Gray card re-visited

Ok I think I have your formula down:
The amount of reflected light is the .18
The (1.0/2.2) is the gamma ratio? And it equals .454545
.18 to the power of .454545 = .459
.459 multiplied by the full range of possible grayscale output (.459*255) = 117

So printing a gray card using RGB 117,117,117 would most closely match the actual shade of a gray card? Has anyone tried this yet? How is the exposure when spot metered and the EV set to 0EV?

Just trying to figure out the specifics here, and I appreciate your including the formula very much.

David Rosenthal said:

The luminance that comes out of you monitor is not linear in RGB
vaule--this is to do with the gamma of the monitor. The color
matching done by all printers compensates for this fact by not
being linear either. Assuming a gamma of 2.0(back of envelope
standard) -> 2.2(srgb) and inverting the resultant equation gives a
solution of the RGB value needed to give a 18% gray card:

[0.18 ^ (1.0 / 2.2)] * 255 = 117
[0.18 ^ (1.0 / 2.0)] * 255 = 108

Click to expand...

--
Shay

My Sony F707 Gallery: http://www.shaystephens.com/portfolio.asp

Shay Stephens said:

Ok I think I have your formula down:
The amount of reflected light is the .18
The (1.0/2.2) is the gamma ratio? And it equals .454545
.18 to the power of .454545 = .459
.459 multiplied by the full range of possible grayscale output
(.459*255) = 117

Click to expand...

You got it.

Shay Stephens said:

So printing a gray card using RGB 117,117,117 would most closely
match the actual shade of a gray card? Has anyone tried this yet?
How is the exposure when spot metered and the EV set to 0EV?

Click to expand...

This should be the best all-around simple way to generate a gray card. If your printer isn't quite adjusted or whatever obviously the result can be off. Glancing at the above posted printouts, it does seem that 128 is too bright and 46 is too dark

Shay Stephens said:

Just trying to figure out the specifics here, and I appreciate your
including the formula very much.

Click to expand...

Sure thing.

Cheers,
Dave

Surprisingly it's very close. Depending on what the angle of the card is, most of the time the EV doesn't change. On some angles, however, it's .3 off. So its somewhere between 0 and -.3.

Shay Stephens said:

I see what your getting at, the printed card does look close to
your gray card, and Yehuda did mention that his printed out too
light when using 128,128,128. How well does the exposure look when
you meter off of the 46,46,46 card?

--
Shay

My Sony F707 Gallery: http://www.shaystephens.com/portfolio.asp

Click to expand...

Indeed the luminance from most monitors are not linear. However, monitors that can be calibrated use gamma correction for this. My monitor has been calibrated and 46-46-46 is the best match against my gray card.

Note that many monitors, especially the old or cheap ones, cannot be calibrated.

So its no wonder why your monitor may show a value between 108 and 117.

David Rosenthal said:

1) (128, 128, 128) is just wrong.

2) (46, 46, 46) is on the right track.

3) Between (108, 108, 108) and (117, 117, 117) is the right answer.

4) It's amazing the "proof" that people come up with for wrong answers

Reason:

The luminance that comes out of you monitor is not linear in RGB
vaule--this is to do with the gamma of the monitor. The color
matching done by all printers compensates for this fact by not
being linear either. Assuming a gamma of 2.0(back of envelope
standard) -> 2.2(srgb) and inverting the resultant equation gives a
solution of the RGB value needed to give a 18% gray card:

[0.18 ^ (1.0 / 2.2)] * 255 = 117
[0.18 ^ (1.0 / 2.0)] * 255 = 108

With a properly calibrated printer this should work fairly well.
And do try to use a matte finish paper is that is an option with
your printer.

Cheers,
Dave

Yehuda Katz said:

Hi friends,

I want to print by myself a gray card. Can you give me rhe R-G-B
value of 18% gray?

Thank you, Yehuda

Click to expand...

Click to expand...

