CIR-Engineering
09-11-08, 11:00 AM
http://test
Hey guys,
I've gotten to calibrate several of the xx833 Diamond sets so far. I must admit that overall I am really impressed with these. Once they are calibrated the colors can be made extremely accurate and the black level and gamma response is quite good. I didn't think these sets would perform all that great, but I was wrong ;)
Test Equipment Used:
PhotoResearch PR-650 Spectroradiometer
Sencore VP-403 Calibration Generator
ColorFacts 7.0 Pro Software Suit
Evaluation Material Viewed:
Samsung BluRay
-Casino Royal
-5th Element
My first impression of the xx833 when I walked in the door was a typical one; way too bright, inaccurate colors, far too much blue in the greyscale, and a poor gamma response. I figured I could probably fix the greyscale, come up with an acceptable gamma curve, and make the color decoder look believable. However, I underestimated the quality of this chassis and was far more impressed with every aspect once I got to work. The colors came out dead on, the gamma looks near perfect, and the greyscale is excellent.
The first thing I did was take pre-calibration readings of greyscale, gamma, black level, white level, and chromacity (color gamut). After that, I moved into the service menu and started to experiment... and that's when my expectations began to increase.
Here is the precalibration gamma response. As you can observe, the precal gamma does not follow the correct curve nor does it meet the industry standard of falling between 2.2 - 2.5. The out of box gamma is far too low. This results in a washed out image that lacks depth, lacks the three dimensional quality of HD, and lacks realism.
http://www.cir-engineering.com/bin/mitsxx833/precalgamma.jpg
As always with digital projection calibrations I started by roughly dialing in black and white level and then to establish a baseline gamma. On many past Mits DLP's the gamma was really tough to adjust because of the poor implementation of their mechanical iris. With this chassis, the iris must remain on, but it is a much better design than what Mits had on earlier models. While I could not hit the target minimum average gamma of 2.2, I was still able to improve the situation dramatically by increasing the average from 1.83 to an impressive 2.13. Furthermore, the graph reveals that the gamma response actually follows the 2.2 industry standard through the entire range except for around 90 IRE. This little blip is what causes the average to be below the 2.2 goal and is actually less significant than if the gamma response were in error across the board. In short, the gamma is actually better than the average would suggest.
http://www.cir-engineering.com/bin/mitsxx833/postcalgamma.jpg
Next I proceeded to dial in saturation and chromacity. Out of the box, the 833 displays the typical blown out gamut that can be observed on most digital displays, as well as several other color errors. As can be seen from the precal gamut plot, green is excessively green, yellow is excessively yellow, and red is excessively red. You may observe this in real world images as red, yellow and green objects glow or seem almost florescent. For example, green trees or grass may seem to be glowing rather than simply being green. Same with red, but people's faces probably look far too red as well. Phase (tint) for magenta and cyan was way way off resulting in inaccuracies for these color components. This takes away from accurate reproduction of all color images. Most manufacturers of digital displays do not adhere to the Rec 709 gamut and blatantly abuse it. However, Mitsubishi is typically the worst offender of this and the xx833 is no exception. The colors out of box were terrible! After calibration however, I was very happy to see that I was able to get the colors nearly perfect :) Magenta and cyan tint came right in line, and the chromacity of red, yellow and green was reduced to near perfection. All this was accomplished with only a small sacrifice to the position of blue.
http://www.cir-engineering.com/bin/mitsxx833/cieprecal.jpg http://www.cir-engineering.com/bin/mitsxx833/ciepostcal.jpg
Furthermore, color gamut is only one aspect that needs attention with respect to color. The luminance (brightness) of the colors also needs to be properly adjusted and not all displays have this ability. Below is a table of the luminace of each color with respect to an ideal luminance. Readings for this table were taken at 75%. The left column is the calculated ideal luminance with respect to the actual white luminance measured. The second column is the measured luminance for each color as seen by my PR-650.
http://www.cir-engineering.com/bin/mitsxx833/colorluma.jpg
It has been stated that the Mits xx833 models are not capable of accurately displaying proper luminance for all colors. As can be observed this is not correct. The data gathered shows that these sets do have excellent color luma response when compared to the ideal luma level.
After color, I moved to greyscale the results of which can be seen below.
http://www.cir-engineering.com/bin/mitsxx833/greytrack.jpg
As is common these days, the Mits 833 was extremely cool (blue) out of the box with an average color temperature of over 13,500 K. At some points on the output the grey was actually reading over 17,000K. This level of inaccuracy is just terrible and will result in a blue tone to everything and colors will be very washed out and inaccurate looking. Also, different color temperatures with respect to luminace will result in black and white images having varying degrees of "color" depending on the output. At lower light levels black and white will look more blue than at higher light levels. The chart only goes to 12,000K so what you are seeing in the precal chart is that the greyscale is literally "off the chart."
The good news is that after calibration, the greyscale came right into line with an average temp of right around 6,500 K, just where it should be.
http://www.cir-engineering.com/bin/mitsxx833/cordinates.jpg
And above are the results for delta "E" before and after calibration. DE is essentially a more precise way to measure error (RMS error), and in this case error to grey tracking.
