Luminous Landscape Forum
Raw & Post Processing, Printing => Colour Management => Topic started by: trinityss on January 29, 2009, 02:31:55 pm
-
Hi,
It is advised to use a native white point when a monitor has no hardware calibration possibilities.
So that is what i want to do but my HP LP2475w has multiple white points?
Is the software fooling me or is that possible?
When i switch between these presets (sRGB - 6500 - 9300 °K) there is no change in:
- the RGB gain values
- Black Level
- gamma curve
[attachment=11239:native.jpg]
Thx!
-
Hi,
It is advised to use a native white point when a monitor has no hardware calibration possibilities.
So that is what i want to do but my HP LP2475w has multiple white points?
Is the software fooling me or is that possible?
When i switch between these presets (sRGB - 6500 - 9300 °K) there is no change in:
- the RGB gain values
- Black Level
- gamma curve
[attachment=11239:native.jpg]
Thx!
Personally I set the white point so that an unprinted sheet of paper in my viewing station matches as closely as possible to my screen white. About 6100k with my 30" ACD. That may require making multiple profiles at varying WP's and trying them at first, but my color balance between screen and print is rarely an issue at all doing this.
-
The native white point of most LCD's is somewhere in the neighborhood of 6500, so that might work if it happens to be close to the white point you need. But the best thing is to do the white paper test like Wayne suggests.
-
The native white point of most LCD's is somewhere in the neighborhood of 6500, so that might work if it happens to be close to the white point you need. But the best thing is to do the white paper test like Wayne suggests.
Actually when i measure my default white point then i get this result:
sRGB= 5300
6500= 6245
9300= 7300
white point value : 100= 6360
So indeed, i can conclude that 6500 is the ONLY real white point of my monitor?
I still find it strange that when i select one of the other white point like sRGB that the value of the RGB gain and the gamma curve is not changed at all.... ?
To minimize the banding effect i should choose 6500, but my white point of my GrafiLiteMode task lamp is 5700... so maybe i should sRGB as well.
Thx,
-
I also have this HP monitor. Though I am not an expert on such matters, it is possible that you are registering multiple white points because the monitor is not uniform/consistent (i.e., in terms of its overall colour output). It seems this is a fairly common problem with this monitor - a number of people have commented on it in this forum (search for threads on LP2475W). One way of testing this, after allowing about 30 minutes for the monitor to warm up, is to display identical B&W photos side by side and look for cast/hue differences, especially between right and left portions of the screen. If the problem is especially bad, you might consider returning/exchaning it.
In any event, hope this might help - good luck!
-
The most reliable way to arrive to the native white point is to RESET the RGB values (right-bottom on your screen shot) and calibrate the monitor to a "native gamma" using calibration device (colorimeter, spectrophotometer) and the corresponding software (e.g., Eye1 Display - Eye-one Match). Calibrating to the native white point ensures that no corrections are done to the video LUT - minimum probability of banding.
Changes in the white white point usually result in changes of brightness and gamma (measured values, not values on the on-screen display). Changes in gamma are small and usually not visible (they can be measured though).
If no software calibration is available, the best choice would be the 6500K (in your case 6245 K) - quite typical white point. All preset values are giving you a white point - just different one. sRGB setting in your case may look yellowish, 9300 may look bluish.
Marcel
more details can be found at:
---------------------------------------------------------
http://www.marcelpatek.com/monitor.html
-
I also have this HP monitor. Though I am not an expert on such matters, it is possible that you are registering multiple white points because the monitor is not uniform/consistent (i.e., in terms of its overall colour output). It seems this is a fairly common problem with this monitor - a number of people have commented on it in this forum (search for threads on LP2475W). One way of testing this, after allowing about 30 minutes for the monitor to warm up, is to display identical B&W photos side by side and look for cast/hue differences, especially between right and left portions of the screen. If the problem is especially bad, you might consider returning/exchaning it.
In any event, hope this might help - good luck!
Hi,
Fortunately i don't have this issue...
The measured values are after a warming up time of 1 hour.
