Daylight lamp as white reference? - Page 3 - AVS Forum
Forum Jump: 
 
Thread Tools
Old 04-21-2012, 06:00 PM
AVS Special Member
 
sotti's Avatar
 
Join Date: Aug 2004
Location: Seattle, WA
Posts: 6,640
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 44 Post(s)
Liked: 175
Quote:
Originally Posted by PlasmaPZ80U View Post

this is going beyond what I know about calibration so I'll just briefly state the point I'm trying to make...

Calibrating grayscale/gamma and gamut is best done with meters/software. The optical comparator approach for grayscale will not provide comparable results in general and can be expensive when done properly (using high quality light source and white card).

I completely agree with that.
Our eyes are the best tool we have to see if colors match or not. But once you move beyond white point (you can't even really do gamma with an optical comparator) the complexity goes well beyond what you can do with a lamp and cards.

Joel Barsotti
SpectraCal
CalMAN Lead Developer
sotti is online now  
Sponsored Links
Advertisement
 
Old 04-21-2012, 06:46 PM
 
Phase700B's Avatar
 
Join Date: Jan 2004
Posts: 2,508
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 24
Probably true. Yet some have used other methods for gamma. Here is one example that some have used:

http://www.quickgamma.de/indexenv4.html

And for grayscale there are also methods using a monochrome monitor as an emmisive source to visually match and correct. It may or may not be as exact as a professional meter, but sometimes better than some meters I've seen.


And grayscale (purity) has been set for years using optical compartors and can correct visible gray scale errors quite successfully. Especially in 2 point or simple RGB high and low cut and gain systems.

And calibrating with meters is best done with those certified as accurate and likely not hobbiest grade meters unless rented from a lab or service that guarantees up to date certification. Plus the person using said meters, software and hardware knows what they are doing. . . which can also be another sort of can.
Phase700B is offline  
Old 04-23-2012, 05:32 PM
AVS Special Member
 
sotti's Avatar
 
Join Date: Aug 2004
Location: Seattle, WA
Posts: 6,640
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 44 Post(s)
Liked: 175
Quote:
Originally Posted by ADU View Post


The CIE values they sent me (x=0.313, y=0.337) are also not a precise match to D65 (x=0.3127, y=0.3290). If the values they gave me are correct, then the bulbs should be a hair off D65 towards green. (There are couple suggestions on how to compensate for that here.)

Looks like the values they gave were decently close to what it measured at.

Granted, the numbers are still only about as close as what alot of sets default to in their warm mode.



Equipment was a CS200 and a NIST traceable reflective tile (99.99% pure reflective).
LL

Joel Barsotti
SpectraCal
CalMAN Lead Developer
sotti is online now  
Old 04-23-2012, 05:34 PM
AVS Special Member
 
sotti's Avatar
 
Join Date: Aug 2004
Location: Seattle, WA
Posts: 6,640
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 44 Post(s)
Liked: 175
Quote:
Originally Posted by Phase700B View Post

And grayscale (purity) has been set for years using optical compartors and can correct visible gray scale errors quite successfully.

While true, it also hasn't been used professionally for almost a decade. This is sort of like advocating slide rules in place of calculators.

Joel Barsotti
SpectraCal
CalMAN Lead Developer
sotti is online now  
Old 04-23-2012, 08:40 PM
AVS Addicted Member
 
DaGamePimp's Avatar
 
Join Date: Jan 2001
Location: WA State
Posts: 15,593
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 54 Post(s)
Liked: 169
Nice CalMan 5 tease there Joel , what is the ETA?

Jason
DaGamePimp is offline  
Old 04-23-2012, 09:24 PM
AVS Special Member
 
sotti's Avatar
 
Join Date: Aug 2004
Location: Seattle, WA
Posts: 6,640
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 44 Post(s)
Liked: 175
Quote:
Originally Posted by DaGamePimp View Post

Nice CalMan 5 tease there Joel , what is the ETA?

Jason

We've got a couple major pieces to put in and some polish, but soon definitely before summers over.