Thanks for the formula!
Here is the greyscale I produced with rgb 117,117,117 using Fireworks:



And this is Shay Stephens' card that was posted earlier ( http://www.dpreview.com/forums/read.asp?forum=1009&message=2804290 ). Note that I just use part of the card for demonstration purposes. These two colors are indeed are very close match (to my monitor, at least):

David Rosenthal said:

Shay Stephens said:

Ok I think I have your formula down:
The amount of reflected light is the .18
The (1.0/2.2) is the gamma ratio? And it equals .454545
.18 to the power of .454545 = .459
.459 multiplied by the full range of possible grayscale output
(.459*255) = 117

Click to expand...

You got it.

David Rosenthal said:

So printing a gray card using RGB 117,117,117 would most closely
match the actual shade of a gray card? Has anyone tried this yet?
How is the exposure when spot metered and the EV set to 0EV?

Click to expand...

This should be the best all-around simple way to generate a gray
card. If your printer isn't quite adjusted or whatever obviously
the result can be off. Glancing at the above posted printouts, it
does seem that 128 is too bright and 46 is too dark

David Rosenthal said:

Just trying to figure out the specifics here, and I appreciate your
including the formula very much.

Click to expand...

Sure thing.

Cheers,
Dave

Click to expand...

--
Tom

I just did a print test using this test pattern:



I metered a scene in a room that had no outside lighting and was lit only by incandescent lights (ok yes it was the bathroom hehehe). I used my real gray card to set the camera exposure to read 0EV then I held up the printed gray card test pattern shown above. Both the 117 and 128 gave me a meter reading of 0EV and the 46 version gave a -.7EV reading.

I printed the test pattern on an Epson Photo Stylus 780 using the black ink only option and a 2.2 gamma from the print dialog box on inkjet paper using the 360dpi ink jet printer paper option. I am using a Windows PC.

I also tried doing a white balance off the printed gray card and it looked close to what the real gray card and white paper result were.

--
Shay

My Sony F707 Gallery: http://www.shaystephens.com/portfolio.asp

Jim Walker said:

Indeed the luminance from most monitors are not linear. However,
monitors that can be calibrated use gamma correction for this. My
monitor has been calibrated and 46-46-46 is the best match against
my gray card.

Click to expand...

When one calibrates their monitor, they are usually not calibrating it to a linear profile. They are usually calibrating it to a gamma of 2.2 so that they can work compatiablly in the sRGB color space. This non-linear color space is the de facto standard of all common imaging devices. It is how cameras output their JPGs, it is how monitors work electrically, and it is what printer drivers are programmed to match.

If someone does indeed use their monitor in a linear mode (46, 46, 46) will be the correct value for 18% gray. But, calibrating your monitor this way is a bad idea. The non-linear sRGB standard was chosen over the simpler linear one because of the positive influences on the encoding of high-dynamic range images and you should use it.

Bottom line: Unless you specifically know otherwise (117, 117, 117) is your best bet.

proteus9 said:

So its no wonder why your monitor may show a value between 108 and
117.

Click to expand...

I didn't test this on my monitor to come up with this number. I used the accepted theory about color correction.

Cheers,
Dave

Thanks for the experimant to back up the theory! I guess that 117 and 128 both came + - 0. 117 should be an 18% gray and 128 should be a 22% gray -- close indeed but I'd go with the 117

Cheers,
Dave

Shay Stephens said:

I just did a print test using this test pattern:



I metered a scene in a room that had no outside lighting and was
lit only by incandescent lights (ok yes it was the bathroom
hehehe). I used my real gray card to set the camera exposure to
read 0EV then I held up the printed gray card test pattern shown
above. Both the 117 and 128 gave me a meter reading of 0EV and the
46 version gave a -.7EV reading.

I printed the test pattern on an Epson Photo Stylus 780 using the
black ink only option and a 2.2 gamma from the print dialog box on
inkjet paper using the 360dpi ink jet printer paper option. I am
using a Windows PC.

I also tried doing a white balance off the printed gray card and it
looked close to what the real gray card and white paper result were.

