When 3 CRTs are side by side, why does the color seem to be off in red, blue, and green (individually)?
August 14, 2006 9:43 AM
When 3 CRTs are placed next to each other and are showing exactly the same image, why does it often seem that one is redder, one is greener, and one is bluer? This is something we've seen where I work on multiple hardware configurations and in random order, but always one of each in a three-channel configuration.
It seems to be true whether all three CRTs are hooked to one PC, or if all three are hooked to separate PCs. I’m not well versed in color calibration or video signals, and I found tons of great information about calibrating here, but no one seemed to describe this particular issue. I was curious if anyone has seen this and has an explanation for what causes it and how to predict or deal with it. In part I wonder whether it’s simply an optical illusion, or if possibly this is caused by video interference between the monitors when they are in close proximity to each other.
It seems to be true whether all three CRTs are hooked to one PC, or if all three are hooked to separate PCs. I’m not well versed in color calibration or video signals, and I found tons of great information about calibrating here, but no one seemed to describe this particular issue. I was curious if anyone has seen this and has an explanation for what causes it and how to predict or deal with it. In part I wonder whether it’s simply an optical illusion, or if possibly this is caused by video interference between the monitors when they are in close proximity to each other.
Knowing nothing about the actual issue underlying the problem:
If the three monitors are each slightly and uniformly different, there's no other way it could turn out, right? There's only three channels - red, green, and blue. So, one of them has to be perceived as the most of each.
posted by dmd at 10:08 AM on August 14, 2006
If the three monitors are each slightly and uniformly different, there's no other way it could turn out, right? There's only three channels - red, green, and blue. So, one of them has to be perceived as the most of each.
posted by dmd at 10:08 AM on August 14, 2006
Color monitors have three electron guns that must be balanced to get accurate colors. They are calibrated to be "close enough," where the value of "close enough" varies proportionally with the price of the monitor, and the circuitry drifts over time. Different copies will drift in different ways. It's not an illusion, it's not interference, it's just analog circuitry.
LCDs are more consistent, because they don't have a light source that must be balanced in this way. Though the backlights may vary somewhat, it's rarely as noticeable.
posted by kindall at 10:13 AM on August 14, 2006
LCDs are more consistent, because they don't have a light source that must be balanced in this way. Though the backlights may vary somewhat, it's rarely as noticeable.
posted by kindall at 10:13 AM on August 14, 2006
dmd hit it on the head... Because there's three, and three tubes, your eyes will be able to tell which monitor has the most red, blue, green.... If you only have two, it's much more difficult to identify which colors, because one will have more of two colors than the other.
The only way it would not seem this way is if two of the monitors were calibrated identically. Otherwise, even if one of the monitors had way more R+B and one had way more G, the third would likely have an abundance of either R or B, which would make it look like the dominant color.
If you actually measure the monitors, you probably won't get measurements to back up your perceptions (for example, no monitor might actually have a red cast, but the monitor with the most red will look like it has a red cast compared to the others).
So yes, it is mostly optical illusion. There's only one way to deal with it, and that's calibrate each monitor using software.
posted by hatsix at 10:40 AM on August 14, 2006
The only way it would not seem this way is if two of the monitors were calibrated identically. Otherwise, even if one of the monitors had way more R+B and one had way more G, the third would likely have an abundance of either R or B, which would make it look like the dominant color.
If you actually measure the monitors, you probably won't get measurements to back up your perceptions (for example, no monitor might actually have a red cast, but the monitor with the most red will look like it has a red cast compared to the others).
So yes, it is mostly optical illusion. There's only one way to deal with it, and that's calibrate each monitor using software.
posted by hatsix at 10:40 AM on August 14, 2006
I think your scenario needs more descriptive info.
First, what type of monitor(s)? Same monitor make, model, and size? Three electron guns or like a Sony Trinitron, one gun? And how close? Is the image you are interpreting containing motion? To assess static convergence, you need a crosshatch, a grey color field, or a dot hatch.
Second, what is being displayed? Not that it makes too much difference, but a scene containing motion would be awfully hard to assess. A stationary gray scale would be the most appropriate field to use for assessment. All phosphors are involved in producing it and it is a better 'level playing field' to use for the monitor to monitor evaluation that you are suggesting.
Beam deflection is magnetic. The fields extend beyond the cabinet, so they have the possibilty of contributing scan errors to adjacent guns in another unit.
Beam deflection/accuracy is also not the only element in intensity. The signal path goes though buffer amps and also the individual electron guns and phosphor sensitivities may differ, intially and over time. Finally, brightness and contrast settings may be involved. Focus is also an issue, as is electron beam astigmatism. CRT/CRT variation is in there, too. They are unique and that's why they have convergence magnets and compensation for magnetic yoke variations.