Joel Barsotti
SpectraCal
CalMAN Lead Developer
sotti is online now  
Old 04-24-2012, 10:45 AM
AVS Special Member
 
PlasmaPZ80U's Avatar
 
Join Date: Feb 2009
Posts: 7,291
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 112 Post(s)
Liked: 204
Quote:
Originally Posted by sotti View Post

While true, it also hasn't been used professionally for almost a decade.

interesting
PlasmaPZ80U is offline  
Old 04-24-2012, 02:34 PM
AVS Special Member
 
HDTVChallenged's Avatar
 
Join Date: Jan 2002
Posts: 8,578
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 3 Post(s)
Liked: 146
Quote:
Originally Posted by PlasmaPZ80U View Post

interesting

There was a time when the technology/accuracy of an i1D2 would have set you back at least $1000 ... and it might have only worked on a DirectView CRT ... and you would have probably had to send it in at least once a year for re-certification ... and in some cases you might have needed a calibration(profile) file for individual display models ...

The "interesting" part about it is that people complain(ed) about the accuracy/durability of the i1D2 when it could be had for $120-$150 ....

Have we made progress??? .... I guess
HDTVChallenged is offline  
Old 04-24-2012, 05:52 PM
AVS Special Member
 
PlasmaPZ80U's Avatar
 
Join Date: Feb 2009
Posts: 7,291
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 112 Post(s)
Liked: 204
Quote:
Originally Posted by HDTVChallenged View Post

There was a time when the technology/accuracy of an i1D2 would have set you back at least $1000 ... and it might have only worked on a DirectView CRT ... and you would have probably had to send it in at least once a year for re-certification ... and in some cases you might have needed a calibration(profile) file for individual display models ...

The "interesting" part about it is that people complain(ed) about the accuracy/durability of the i1D2 when it could be had for $120-$150 ....

Have we made progress??? .... I guess

Well, now that the D3 family of meters has taken the place of the D2 and with bundles like these two available for those getting into DIY calibration, there's no need for the D2 either.

Whether the D2 or the optical comparator approach is better on any given set would be dependent on the quality of the light source and the white card used in the optical comparator and the accuracy of the D2 on the given display. Unless you spend about $300 or so on the right parts, the optical comparator approach would be proabably be considerably worse than the initial accuracy of the D2, not to say that either of these options are worthwhile at this time given the availability of the D3 family of meters.

Also, CalMAN or ChromaPure software is worth the extra cost over the free ColorHCFR software and you have to factor in that cost when you look at the two bundles I linked to above.
PlasmaPZ80U is offline  
Old 04-24-2012, 11:28 PM
AVS Special Member
 
HDTVChallenged's Avatar
 
Join Date: Jan 2002
Posts: 8,578
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 3 Post(s)
Liked: 146
Quote:
Originally Posted by PlasmaPZ80U View Post

Well, now that the D3 family of meters has taken the place of the D2 and with bundles like these two available for those getting into DIY calibration, there's no need for the D2 either.

... unless you still find it silly to spend $400-$600 to calibrate a $525 TV set.

Granted the "rent-a-kit option" is probably a step in the right direction, if only the software would include a few more options.

I discovered that I had the foresight to save a base/CMSless run of my LK450's primaries/secondaries 8 months ago, and I just figured out how to use XYZ adjustment matrix in HCFR without bringing the measured whitepoints into the mix. So in theory, I should be able to detect and possibly correct any meter drift based on only the primary measurements ... assuming the set's backlight spectrum and panel chromaticity remain stable ... which are *big* (and foolish) assumptions.

Clearly, sooner or later, no matter what instrument you use, you're going to have to re-certify/profile it against a reference instrument. You can spend $150 or $600 or $1000 on that instrument .... or you can just flail away with greycards and quasi-6500K bulbs which works about as well as expecting a room full of Chimpanzees to produce the collected works of Shakesphere. ... Consumer's choice ...