--
Shay

My Sony F707 Gallery: http://www.shaystephens.com/portfolio.asp

Click to expand...

I agree, and the math and metering results would back that up too. I have entered 117,117,117 in my note book for printing gray cards in a pinch. Thanks a million for the info!

--
Shay

My Sony F707 Gallery: http://www.shaystephens.com/portfolio.asp

David Rosenthal said:

When one calibrates their monitor, they are usually not
calibrating it to a linear profile. They are usually calibrating
it to a gamma of 2.2 so that they can work compatiablly in the sRGB
color space.
This non-linear color space is the de facto standard
of all common imaging devices. It is how cameras output their
JPGs, it is how monitors work electrically, and it is what printer
drivers are programmed to match.
If someone does indeed use their monitor in a linear mode (46, 46,
46) will be the correct value for 18% gray. But, calibrating your
monitor this way is a bad idea.

Click to expand...

The whole idea of using RGB values is to match what the human senses.

The RGB values become meaningless if two people set their monitor to match a color and then come up with two different RGB values.

David Rosenthal said:

Bottom line: Unless you specifically know otherwise (117, 117, 117)
is your best bet.

Click to expand...

My monitor is calibrated. Setting the values to 117-117-117 makes it far to light compared to my grey card.

David Rosenthal said:

David Rosenthal said:

So its no wonder why your monitor may show a value between 108 and
117.

Click to expand...

I didn't test this on my monitor to come up with this number. I
used the accepted theory about color correction.

Click to expand...

Ah that's the problem with trying to rely on misplaced theories. Try looking!

I looked at it and saw that your numbers don't match. Furthermore, it also agrees with my calibrated monitor, the printouts and the percentage numbers.

Shay Stephens said:

I just did a print test using this test pattern:



I metered a scene in a room that had no outside lighting and was
lit only by incandescent lights (ok yes it was the bathroom
hehehe). I used my real gray card to set the camera exposure to
read 0EV then I held up the printed gray card test pattern shown
above. Both the 117 and 128 gave me a meter reading of 0EV and the
46 version gave a -.7EV reading.

Click to expand...

Your 117 pattern, and setting my color wheel to 117, produces a lighter shade than the grey card.

Shay Stephens said:

I printed the test pattern on an Epson Photo Stylus 780 using the
black ink only option and a 2.2 gamma from the print dialog box on
inkjet paper using the 360dpi ink jet printer paper option. I am
using a Windows PC.

I also tried doing a white balance off the printed gray card and it
looked close to what the real gray card and white paper result were.

--
Shay

My Sony F707 Gallery: http://www.shaystephens.com/portfolio.asp

Click to expand...

proteus9 said:

The whole idea of using RGB values is to match what the human senses.
...
Ah that's the problem with trying to rely on misplaced theories.
Try looking!

Click to expand...

Give me a break. This condescention doesn't interest me. Shay didn't mind my "misplaced theories" that gave him a perfect result in his experiment. If you think that I don't know what I'm talking about I can assure you that you are wrong.

proteus9 said:

I looked at it and saw that your numbers don't match. Furthermore,
it also agrees with my calibrated monitor, the printouts and the
percentage numbers.
...
My monitor is calibrated. Setting the values to 117-117-117 makes
it far to light compared to my grey card.

Click to expand...

I tried to explain this previously--and i'm sure that you are not lying when you say 46 works for your setup which is (incorrectly) calibrated to a linear profile. I explained this previously.

give me a break,
Dave

To make that equation useful, you'd have to pin the values between the percentage reflectance of white paper, and the percentage reflectance of pure black ink covering your paper.

Right now, the equation wrongly would tell you that you'd get 100% reflectance by printing 255 RGB (white paper), and that you'd get 0% reflectance with and RGB value of 0. Even a perfectly clean mirror doesn't reflect 100% of light, so we shouldn't assume that printing a bright enough RGB value would make paper into a surface that could reflect 100% of light when it came out of the printer.

-jeremy

David Rosenthal said:

1) (128, 128, 128) is just wrong.

