What does this chart tell you?

When you look at a projector or display review, you’ll often see a picture of the common CIE diagram accompanying it.  Many people will look at the diagram and say “That looks good”, but is that diagram really telling you anything useful?  While looking at a review of the new Panasonic AE8000U projector at ProjectorCentral.com, I came across this chart in their performance section:

Image used from ProjectorCentral.com Panasonic AE8000 Review.

Looking at this you would think that the AE8000U performs pretty well.  The targets and measured values are close together, so that means you have good performance, right?  There are a few reasons why this isn’t the case, and why without the actual numbers you can’t tell how good this projector is.

First, this chart has very different scales depending on where you are on the chart.  Being 10 pixels off on this chart for blue is very different than being 10 pixels off from green or red.  There is a better chart available, the 1976 CIE uv (CIELUV) chart, that is spatially uniform so that being 10 pixels off in one area of the chart is the same as 10 pixels off in another area of the chart.  Many people find this confusing, unsure why the chart looks different than the 1931 CIE xy chart, so reviewers have been slow to adopt it.  Given how it is more accurate and useful, I feel it would be beneficial for everyone to switch to it at this point.

Second, color is made up of three components: Hue, Saturation, and Luminance.  On the CIE chart pictured above, the way to understand this is that hue rotates the color point around the white point, or center, of the chart while saturation controls the distance of the point from the center.  Reducing saturation will move the point closer to the white point of the chart, and increasing the saturation of a color will move it closer to that color point on the chart.  So where does luminance come into play?  Look at the samples below.

Do those colors look different?  They should, as they all have different amounts of luminance from each other.  However, they all have the exact same xy coordinates on this chart, so they would all be identical on there.  Of course, compared to the ideal target, one of these will be good and the others will have a high dE value because they have the incorrect luminance value.  If you look at a program like CalMAN, next to the CIE xy or uv chart you will often see a bar graph of the luminances of the primary and secondary colors.  This makes it easier to know if they are close to accurate, or far off.  If your red luminance is too high, everyone will appear sunburned, but you’d never be able to tell this from the chart above.

Third, this chart provides no error numbers, like a DeltaE value using the 1994 or 2000 formulas.  As I mentioned earlier, you can’t tell how close a value is here by looking at the chart.  Similarly you can’t tell what the dE value is going to be by this chart.  The lower the dE value the more accurate the color, though humans are more sensitive to errors in green and less sensitive to errors in blue.  We might look and this chart and say that it looks like green is closer to the target than blue, but if green has a dE of 2.0 and blue has a dE of 1.0, then we would be better off to adjust green to be more accurate, even at the expense of blue.

There are many more things we can’t tell from this chart at all.  We have no idea if the color temperature is close to 6500K, or 5500K, or 8000K, as that isn’t mentioned anywhere.  We also don’t know if the gamma is accurate, or if it might be high or low, or an S-curve.  Really the only thing you can accurately tell from this chart would be if something were really wrong.  If the xy coordinates were really far off, and the targets didn’t even come close to the measured values, you could say that something is wrong, but you couldn’t really say what.

Charts and graphs in a review have their place, but just throwing in a chart without the appropriate numbers and data to go along with it doesn’t really tell you anything.  I’m not picking on Projector Central, as there are many review sites that put up these charts without the corresponding data.  I also know that if I put up the raw xyY values for the primary and secondary colors, that will mean a lot to very technical readers but won’t mean anything to the more casual reader of these reviews.

It’s important to try to strike a balance between having all the appropriate data to make a relevant decision about something, and to making that data approachable and understandable to the casual reader.  Just presenting this chart really doesn’t help either of those groups out.  Casual readers think they know something, but they really don’t know because of the lack of details, and technical readers aren’t getting a single hard number to help them interpret the data.  Finding a way to serve both groups equally is important, and something we all should try to do in our reviews.

, , , , , , , ,