Which is all a long way of saying that these "issues" have *always* been present with the instrumentation ... I just find it "curious" that it only became a major show stopping event!!! worthy of much weeping and gnashing of teeth after the cost of the meters fell below $200ish thus becoming affordable to the non-professional.
HDTVChallenged is offline  
Old 04-25-2012, 11:36 AM
AVS Special Member
 
PlasmaPZ80U's Avatar
 
Join Date: Feb 2009
Posts: 7,291
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 112 Post(s)
Liked: 204
Quote:
Originally Posted by HDTVChallenged View Post

... unless you still find it silly to spend $400-$600 to calibrate a $525 TV set.

Well, I never fully understood that logic. Does a $1000 or $1500 set necessarily calibrate better or provide a more satisfying calibration than a $500 one? I believe it's not the price tag that matters but rather what picture settings are available for calibration and how well they work/whether they have sufficient range and precision. If LG offers it's full set of calibration controls in its entry-level LCD models, than why not get the most of those sets with proper, complete calibration? Samsung offers 10-pt grayscale/gamma and full 3D CMS in its 600/6000 series sets and higher. I'd say any set that can be calibrated well is worth calibrating right and the price tag is not a relevant factor in determining how well a given set calibrates.
PlasmaPZ80U is offline  
Old 04-25-2012, 11:41 AM
AVS Special Member
 
PlasmaPZ80U's Avatar
 
Join Date: Feb 2009
Posts: 7,291
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 112 Post(s)
Liked: 204
Quote:
Originally Posted by HDTVChallenged View Post

Which is all a long way of saying that these "issues" have *always* been present with the instrumentation ... I just find it "curious" that it only became a major show stopping event!!! worthy of much weeping and gnashing of teeth after the cost of the meters fell below $200ish thus becoming affordable to the non-professional.

A very interesting and valid point. I guess people's standards go up as technology advances and price points drop on older technologies. If there weren't any meters available today that outperformed the D2, I'm willing to bet people would be far less critical of it.
PlasmaPZ80U is offline  
Old 04-25-2012, 11:59 AM
AVS Special Member
 
HDTVChallenged's Avatar
 
Join Date: Jan 2002
Posts: 8,578
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 3 Post(s)
Liked: 146
Quote:
Originally Posted by PlasmaPZ80U View Post

Well, I never fully understood that logic. Does a $1000 or $1500 set necessarily calibrate better or provide a more satisfying calibration than a $500 one?

.... No but it would probably be a *bigger* one. The balk is that you (or I) would be pretty much doubling the (originally reasonably) priced set.

Now if I had more than one set to calibrate or a display that required frequent rechecks (front projection) the cost/benefit ratio definitely changes ...

To be honest, I never intended for my LK450 to be my primary display ... I originally intended to use it for daylight and non-critical TV viewing ... and gaming ... mounted "behind" a drop down screen ... plans, mice, men ... et all.
HDTVChallenged is offline  
Old 05-08-2012, 06:42 PM
ADU
AVS Special Member
 
ADU's Avatar
 
Join Date: Apr 2003
Posts: 6,482
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 132 Post(s)
Liked: 176
Quote:
Originally Posted by sotti View Post

Looks like the values they gave were decently close to what it measured at.

Very close indeed (to the values they sent me, that is)! And just a bit off D65 towards green or yellow-green, as anticipated. (A couple tips on how to compensate for that here, as mentioned previously.)

Thank you for checkin this Joel. Was this a "cold" or "warm" reading btw (if you know)?

Quote:


Granted, the numbers are still only about as close as what alot of sets default to in their warm mode.

Can't really comment about that in general. But I believe there's alot of inconsistency in the Warm settings on displays. I know that the Warm setting on my own TV was certainly much worse than this to start with. I dug up a couple old reviews, and they both say the Warm setting on the 34XBR800 was too blue out of the box.

http://www.hometheater.com/content/2...ny-kv-34xbr800
http://reviews.cnet.com/Sony_WEGA_KV...7-8879879.html

And removing the anti-glare filter from the TV made the picture even bluer. (The left side of the screen has filter removed in the picture below.)