Kr,
-
The most reliable way to arrive to the native white point is to RESET the RGB values (right-bottom on your screen shot) and calibrate the monitor to a "native gamma" using calibration device (colorimeter, spectrophotometer) and the corresponding software (e.g., Eye1 Display - Eye-one Match). Calibrating to the native white point ensures that no corrections are done to the video LUT - minimum probability of banding.
Changes in the white white point usually result in changes of brightness and gamma (measured values, not values on the on-screen display). Changes in gamma are small and usually not visible (they can be measured though).
If no software calibration is available, the best choice would be the 6500K (in your case 6245 K) - quite typical white point. All preset values are giving you a white point - just different one. sRGB setting in your case may look yellowish, 9300 may look bluish.
Marcel
more details can be found at:
---------------------------------------------------------
http://www.marcelpatek.com/monitor.html
Hi Marcel,
The value is 255 / 255 / 255 for the RGB option for ALL 3 presets (sRGB / 6500 / 9300).
When i use these presets and select native white point during calibration (basICColor) then the LUT is slightly changed...
This really confuses me, the backlight should only have 1 color temperature, right?
The brightness nor the gamma is changed either, check this movie:
http://screencast.com/t/5txjqmzb9Kt (http://screencast.com/t/5txjqmzb9Kt)
THx,
-
Hi Marcel,
The value is 255 / 255 / 255 for the RGB option for ALL 3 presets (sRGB / 6500 / 9300).
When i use these presets and select native white point during calibration (basICColor) then the LUT is slightly changed...
This really confuses me, the backlight should only have 1 color temperature, right?
The brightness nor the gamma is changed either, check this movie:
http://screencast.com/t/5txjqmzb9Kt (http://screencast.com/t/5txjqmzb9Kt)
THx,
Values R=G=B=255 are correct. Any white point has exactly RGB=255. You are right - the monitor can have only one white point (color temperature) at the time.
Looking at the movie, it looks like the changes in your White point presets do not alter appearance of the display immediately. However, your measurements indicate that presets work fine (sRGB= 5300K, 6500= 6245K, 9300= 7300K). So the changes must have happened after you finished your on-screen setup.
One clarification - when changes (brightness, gamma, white point) are made through the OSD window (your HP Display Assistant), these are direct hardware changes taking effect through the monitor hardware - not the video lookup table (LUT of your video card). That is why no changes are seen in the Calibration Tester. Calibration Tester reads only values in your video LUT - and nothing has changed there by setting the values directly in your monitor. You would need to calibrate the monitor to see some changes in Calibration Tester. Furthermore, the video LUT has only "correction" data to your monitor preset used in calibration. After calibration to the native white point (and native gamma) the line would be nearly linear. Here is a screenshot of such LUT:
[attachment=11276:Calibration_shot.png]
My recommendation would be to RESET your monitor through the HP Display Assistant and calibrate it using BasICColor and native white point selection there. If Preset White point has to be chosen, I would go with 6500 (in your case 6245K), which is very close to LCD native white points (typically 6000-6400K).
After calibration, you should see some changes (corrections) in the LUT table (via Calibration Tester). Re-measure your new white point directly off the screen (white background, XYZ or xy values). It should be between 6000-6500K.
Marcel
-
This really confuses me, the backlight should only have 1 color temperature, right?
The backlight may well only have one temperature on its own, but the panel doesn't - the backlight shines through the pixels, so depending on how you have the RGB settings configured, the color temperature of the panel as a whole does change. That's how the presets (or whatever customs setting you have) work.
Sandy
-
The various settings of course produce a change. But HOW is the changed made? Well there's no physical alterations of white point available on CCFL LCDs. The native white point is what it is. You can alter this by altering a color lookup table, most displays are doing so in 8-bits which just introduces banding. With such displays, far better to profile this native behavior, not alter it (let the CMS do this). On higher end units, high bit adjustments are conducted in the panel, high bit. So you're not forced into using a Native white point.
The right target white point is the one that results in a print to screen match, the illuminant of which you view the prints will play a significant role in what you would ultimately select. YMMV.
-
Values R=G=B=255 are correct. Any white point has exactly RGB=255. You are right - the monitor can have only one white point (color temperature) at the time.