Also, they can't display the 'same' image. They can all be connected to the same image source, but the signal has to be split in some way or anthor to go to all three. Your scenario presumes that the same image comes from a common point. There may be dynamics involved if you have one source, one splitter, three cables. Or one source, one distribution amp, three output channels, three cables. Or three different computers, three graphic cards, three cables, all displaying the 'same' image, but with tiny asynchonisms.
This simple question gets complicated quickly.
It would be easy to misadjust all three monitors for blue tint, green or red. There are enough variables that it would be difficult to get an absolutely perfect monitor to monitor match, but I think you could get close enough that all three would be within a range that would be difficult to tell apart.
Bottom line... it's not always the case that one will be bluer, one redder, and one greener.... one of each.
posted by FauxScot at 11:36 AM on August 14, 2006
First, what type of monitor(s)? Same monitor make, model, and size? Three electron guns or like a Sony Trinitron, one gun? And how close? Is the image you are interpreting containing motion? To assess static convergence, you need a crosshatch, a grey color field, or a dot hatch.
Second, what is being displayed? Not that it makes too much difference, but a scene containing motion would be awfully hard to assess. A stationary gray scale would be the most appropriate field to use for assessment. All phosphors are involved in producing it and it is a better 'level playing field' to use for the monitor to monitor evaluation that you are suggesting.
Beam deflection is magnetic. The fields extend beyond the cabinet, so they have the possibilty of contributing scan errors to adjacent guns in another unit.
Beam deflection/accuracy is also not the only element in intensity. The signal path goes though buffer amps and also the individual electron guns and phosphor sensitivities may differ, intially and over time. Finally, brightness and contrast settings may be involved. Focus is also an issue, as is electron beam astigmatism. CRT/CRT variation is in there, too. They are unique and that's why they have convergence magnets and compensation for magnetic yoke variations.
Also, they can't display the 'same' image. They can all be connected to the same image source, but the signal has to be split in some way or anthor to go to all three. Your scenario presumes that the same image comes from a common point. There may be dynamics involved if you have one source, one splitter, three cables. Or one source, one distribution amp, three output channels, three cables. Or three different computers, three graphic cards, three cables, all displaying the 'same' image, but with tiny asynchonisms.
This simple question gets complicated quickly.
It would be easy to misadjust all three monitors for blue tint, green or red. There are enough variables that it would be difficult to get an absolutely perfect monitor to monitor match, but I think you could get close enough that all three would be within a range that would be difficult to tell apart.
Bottom line... it's not always the case that one will be bluer, one redder, and one greener.... one of each.
posted by FauxScot at 11:36 AM on August 14, 2006
Yeah, what Faux said: the thing that makes you discern them as redder, greener, or bluer, is not generally their differences from one another, it's their differences from the neutral grey spot in the middle of the color palette.
If all of them are off by varying degrees in the general direction of red, they're *all* gonna look red.
And yes, 50% grey is what you usually use to check color balance, as well as 20% and 80% (roughly) for black and white balance.
Assuming you don't have a colorimeter/spectrophotometer, in which case you let *it* do the work.
posted by baylink at 12:13 PM on August 14, 2006
If all of them are off by varying degrees in the general direction of red, they're *all* gonna look red.
And yes, 50% grey is what you usually use to check color balance, as well as 20% and 80% (roughly) for black and white balance.
Assuming you don't have a colorimeter/spectrophotometer, in which case you let *it* do the work.
posted by baylink at 12:13 PM on August 14, 2006
If you place three points on a piece of paper, they always describe the corners of a triangle.
posted by Steven C. Den Beste at 2:56 PM on August 14, 2006
posted by Steven C. Den Beste at 2:56 PM on August 14, 2006
Let me try to be more clear about what I mean. Consider this diagram (which I just put together with about 5 minutes effort):
Place three points in the colored section at random. If you do, one of the three will be closest to the red vertex, another to the blue vertex, and the third to the green vertex. That's what you're seeing.
Of course, you could deliberately select yellow, magenta, and cyan, but the odds are strongly against random points doing that.
posted by Steven C. Den Beste at 5:10 PM on August 14, 2006
Place three points in the colored section at random. If you do, one of the three will be closest to the red vertex, another to the blue vertex, and the third to the green vertex. That's what you're seeing.
Of course, you could deliberately select yellow, magenta, and cyan, but the odds are strongly against random points doing that.
posted by Steven C. Den Beste at 5:10 PM on August 14, 2006
These are unbelievably helpful answers. From the simplest to the most complex possibility has been explained, and while Steven's picture is worth a thousand words, FauxScot really nailed it on the head, as well. Thank you all very much for your informative answers - this is much less mysterious and hopefully will be much less frustrating in the future!!
posted by maldrin at 2:37 PM on August 15, 2006
posted by maldrin at 2:37 PM on August 15, 2006
This thread is closed to new comments.
posted by cillit bang at 10:06 AM on August 14, 2006