ADU
ADU is offline  
Old 05-08-2012, 09:43 PM
AVS Special Member
 
sotti's Avatar
 
Join Date: Aug 2004
Location: Seattle, WA
Posts: 6,640
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 44 Post(s)
Liked: 175
Quote:
Originally Posted by ADU View Post

Thank you for checkin this Joel. Was this a "cold" or "warm" reading btw (if you know)?

It was warm, it had been on for at least an hour.

Joel Barsotti
SpectraCal
CalMAN Lead Developer
sotti is online now  
Old 05-09-2012, 05:43 PM
ADU
AVS Special Member
 
ADU's Avatar
 
Join Date: Apr 2003
Posts: 6,482
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 132 Post(s)
Liked: 176
^ Cool.

Quote:
Originally Posted by ADU View Post

And just a bit off D65 towards green or yellow-green, as anticipated. (A couple tips on how to compensate for that here, as mentioned previously.)

Actually, since the CIE xy values are known for the GE daylight CFL bulbs, there's probably a simpler and potentially more reliable way to compensate for the difference in color between them and D65 than just fudging the green setting lower on the display, as described in the link above.

The "error" in the GE bulbs could instead be built into a special set of tinted or "profiled" grayscale patterns designed to match the color of the GE bulbs on an accurate D65 display. If the "profiled" patterns are used for optical comparison with the GE bulbs instead of regular neutral gray patterns, the result should be more precise D65 grays on the display (without the need to tweak green levels on the display after the fact).

To put it simply, if you match the "tinted" patterns on the display to the color of the GE Daylight bubls, then the neutral grays on the display should wind up pretty damn close to D65; or at least as close as just about any eyeballing method could possibly allow. By building the difference between D65 and the GE bulbs into the patterns you save yourself the trouble and guesswork of how to tweak the green values lower later on, and also end up with a more accurate result.

Assuming the colors on Joel's graph above are accurate (and they usually seem to be ), then the closest "white" color match to a GE Daylight bulb on a D65 display should be in the neighborhood of R=238, G=255, B=231*. And the closest 50% gray should be about half that, or around R=119, G=128, B=116*. (This is in 0-255 "full swing" RGB values, as opposed to 16-235 compressed video levels.)

I haven't done any xyY -> sRGB math to confirm this yet. But in theory, if you use "grayscale" patterns which are based on percentages of the above R=238, G=255, B=231* white values for optical comparison with the GE bulbs, then the neutral grays on the display ought to come out pretty close to D65.

For this to work best, you should have a player or other output device (ie, PC with HDMI out) that can output correct color to begin with (which could be a stretch for some players). And the RGB primaries on the display should be as close to the HDTV/Rec. 709 spec as possible. (IOW, a "wide gamut" display should be switched to the "normal gamut" setting, or whichever gamut is closest to Rec. 709, if that's possible.)

*Edit: please see updated values in post below.

ADU
ADU is offline  
Old 05-09-2012, 05:54 PM
AVS Special Member
 
sotti's Avatar
 
Join Date: Aug 2004
Location: Seattle, WA
Posts: 6,640
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 44 Post(s)
Liked: 175
Quote:
Originally Posted by ADU View Post

D65 display should be in the neighborhood of R=238, G=255, B=231. And the closest 50% gray should be about half that, or around R=119, G=128, B=116. (This is in 0-255 "full swing" RGB values, as opposed to 16-235 compressed video levels.)

I haven't done any xyY -> sRGB math to confirm this yet. But in theory, if you use "grayscale" patterns which are based on percentages of the above R=238, G=255, B=231 white values for optical comparison with the GE bulbs, then the neutral grays on the display ought to come out pretty close to D65.

This stuff doesn't work because the XYZ -> RGB matrix's needs to be based off the native chromaticity of the display to even come close to working correctly.

Then couple that with the fact that the RGB balance to mix to any particular color requires precise knowledge of the displays gamma and you're back to mine as well put the TV in warm/warm2 and forget about it.