Looking at the movie, it looks like the changes in your White point presets do not alter appearance of the display immediately. However, your measurements indicate that presets work fine (sRGB= 5300K, 6500= 6245K, 9300= 7300K). So the changes must have happened after you finished your on-screen setup.
There is something wrong with the capture software, there is noticable difference of the colors on screen when i change the presets... you don't see it in the movie :-(
One clarification - when changes (brightness, gamma, white point) are made through the OSD window (your HP Display Assistant), these are direct hardware changes taking effect through the monitor hardware - not the video lookup table (LUT of your video card). That is why no changes are seen in the Calibration Tester. Calibration Tester reads only values in your video LUT - and nothing has changed there by setting the values directly in your monitor. You would need to calibrate the monitor to see some changes in Calibration Tester. Furthermore, the video LUT has only "correction" data to your monitor preset used in calibration. After calibration to the native white point (and native gamma) the line would be nearly linear. Here is a screenshot of such LUT:
[attachment=11276:Calibration_shot.png]
Hmm, i don't have a gamma option... Only brightness - contrast - color (RGB).
When i select "monitor's native" for the calibration part then i get a similar curve like yours. I thought that it would stay lineair, after all you say "monitor's native"... strange!.
By the way, i don't have a native gamma option in BasICColor? Would that only be available for the high-end display's (NEC - EIZO)?
THx!
-
The various settings of course produce a change. But HOW is the changed made? Well there's no physical alterations of white point available on CCFL LCDs. The native white point is what it is. You can alter this by altering a color lookup table, most displays are doing so in 8-bits which just introduces banding. With such displays, far better to profile this native behavior, not alter it (let the CMS do this). On higher end units, high bit adjustments are conducted in the panel, high bit. So you're not forced into using a Native white point.
The right target white point is the one that results in a print to screen match, the illuminant of which you view the prints will play a significant role in what you would ultimately select. YMMV.
Hi Digitaldog,
I have indeed a good match when i select the white point of my GrafiLiteMode but wanted to know i the match would still be good and the banding better when select a native white point...
Somehow i have the feeling that the DisplayAssistant is fooling me, i think i'm going to contact HP's support...
Kr,
-
The various settings of course produce a change. But HOW is the changed made? Well there's no physical alterations of white point available on CCFL LCDs. The native white point is what it is. You can alter this by altering a color lookup table, most displays are doing so in 8-bits which just introduces banding. With such displays, far better to profile this native behavior, not alter it (let the CMS do this). On higher end units, high bit adjustments are conducted in the panel, high bit. So you're not forced into using a Native white point.
No, no really - LCD displays are fundamentally very simple devices - the amount of light transmitted is controlled by the voltage on a pixel. So, simplistically 0 volts = black, 1 Volt = all light transmitted. The 8-bit signal (or more bits in high end displays) just divides that up proportionally, e.g., 0 = 0V, 255= 1 Volt.
When you change the whitepoint of the display, you're changing the voltage that equals 255. So, e.g.,
1. a 9500 whitepoint may mean: Red max voltage = 0.8V, Green max voltage = 1V, Blue max voltage = 1V
1. a 6500 whitepoint may mean: Red max voltage = 0.9V, Green max voltage = 0.9V, Blue max voltage = 1V
Max voltage always equals 255 (in an 8 bit display).
There's no need for a LUT in this process. The display might, depending on how it works internally, use a LUT, but that's not a necessary part of the equation.
Note the above is vastly simplified - e.g., I've left out gamma, black levels, contrast controls, etc. Also most LCD panels (for technical reasons) internally work with 0V = white and max volts = black, etc, but the principle remains the same.
The one thing you need to avoid is setting the display to settings such that it is in saturation, e.g., voltages of above 1V in the example above. How you tell is easy - when you calibrate the panel, check that the transfer curve doesn't have a kink in it at the top or bottom.
Sandy
-
No, no really - LCD displays are fundamentally very simple devices - the amount of light transmitted is controlled by the voltage on a pixel.
I'm not referring to the amount of light, I'm referring to the color of light (in this case white).