2) (46, 46, 46) is on the right track.

3) Between (108, 108, 108) and (117, 117, 117) is the right answer.

4) It's amazing the "proof" that people come up with for wrong answers

Reason:

The luminance that comes out of you monitor is not linear in RGB
vaule--this is to do with the gamma of the monitor. The color
matching done by all printers compensates for this fact by not
being linear either. Assuming a gamma of 2.0(back of envelope
standard) -> 2.2(srgb) and inverting the resultant equation gives a
solution of the RGB value needed to give a 18% gray card:

[0.18 ^ (1.0 / 2.2)] * 255 = 117
[0.18 ^ (1.0 / 2.0)] * 255 = 108

With a properly calibrated printer this should work fairly well.
And do try to use a matte finish paper is that is an option with
your printer.

Cheers,
Dave

Yehuda Katz said:

Hi friends,

I want to print by myself a gray card. Can you give me rhe R-G-B
value of 18% gray?

Thank you, Yehuda

Click to expand...

Click to expand...

--
Jeremy Birn
http://www.3dRender.com/

David Rosenthal said:

David Rosenthal said:

The whole idea of using RGB values is to match what the human senses.
...
Ah that's the problem with trying to rely on misplaced theories.
Try looking!

Click to expand...

Give me a break. This condescention doesn't interest me. Shay
didn't mind my "misplaced theories" that gave him a perfect result
in his experiment. If you think that I don't know what I'm talking
about I can assure you that you are wrong.

Click to expand...

I'm not trying to be condescending here. Rrather, I'm trying to point out that if humans see a difference in a color match and a theory says they match, that means the theory is incorrect.

David Rosenthal said:

David Rosenthal said:

I looked at it and saw that your numbers don't match. Furthermore,
it also agrees with my calibrated monitor, the printouts and the
percentage numbers.
...
My monitor is calibrated. Setting the values to 117-117-117 makes
it far to light compared to my grey card.

Click to expand...

I tried to explain this previously--and i'm sure that you are not
lying when you say 46 works for your setup which is (incorrectly)
calibrated to a linear profile. I explained this previously.

Click to expand...

Now you give me a break! How in the world could you possibly know whether or not my setup is calibrated incorrectly?

Why not find a grey card, take it to a paint store or a hardware store, and talk the clerk into measuring it with their paint-matching machine & mixing a batch of "18% gray" paint??

Two things to consider though: DO NOT try to do this if the store is busy or the clerk is not receptive to your request (!) and look at their formula before they mix. It should be the equivalent of lamp-black and white. Any other tints might be a bad card (the cards that I've seen & purchased tend to have excess green), a bad reading, or compensating for the base tint that they're going to use.

Finally (if you do this) keep in mind what you'll be trying to paint (over) not all paint are opaque, but several coats should take care of that.

cvt

Jeremy Birn61459 said:

Jeremy Birn61459 said:

Yehuda Katz said:

Hi friends,

I want to print by myself a gray card. Can you give me rhe R-G-B
value of 18% gray?

Thank you, Yehuda

Click to expand...

Click to expand...

--
Jeremy Birn
http://www.3dRender.com/

Click to expand...

--
C. V. Tintera

Not sure that would work. Does the pantone scale directly correspond with the RGB scale?

Chuck Tintera said:

Why not find a grey card, take it to a paint store or a hardware
store, and talk the clerk into measuring it with their
paint-matching machine & mixing a batch of "18% gray" paint??

Click to expand...

--

Ulysses

It appears that there's a belief that an RGB value between 108 and 117 will give the correct printed out value to match a gray card.

I experimented further and took a photos of a gray card against printed values of RGB 46-46-46, RGB 108-108-108, RGB 117-117-117, and RGB 128-128-128.

The top photo was of a laserjet printout, and the bottom photo of an inkjet printout. As you can see the gray card still best matches RGB 46-46-46 in both cases.