Joel Barsotti
SpectraCal
CalMAN Lead Developer
sotti is online now  
Old 05-16-2012, 07:33 PM
ADU
AVS Special Member
 
ADU's Avatar
 
Join Date: Apr 2003
Posts: 6,482
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 132 Post(s)
Liked: 176
Quote:
Originally Posted by sotti View Post

This stuff doesn't work because the XYZ -> RGB matrix's needs to be based off the native chromaticity of the display to even come close to working correctly.

Then couple that with the fact that the RGB balance to mix to any particular color requires precise knowledge of the displays gamma and you're back to mine as well put the TV in warm/warm2 and forget about it.

If the reference light source was not so close to D65 in color, then the issues above might be more of a concern. However, since the GE bulbs are already pretty close to D65, any differences in display gamma or color primaries should have a negligible effect on the end result.

For those desiring better accuracy though, I calculated the GE bulb's RGB equivalents for displays ranging between a "flat" 2.0 and 2.6 gamma. And here are the 0-255 RGB codes for both your measured CIE xy coordinates and the ones provided by GE.

MEASURED CIE CHROMATICITY: x=.314038, y=.339542
CCT: 6363K



 

Display Gamma: Red: Green: Blue: Notes:
2.0 246 255 243  
2.1 246 255 243  
2.2 247 255 244 "White" codes used for GE gradient above.
2.3 247 255 244  
2.4 247 255 245  
2.5 248 255 245  
2.6 248 255 246  

sRGB stimulus, rel. luminance, etc. used to compute above values (calculated with Bruce's CIE color calculator and the sRGB forward/reverse transformation)...
 

sRGB Color Component Stimulus 0-255 RGB Code Rel. Luminance Effective sRGB Gamma
Red: 0.968969 247 .930856 2.2730
Green: 1.0 255 1.0 2.2749
Blue: 0.957575 244 .906190 2.2723


GE'S REPORTED CIE CHROMATICITY: x=.313, y=.337
CCT: 6431K




Display Gamma: Red: Green: Blue: Notes:
2.0 247 255 246  
2.1 247 255 246  
2.2 248 255 246 "White" codes used for GE gradient above.
2.3 248 255 247  
2.4 248 255 247  
2.5 248 255 248  
2.6 249 255 248  

sRGB data used for calculations...
 

sRGB Color Component Stimulus 0-255 RGB Code Rel. Luminance Effective sRGB Gamma
Red: 0.971798 248 .937040 2.2732
Green: 1.0 255 1.0 2.2749
Blue: 0.967702 247 .928094 2.2729


 

I'm not sure why my sRGB values came out different than the RGB color values on your graph. Maybe the Calman software is designed to slightly exaggerate the color differences between D65 and other measured stimuli (the GE bulb, in this case) to make it easier for users to see the difference?

However, if you're looking at this on an ~2.2 display (in the D65 ballpark), then the gradients above should give a pretty good idea of the difference in color between the GE bulb and neutral grays on a D65 display. (The difference is pretty subtle. You might be able to see it better if you increase the brightness of the monitor, or switch the AVS forum skin to "Black" mode using the Quick Style Chooser at the lower lefthand corner of the webpage.)

If the exact display gamma isn't known, then either the sRGB values or something in the middle of the RGB ranges above could probably still be used with pretty decent results.
 

Quote:
Originally Posted by sotti View Post

This stuff doesn't work because the XYZ -> RGB matrix's needs to be based off the native chromaticity of the display to even come close to working correctly.

Obviously, the closer the display's primaries are to Rec. 709, the better the results will be on highly saturated colors with an approach like this. Just because the neutral gray values are adjusted close to D65, that doesn't necessarily mean that more saturated colors on the display will also be accurate. Since all you're doing with eyeballing approachs like this is adjusting the neutral grays close to D65, you pretty much have to accept that there'll probably be some error in the most saturated colors. There are some common sense ways to mitigate that error though, like avoiding the wide gamut modes, and reading reviews and others' results on which gamut (and temperature) settings are most accurate for your display.

Since the GE bulb is so close to D65 though, IMHO, the display primaries/gamut would have to be way off the mark to have any noticeable effect on the accuracy of the grays with this kind of approach.

The biggest variable in this is probably the accuracy of the player used to send patterns to the display, because the color output on consumer equipment is often questionable/unreliable.