You've got a single Lightsource with a native WP. You're saying the display can "alter" the color of white not by introducing a LUT somewhere but by altering the voltage?
-
I'm not referring to the amount of light, I'm referring to the color of light (in this case white).
You've got a single Lightsource with a native WP. You're saying the display can "alter" the color of white not by introducing a LUT somewhere but by altering the voltage?
Color is just the combination of the amounts of red, green and blue light.
And yes, altering color by voltage is exactly what an LCD panel does. Voltage modulates the amount of light that each pixel transmits. If you have RGB pixels, and you change the voltage on the pixels, the color that results from the combination of those pixels changes. You can't change the color of the backlight, and you can't change the color temperature that you would get if all pixels were in their fully transmissive state, by you certainly can (and do) alter the transmissivity (and hence color) of the display by altering those voltages. You can argue that there is some kind of "natural" white point that occurs when all the pixels are maximally transmissive (and thus it would be equal to the backlight plus the impact of the RGB filtration used), but that is really of no practical use for (or interest to) a photographer. All other things aside, (a) the RGB filtration is usually chosen to optimize the gamut across multiple white points, not to give any particular white point, and ( backlights drift over time.
See here for the basics of voltages and pixels: http://en.wikipedia.org/wiki/Liquid_crystal_display (http://en.wikipedia.org/wiki/Liquid_crystal_display)
Sandy
-
There is something wrong with the capture software, there is noticable difference of the colors on screen when i change the presets... you don't see it in the movie :-(
That is probably the case. I do not see any color temperature or brightness changes in the movie. Changes though make sense. You should see a yellowish, neutral, and bluish tint when going through the sRGB, 6500, 9300 presets. These are driven by internal monitor hardware changes and not by video LUT - that's why CalibrationTester shows linear (unchanged) in vs out response.
Hmm, i don't have a gamma option... Only brightness - contrast - color (RGB).
When i select "monitor's native" for the calibration part then i get a similar curve like yours. I thought that it would stay lineair, after all you say "monitor's native"... strange!.
By the way, i don't have a native gamma option in BasICColor? Would that only be available for the high-end display's (NEC - EIZO)?
THx!
Gamma option through the monitor hardware (OSD panel) may indeed by available only for higher-end monitors. Don't worry about it. Calibration to a native gamma further minimizes possibility of banding, particularly in shadows. If BasICColor does not have this feature, your calibration still can perform very well. Manufacturers typically adjust the monitor circuitry to match the gamma of 2.2 In your calibration software I would set the gamma to 2.2 and white point to "Native". Reset the monitor in HPDisplay Assistant first. After the calibration is finished, CalibrationTester will show nearly linear response - that would be a correction to your monitor's hardware settings. It never is precisely linear. The closer to the linear response you are (in Calibration Tester), the better for quality of your images - minimum changes were done in videoLUT (that's good).
Marcel
-
I don't know if this might be of assistance, but I just want to point something else out: when calibrating with BasICColor, my LP2475W display exhibited some odd behaviour. Although it should not be possible, BasICColour would sometimes actually allow a hardware calibration via DDC - it would treat the HP monitor as it does my Eizo, which does not need the additional step of monitor settings adjustment. (The first page of BasICColour's set-up options would have DDC option available, so I chose it to see what would happen.) It actually worked well for a while (even though it is not supposed to work at all), and then became unstable (weird flickering, random colour changes etc.) and so I reverted to regular non-DDC calibration of the HP. I have no explanation for any of this, alas - so just to say that the interaction between the monitor and the calibration software is worth considering as a possible (though probably unlikely) source of the trouble.
-
Color is just the combination of the amounts of red, green and blue light.
Yes it is. And we're talking about the color of white. Can you point out in that article where I'd find how white (all elements twisted to produce white to pass the backlight), one controls the color of white?
-
Yes it is. And we're talking about the color of white. Can you point out in that article where I'd find how white (all elements twisted to produce white to pass the backlight), one controls the color of white?
