has been converted to electronic pulses.

 

In the case of sensors if one had a sensor that had 6 or 12 bands of 'color' sensor, like a color that is  R, G, B that one has today, into 12 separate colors, then perhaps by processing the 6 colors with some sort of 'weighting' one could mitigate the spikes in some way in the the electronic domain.

 

 

In any case, by the time one puts on enough filters to knock down several spikes, one probably has a very poor output and not worth the effort.


Edited by John E Clark, 25 August 2015 - 11:03 PM.

  • 0

#10 Kelvin Xu

Kelvin Xu
  • Basic Members
  • PipPip
  • 17 posts
  • Colorist
  • Singapore

Posted 26 August 2015 - 12:35 AM

 

Unfortunately, LEDs and other 'spiky' sources would continue to give rise to spiky spectra, which is the problem when compared to continuous spectra such as Sunlight and Tungsten.

 

Optical filters, analogous to audio filters, such as 'band pass' or 'notch' are produced, but they are not available in the same way that one deals with audio filters, or audio DSP processing in the digital domain to give effects. If one thinks about how one 'tunes' a room for speakers, one hears of 'bass traps' or other physical elements, that are used, and 'custom' configured for the particular setup one has. Where most analog audio filtering happens, is in the electronic domain once the sound pressure has been converted to electronic pulses.

 

In the case of sensors if one had a sensor that had 6 or 12 bands of 'color' sensor, like a color that is  R, G, B that one has today, into 12 separate colors, then perhaps by processing the 6 colors with some sort of 'weighting' one could mitigate the spikes in some way in the the electronic domain.

 

 

In any case, by the time one puts on enough filters to knock down several spikes, one probably has a very poor output and not worth the effort.

 

Thank you John for the explanation. 


  • 0

#11 Stuart Allman

Stuart Allman
  • Basic Members
  • PipPipPip
  • 149 posts
  • Cinematographer
  • San Diego, CA

Posted 26 August 2015 - 07:44 AM

Kelvin,

 

Gels are very lossy, so you would lose significant light output trying to filter an LED source to match an "ideal" spectrum.  It's not worth it.

 

You can mix LEDs, however that's a very difficult problem.  The LED technologies involved in producing different colors have different linearity and thermal responses.  It would require nearly real-time calibration to keep the LEDs semi-sane.

 

What is you goal in all of this?  Are you looking for an ideal D65 backlight for your color grading suite?  I'm no expert on this topic, but I would think any reasonable D65 lamp would do.  You could even use the Kino K55 bulbs since they run in the 6500-6900K (CCT) region.  That's probably an excellent question for the color grading forum folks to see what works best.

 

Stuart

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

illuma.blogspot.com


  • 0

#12 Kelvin Xu

Kelvin Xu
  • Basic Members
  • PipPip
  • 17 posts
  • Colorist
  • Singapore

Posted 27 August 2015 - 01:10 AM

Kelvin,

 

Gels are very lossy, so you would lose significant light output trying to filter an LED source to match an "ideal" spectrum.  It's not worth it.

 

You can mix LEDs, however that's a very difficult problem.  The LED technologies involved in producing different colors have different linearity and thermal responses.  It would require nearly real-time calibration to keep the LEDs semi-sane.

 

What is you goal in all of this?  Are you looking for an ideal D65 backlight for your color grading suite?  I'm no expert on this topic, but I would think any reasonable D65 lamp would do.  You could even use the Kino K55 bulbs since they run in the 6500-6900K (CCT) region.  That's probably an excellent question for the color grading forum folks to see what works best.

 

Stuart

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

illuma.blogspot.com

 

I am new to this area so asking a lot of questions may get me to know new ideas and methods. I understand Kino has a good portfolio of lights. 


  • 0


Rig Wheels Passport

CineLab

Visual Products

Ritter Battery

Media Blackout - Custom Cables and AKS

The Slider

Pro 8mm

Technodolly

Paralinx LLC

Abel Cine

CineTape

rebotnix Technologies

Aerial Filmworks

Tai Audio

Willys Widgets

Glidecam

Zylight

Broadcast Solutions Inc

Glidecam

Media Blackout - Custom Cables and AKS

CineLab

Visual Products

Broadcast Solutions Inc

Tai Audio

rebotnix Technologies

Abel Cine

Willys Widgets

Rig Wheels Passport

Ritter Battery

CineTape

Pro 8mm

Zylight

Paralinx LLC

Technodolly

The Slider

Aerial Filmworks