ADU
ADU is offline  
Old 05-20-2012, 08:50 PM
ADU
AVS Special Member
 
ADU's Avatar
 
Join Date: Apr 2003
Posts: 6,482
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 132 Post(s)
Liked: 176
Quote:
Originally Posted by ADU View Post

I'm not sure why my sRGB values came out different than the RGB color values on your graph. Maybe the Calman software is designed to slightly exaggerate the color differences between D65 and other measured stimuli (GE bulb, in this case) to make it easier for the user to see those differences?

Joel,

It looks to me like the RGB colors on your CalMAN graph could be the result of simple linear XYZ -> linear RGB translation, without any gamma correction applied. I suspect that's why the colors on the graph look somewhat exaggerated, and are different than my sRGB and 2.2 gamma RGB codes above.

This may make it easier for your users to see a difference between reference white (D65, in this case), and the stimulus being measured (GE bulb). But I don't think it really gives an accurate portrayal of the difference in color between the two, because the linear RGB values would tend to enhance the slight green bias in the GE bulb, making it look a bit worse than it really is.

In case you want to check my math on this, here are the steps used to calculate my encoded sRGB values for the GE bulb from your measured x=.314038, y=.339542 CIE chromaticity coordinates.

 

sRGB Matrix Used for
XYZ->RGB Translation
CIE xyY CIE XYZ

X=Yx/y
Y=Y
Z=Y(1-x-y)/y
Linear RGB

(Calculated with XYZ->RGB matrix indicated in first column.)
sRGB Stimulus

1.055[RGBlinear^(1/2.4)]-.055

(Note: this is where gamma correction is applied.)
0-255 Encoded sRGB

sRGBstim*255
1. Lindbloom Matrix x=0.314038
y=0.339542
Y=0.978524
X=0.905024
Y=0.978524
Z=0.998346
Rlinear=(3.2404542)X+(-1.5371385)Y+(-0.4985314)Z=0.930856

Glinear=(-0.9692660)X+(1.8760108)Y+(0.0415560)Z=1.0

Blinear=(0.0556434)X+(-0.2040259)Y+(1.0572252)Z=0.906191
sRstim=0.968969
sGstim=1.0
sBstim=0.957575
sRcode=247
sGcode=255
sBcode=244
2. Wikipedia Matrix x=0.314038
y=0.339542
Y=0.978457
X=0.904962
Y=0.978457
Z=0.998277
Rlinear=(3.2406)X+(-1.5372)Y+(-0.4986)Z=0.930795

Glinear=(-0.9689)X+(1.8758)Y+(0.0415)Z=1.0

Blinear=(0.0557)X+(-0.2040)Y+(1.0570)Z=0.905980
sRstim=0.968941
sGstim=1.0
sBstim=0.957477
sRcode=247
sGcode=255
sBcode=244


As indicated above, I computed the sRGB codes using both the XYZ->RGB matrix on Bruce Lindbloom's site, and the matrix on Wikipedia's sRGB page. The Y luminance value is scaled to ~0.9785, in both cases, so the brightest RGB component (Green) equals 1.0, or 255 in 0-255 coding. The results are pretty similar, and the final 0-255 encoded sRGB values are identical with both matrices. (The Wikipedia sRGB matrices supposedly come from the official sRGB spec, but Bruce's figures invert better than the more rounded Wikipedia values.)

There are several different ways of deriving equivalent 0-255 RGB codes for displays with "flat gamma" from the data above. The easiest is simply to apply the inverse of the display's gamma to the linear RGB values, and then multiply the resulting RGB stimulus values by 255, as shown on the table below. The values below were also computed using linear RGB data from both the Lindbloom and Wikipedia matrices, as indicated. And the final gamma-corrected RGB codes were the same for both matrices, with only one exception*.