Umm, I think we may be talking past each other, because that's a 101 question, and you're well beyond 101 questions. But anyway:
Display white is the combination of the light coming from the red, green and blue pixels. If you control the transmissivity of the red, green and blue pixels, the color of white changes. Simply put, you have a display for which RGB signals (255,255,255) = pixel transmissivity (.8, .9, .9), and you then change the display's controls such that RGB signals (255,255,255) = pixel transmissivity (.7, .95, .95), then you have changed the white point. That's what LCD displays with fixed backlights do.
This NEC document may be useful: http://www.digitaladlab.co.uk/content/down...s%20Release.doc (http://www.digitaladlab.co.uk/content/downloads/files/2005%20LED%20Backlisess%20Release.doc)
It contrasts how you adjust white point of the CCFL type displays we're talking about here with the high end LED backlight displays that actually can vary the color temperature of the backlight. In the case of those displays of course, there is not even the concept of a "native white point"
To quote the critical sentence: "On top of this, the LED backlight allows for accurate adjustment of the white point between 5000K and 9300K without loss of brightness, as the LCD layer does not change the white colour temperature as in the case of LCD’s with conventional CCFL backlighting."
Sandy
-
Sandy, I'm aware of the LED advantages of the NEC (I had one) so I agree, that's a totally different beast here. I thought we were talking about CCFL? Is there provisions outside of the native white point that is controllable san's LUTs?
-
Sandy, I'm aware of the LED advantages of the NEC (I had one) so I agree, that's a totally different beast here. I thought we were talking about CCFL? Is there provisions outside of the native white point that is controllable san's LUTs?
To confirm, you don't need a LUT to change the whitepoint of a CCFL LCD display. A LUT is a possible design choice, but just one among many.
I don't know how to put this more clearly. The NEC document confirms the mechanism by which white points are changed in conventional CCFL panels, the Wikipedia article describes underlying technology, so at this point I'm not clear what it is that isn't clear. If you have a specific question around this, please ask it. Or if you have references that suggests that the above isn't the case, please post it.
Sandy
-
"The closer to the linear response you are (in Calibration Tester), the better for quality of your images - minimum changes were done in videoLUT (that's good)."
This is only true if either you don't you care about a critical color match between screen and print or by shear coincidence the native color temp of the monitor matches the color temp of your viewing light. If you do not have a match and you want one, native is not the right choice. Suffering a little banding to get a critical color match is the better choice. Most current calibration solutions do a very good job of smoothing out these banding issues. This "problem" is now considerably out dated and should be cast into the dustbin of history.
And for the record this can only effect the view of your images on your monitor, not their "quality."
-
I don't know if this might be of assistance, but I just want to point something else out: when calibrating with BasICColor, my LP2475W display exhibited some odd behaviour. Although it should not be possible, BasICColour would sometimes actually allow a hardware calibration via DDC - it would treat the HP monitor as it does my Eizo, which does not need the additional step of monitor settings adjustment. (The first page of BasICColour's set-up options would have DDC option available, so I chose it to see what would happen.) It actually worked well for a while (even though it is not supposed to work at all), and then became unstable (weird flickering, random colour changes etc.) and so I reverted to regular non-DDC calibration of the HP. I have no explanation for any of this, alas - so just to say that the interaction between the monitor and the calibration software is worth considering as a possible (though probably unlikely) source of the trouble.
Hi,
I don't have this behaviour... i can only perform a software calibration => Software calibration (Video LUTs).
Maybe it is version related, i use 4.1.9
Kr,
-
No, no really - LCD displays are fundamentally very simple devices - the amount of light transmitted is controlled by the voltage on a pixel. So, simplistically 0 volts = black, 1 Volt = all light transmitted. The 8-bit signal (or more bits in high end displays) just divides that up proportionally, e.g., 0 = 0V, 255= 1 Volt.
When you change the whitepoint of the display, you're changing the voltage that equals 255. So, e.g.,
1. a 9500 whitepoint may mean: Red max voltage = 0.8V, Green max voltage = 1V, Blue max voltage = 1V
1. a 6500 whitepoint may mean: Red max voltage = 0.9V, Green max voltage = 0.9V, Blue max voltage = 1V
Max voltage always equals 255 (in an 8 bit display).
There's no need for a LUT in this process. The display might, depending on how it works internally, use a LUT, but that's not a necessary part of the equation.