I also experimented by setting exposures against all four printed RGB values for the laser and inkjet printouts after calibrating against a gray card. The results are as follows:

For the laserjet printout:

The RGB 128 was off by +1EV
The RGB 117 and RGB 108 were off by +.7EV
The RGB 46 was within 0EV and -.3EV

For the inkjet printout:

The RGB 117 and 128 was off by +1.7EV
The RGB 108 was within +1.3EV and +1.7EV
The RGB 46 was within +.3EV and +.7EV

Because the characteristics of the printer, the ink density can vary (darkest with a new cartridge, lighter when the ink is low), thus results will vary. The laser printer I used had a brand new ink cartridge, which I think gives the reason why RGB 46 appeared a little darker than the grey card (and thus an EV reading within -.3EV). Adjusting the RGB values up or down would improve this for a particular printer. On the inkjet printer, however, the cartridge had been used awhile, and you'll notice the RBG 46 appears a little lighter than the gray card. You will also get varying results if you set the printer for best or draft quality.

A better test would be to use a calibrated gray swatch, set to RGB 46-46-46, and then do the same tests above (but I have no idea where to get one).

However, it's becoming clearer that the best RGB value to match an 18% gray card is at or very near 46-46-46. There are at least 4 different methods which confirm this theory:

1) It agrees closely with color matching a gray card against my calibrated monitor with RGB 46-46-46.
2) RGB 46 agrees with the numbers (calculating 18%) from PSE's ColorPicker.

3) It closely agrees with color matching a gray card (by eye) against a printed RGB 46.
4) It closely agrees with the exposure settings compared against a gray card.

In conclusion, it appears that using an RGB printout is not a very good idea unless you can adjust the RGB values against a gray card (but then you wouldn't need a printout in the first place). However, if you don't have a gray card handy, it works better than nothing, and your best chances of matching a gray card is by setting the RGB values to 46-46-46.

And just what type of paint would they mix up for you? It would have to be a special non-reflective paint that would dry to a matte type finish. I can't see how this would be a practical alternative.

• David

Chuck Tintera said:

Chuck Tintera wrote:
Why not find a grey card, take it to a paint store or a hardware
store, and talk the clerk into measuring it with their
paint-matching machine & mixing a batch of "18% gray" paint??

Two things to consider though: DO NOT try to do this if the store
is busy or the clerk is not receptive to your request (!) and look
at their formula before they mix. It should be the equivalent of
lamp-black and white. Any other tints might be a bad card (the
cards that I've seen & purchased tend to have excess green), a bad
reading, or compensating for the base tint that they're going to
use.

Finally (if you do this) keep in mind what you'll be trying to
paint (over) not all paint are opaque, but several coats should
take care of that.

cvt

Chuck Tintera said:

Chuck Tintera said:

Yehuda Katz said:

Hi friends,

I want to print by myself a gray card. Can you give me rhe R-G-B
value of 18% gray?

Thank you, Yehuda

Click to expand...

Click to expand...

--
Jeremy Birn
http://www.3dRender.com/

Click to expand...

--
C. V. Tintera

Click to expand...

At Home Depot they use a mechanical eye of some sort, connected to a computer, which then matches the color. You put your color swatch against the eye and the machine mixes the paint to match it. I asked the guy what color scale it used, but he had no idea how the thing works.

Ulysses said:

Chuck Tintera said:

Why not find a grey card, take it to a paint store or a hardware
store, and talk the clerk into measuring it with their
paint-matching machine & mixing a batch of "18% gray" paint??

Click to expand...

--

Ulysses

Click to expand...

I would guess that a flat or slate gray would work best.

Home Depot has free color matching squares that you can take home and test. I picked up several and tried it. The one called "Palais(sic?) Royal - TH12 worked best. I metered off of it and it appeared to work just like a gray card.

However, it could still be off by ±.2EV (within the tolerance of the lowest EV unit), so I wouldn't trust it over an actual calibrated gray card.

David_S85 said:

And just what type of paint would they mix up for you? It would
have to be a special non-reflective paint that would dry to a matte
type finish. I can't see how this would be a practical alternative.

Click to expand...