- Written by Chris Heinonen
- Published on 19 December 2012
Sony VPL-HW50ES Projector On The Bench
There are a number of preset modes on the Sony VPL-HW50ES projector to test, but two of them will be of the most interest to readers. The Reference mode tested the best and is used for all of the charts here. The BrightTV mode was the brightest mode, with almost 1700 lumens at its default settings. I did find that turning the Contrast down from Max to 85 will remove clipping in red, which leads to horrible bluish whites, but had a negligible effect on light output by eye while producing a much better image.
Using the Reference mode, the Sony VPL-HW50ES was relatively accurate on my Solar HD screen. With the iris set to manual mode and half open, it still produced 19.5 fL on a 96", 1.3 gain screen. It delivered a 5500:1 contrast ratio here with the iris on manual and the lamp on low. The grayscale was a bit blue with an average CCT of 7168 and an average dE2000 of 3.8. The gamma was 2.24 on average, but wasn't flat as can be seen in the graph.
The color gamut was good, if a bit under-saturated at the extremes. The luminance error was low at 0.57 dE94_L on average, and an overall average dE2000 for colors of 2.57. When we look at the color checker averages, we see an average dE2000 of 2.89 that is pretty good for a display with no adjustments done.
Finally looking at the saturations for the Sony, we do see that colors are a bit under-saturated across the board, but not really badly. The average dE2000 for all the saturations is 2.81, very close to the average dE2000 we found in the color checker chart as well. Overall in Reference mode, the Sony did a good job on a neutral screen without any adjustments. With gray or other screens, you might find more of a color shift with the default settings.
For calibration I utilized the user mode, targeting 16 fL for peak light output. On a 96" 1.3 gain Solar HD screen, I got 15.98 fL of light with the lamp on low and the iris manually set to 50. This allows for a lot of room to open the iris and switch the lamp to high as it ages without the screen dimming. The black level was 0.0029 fL, giving a contrast ratio of 5550:1. With the dynamic iris engaged, the black level was too low to measure.
Using the two-point grayscale, the color balance was nearly perfect, though overall it did run a little bit warm with a CCT of 6419 on average. The gamma was almost a perfect 2.2 other than a spike at 5%. If I had measured at every 10% instead of 5% as most people do it would look perfect, but overall the gamma was very good.
The overall dE average was 0.75 across the whole grayscale, with the worst values at 10% and 100% only being around 1.75. Looking at the grayscale it will be totally neutral with gamma that is near ideal.
On the CIEuv diagram, the points are very close, though the red is a bit under-saturated at the maximum. I was able to correct this using the Real Color Processing, but that caused very bad results on the red values below 100%. Green suffers from this same issue. Instead I found I got much better values by targeting 75% instead of 100% saturation, and limiting my adjustments in the RCP module to +/-5 instead of the maximum of 30. When done this way this gamut looks slightly off, but the saturations and color checker charts later will look much better.
Even with these issues, the maximum color dE at 100% saturation was only 3 for Red, and close to that for green. The average dE for fully saturated colors is only OK at 1.747. However, we can look at the color saturations and see a different story.
Now we see that the largest errors happen at 100%, and everything below 100% is much more accurate. As values below 100% are much more common than 100%, this leads to some charts that look worse, but real world content that looks much better. It's very possible to have perfect regular charts and an image that looks horrible, so these extra charts provide a fuller picture of what a display can do. The average dE for all these saturations was only 1.3939, so the error overall is not visible.
The color checker chart contains common colors from nature, and has shades that you would not target in a CMS system, so you can't fake this data as easily and still have a bad picture. Our average error for the Color Checker chart is 1.4194 dE2000, which is very good. The largest issue is in a couple shades of blue, which is the color we want the most error in, so this is very good to see.
Pre-calibration, the Sony VPL-HW50ES is good when set to Reference mode, but after a calibration it can really excel. The biggest problem is that the RCP control lacks enough bit-depth to allow for large changes, and you can get better results with the combination of using 75% saturation targets and keeping the RCP changes at 5 or below. Sony did a very good job with the 50ES providing a very nice, very accurate image.