 

  1. Lindbloom Matrix 2. Wikipedia Matrix
Display Gamma RGB Stimulus

RGBlinear^(1/Display Gamma)

where...
Rlinear=0.930856
Glinear=1.0
Blinear=0.906191


(Note: this is where gamma correction is applied.)
0-255 Encoded RGB

RGBstim*255
RGB Stimulus

RGBlinear^(1/Display Gamma)

where...
Rlinear=0.930795
Glinear=1.0
Blinear=0.905980


(Note: this is where gamma correction is applied.)
0-255 Encoded RGB

RGBstim*255
Gamma=2.0 Rstim=0.964809
Gstim=1.0
Bstim=0.951941
Rcode=246
Gcode=255
Bcode=243
Rstim=0.964777
Gstim=1.0
Bstim=0.951830
Rcode=246
Gcode=255
Bcode=243
Gamma=2.1 Rstim=0.966456
Gstim=1.0
Bstim=0.954176
Rcode=246
Gcode=255
Bcode=243
Rstim=0.966426
Gstim=1.0
Bstim=0.954070
Rcode=246
Gcode=255
Bcode=243
Gamma=2.2 Rstim=0.967956
Gstim=1.0
Bstim=0.956212
Rcode=247
Gcode=255
Bcode=244
Rstim=0.967927
Gstim=1.0
Bstim=0.956111
Rcode=247
Gcode=255
Bcode=244
Gamma=2.3 Rstim=0.969328
Gstim=1.0
Bstim=0.958076
Rcode=247
Gcode=255
Bcode=244
Rstim=0.969300
Gstim=1.0
Bstim=0.957979
Rcode=247
Gcode=255
Bcode=244
Gamma=2.4 Rstim=0.970587
Gstim=1.0
Bstim=0.959787
Rcode=247
Gcode=255
Bcode=245
Rstim=0.970560
Gstim=1.0
Bstim=0.959694
Rcode=247
Gcode=255
Bcode=245
Gamma=2.5 Rstim=0.971747
Gstim=1.0
Bstim=0.961364
Rcode=248
Gcode=255
Bcode=245
Rstim=0.971721
Gstim=1.0
Bstim=0.961275
Rcode=248
Gcode=255
Bcode=245
Gamma=2.6 Rstim=0.972818
Gstim=1.0
Bstim=0.962822
Rcode=248
Gcode=255
Bcode=246*
Rstim=0.972794
Gstim=1.0
Bstim=0.962736
Rcode=248
Gcode=255
Bcode=245*


^ This works for sRGB content btw because the reverse sRGB transformation (ie "decoding gamma") perfectly inverts the forward sRGB encoding transformation. So the linear RGB values are the same as the relative luminance of the encoded sRGB values on a properly configured sRGB display (in theory anyway). One could not make the same assumption for Rec. 709 encoded content though, because the "reverse transformation" (EOTF) contained in Rec. 1886 is not the exact inverse of the "forward transformation" (OETF) in Rec. 709. But I digress...

If I multiply the linear RGB values (shown on the first table) by 255 without applying a 1/display gamma correction, then I get 0-255 encoded values which are similar to the RGB values shown on your original graph of the GE bulb (Graph A below). The un-corrected RGB color values in Graph A would look right on a display with a linear 1.0 gamma, but not on a typical ~2.2 gamma computer monitor, or ~2.4 gamma HDTV.

If I apply a 1/2.2 gamma correction to the graph though (as shown in Graph B), then the colors are more in line with the sRGB and 2.2 display gamma RGB codes shown on the tables above. IMO, the gamma-corrected Graph B gives a more accurate representation of the difference in color between D65 (the square in the middle of the graph), and the GE bulb (small white dot above it).





If any of the above math doesn't look right, please say so. Most of it seemed pretty straight forward, and I've run through the data a couple times with a pretty fine-toothed comb, so I'm fairly sure most of it is correct. Alot of number-crunching to grind out the final RGB values though (since I did not use Bruce's CIE calculator this time around).

The xy values shown along the edges of the graph also look a bit off, and make it appear that D65 is situated between y=.33 and y=.34, which (as you know) isn't correct. (Maybe a rounding issue?)


ADU
ADU is offline  
 
Thread Tools


Forum Jump: 

Posting Rules  
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off