Note the above is vastly simplified - e.g., I've left out gamma, black levels, contrast controls, etc. Also most LCD panels (for technical reasons) internally work with 0V = white and max volts = black, etc, but the principle remains the same.
The one thing you need to avoid is setting the display to settings such that it is in saturation, e.g., voltages of above 1V in the example above. How you tell is easy - when you calibrate the panel, check that the transfer curve doesn't have a kink in it at the top or bottom.
Sandy
Hi Sandy,
Is your described theory valid for all panel types? I've found an article that i think confirms your theory:
http://monitordelcd.com/en/monitor/manuten...monitor-de-lcd/ (http://monitordelcd.com/en/monitor/manutencao-em-monitor-de-lcd/)
"Controlling the level of voltage applied between the polarized and can vary the level of light that pass through the display.
...
If all the bits are 0 that subpixel is erased. If some other bits are 0 and 1 are, if the subpixel lights off eight times and very fast so that our eye enxergará a weaker brightness.
Since each subpixel (color) receives 8 bits at a time, he can make 256 levels of brightness"
So to summarize:
The level of voltage between the polarize filters (for each subpixel) determines the color temperature.
The 8 bits define the level of brigthness for R / G / B thus creating a color.
That makes sense checking the options while playing with the white point option.
If i switch between the presets sRGB/6500/9300 there is no change for the RGB values.
sRGB= 5432/5333
6500= 6245
9300= 7300/7100
maximum= 6360
So the maximum preset is when the voltage value is maximum?
But why do i need to change the RGB values in BasICColor to achieve a certain color temperature?
This is a totally other approach to change my white point...
Looks like i can change the white point using a hardware way (presets) and using a software way (RGB values)?
Can i conclude if all i said is true that i have 4 native white points? My 4 presets...
Kr,
-
But why do i need to change the RGB values in BasICColor to achieve a certain color temperature?
Well, there are really two white points that you are dealing with (and nether of them is native white point!):
- the white point of the display on its own (so the color you get when the input to the display is 255,255,255), which is what the previous discussion on this thread was about
- the white point of the system as a whole - that is, the color that you will get when a color managed application asks for lab 1,0,0
Those are very often a bit different in a calibrated system. Reason is that the Display white point is often a bit out - so if set it to say the 6500 preset, it will actually give 6400. When you then calibrate with an I1 or whatever to 6500, the ICC profile that is created "tells" the color management system that to give pure white it should give values of 255, 250, 253 (or whatever).
End result is that while the white point of the monitor is 6400, the color temperature of a white that is displayed by a color managed application will be 6500.
Its best to have the monitor white point and the system white point you're calibrating to as close together as possible. Otherwise, a grey in a non-color-managed part of the screen will look different to a grey in the color managed part, which pretty much defeats the point of calibration. Most color calibration systems let you do this by setting a monitor's custom white point by moving the monitor's RGB controls.
You can see this difference quite easily on most calibrated systems, btw - in Aperture on a Mac, for example, the grey surround is not color managed, while the actual image is. So if you use the DigitalColor Meter tool to look at the RGB values of a white point adjusted grey in the image, and a grey in the surround, there will probably be a slightly different ratio of R to G to B.
Hope that helps.
Sandy
-
As a possible aside, there is one disadvantage to using a white point other than native: the brightness is reduced as the monitor compensates by lowering the R, G or B to compensate. It's neither good or bad, just a trade-off between using the native white point (brighter) or an adjusted one (which is truer white).
-
As a possible aside, there is one disadvantage to using a white point other than native: the brightness is reduced as the monitor compensates by lowering the R, G or B to compensate. It's neither good or bad, just a trade-off between using the native white point (brighter) or an adjusted one (which is truer white).
I can see that, but since nearly every monitor has to be dimmed dramatically to be used correctly, does this really matter? Plenty of headroom to tweak the backlight up slightly to compensate for the very slight dimming resulting from the LUT's adjustment to achieve the target white point.
Personally to get anywhere close to matching color in my viewing station and my monitor, my display profile is set at 6100k ... native and 6500k just aren't even close.