As can be observed, before calibration the DE was running around 65 across the entire luminance range. DE of 10 or less is considered ideal for a digital display so obviously 65 is way too high. But after calibration the DE came out really well averaging less than 3 at all luminance levels. You may observe that at 10 IRE and 20 IRE IRE the DE was measured at 35 and 15 respectively, but the windows looked grey to me and there was a little ambient light in the room. At 10 and 20 IRE it can be difficult to get a good reading with any ambient light because those luma levels are pretty darn dim. Therefore, the DE readings for 10 and 20 IRE should be taken with a grain of salt because they probably aren't accurate anyway.
And finally for greyscale here are the plots for color tracking. As can be seen, out of box there was so much blue that it was off the chart. There was also not enough red in the grey. After calibration all three colors combined brilliantly to make D65 grey.
http://www.cir-engineering.com/bin/mitsxx833/colortrack.jpg
I went back and touched up grayscale, black level, white level, gamma, and color and found that there was a lot of interactions between settings. This required several more calibration iterations and was time consuming... but in the end it was worth it.
Finally we got to view some material. BluRay and HD cable looked phenomenal. Colors were realistic and very accurate. Black level was excellent, about as good as I've seen on any digital display. I did not watch any 480i as the cable box on hand could only output one resolution and was set to 1080i.
I have watched 5th Element on every display I've calibrated for the past seven plus years and I will say it looked great on this Mits. Parts of this movie are like test patterns for me because I have seen them on so many units. This digital really looks fantastic. Casino Royal is a new addition to my reference material and I have seen it on a hundred or so units. And again, the Mits kicked!
So all in all I think Mits is moving in the right direction. These DLP units with color management are great and I'd be pleased to own one myself.
One final note, all displays require unique service and user menu settings for accurate imagery. It is a well established fact that simply coping settings from one unit to another will not improve accuracy. Therefore, any emails or inquiries asking for service menu settings will be deleted and not read.
Happy viewing :)
craigr
Hey guys,
I've gotten to calibrate several of the xx833 Diamond sets so far. I must admit that overall I am really impressed with these. Once they are calibrated the colors can be made extremely accurate and the black level and gamma response is quite good. I didn't think these sets would perform all that great, but I was wrong ;)
Test Equipment Used:
PhotoResearch PR-650 Spectroradiometer
Sencore VP-403 Calibration Generator
ColorFacts 7.0 Pro Software Suit
Evaluation Material Viewed:
Samsung BluRay
-Casino Royal
-5th Element
My first impression of the xx833 when I walked in the door was a typical one; way too bright, inaccurate colors, far too much blue in the greyscale, and a poor gamma response. I figured I could probably fix the greyscale, come up with an acceptable gamma curve, and make the color decoder look believable. However, I underestimated the quality of this chassis and was far more impressed with every aspect once I got to work. The colors came out dead on, the gamma looks near perfect, and the greyscale is excellent.
The first thing I did was take pre-calibration readings of greyscale, gamma, black level, white level, and chromacity (color gamut). After that, I moved into the service menu and started to experiment... and that's when my expectations began to increase.
Here is the precalibration gamma response. As you can observe, the precal gamma does not follow the correct curve nor does it meet the industry standard of falling between 2.2 - 2.5. The out of box gamma is far too low. This results in a washed out image that lacks depth, lacks the three dimensional quality of HD, and lacks realism.
http://www.cir-engineering.com/bin/mitsxx833/precalgamma.jpg
As always with digital projection calibrations I started by roughly dialing in black and white level and then to establish a baseline gamma. On many past Mits DLP's the gamma was really tough to adjust because of the poor implementation of their mechanical iris. With this chassis, the iris must remain on, but it is a much better design than what Mits had on earlier models. While I could not hit the target minimum average gamma of 2.2, I was still able to improve the situation dramatically by increasing the average from 1.83 to an impressive 2.13. Furthermore, the graph reveals that the gamma response actually follows the 2.2 industry standard through the entire range except for around 90 IRE. This little blip is what causes the average to be below the 2.2 goal and is actually less significant than if the gamma response were in error across the board. In short, the gamma is actually better than the average would suggest.
http://www.cir-engineering.com/bin/mitsxx833/postcalgamma.jpg
Next I proceeded to dial in saturation and chromacity. Out of the box, the 833 displays the typical blown out gamut that can be observed on most digital displays, as well as several other color errors. As can be seen from the precal gamut plot, green is excessively green, yellow is excessively yellow, and red is excessively red. You may observe this in real world images as red, yellow and green objects glow or seem almost florescent. For example, green trees or grass may seem to be glowing rather than simply being green. Same with red, but people's faces probably look far too red as well. Phase (tint) for magenta and cyan was way way off resulting in inaccuracies for these color components. This takes away from accurate reproduction of all color images. Most manufacturers of digital displays do not adhere to the Rec 709 gamut and blatantly abuse it. However, Mitsubishi is typically the worst offender of this and the xx833 is no exception. The colors out of box were terrible! After calibration however, I was very happy to see that I was able to get the colors nearly perfect :) Magenta and cyan tint came right in line, and the chromacity of red, yellow and green was reduced to near perfection. All this was accomplished with only a small sacrifice to the position of blue.
http://www.cir-engineering.com/bin/mitsxx833/cieprecal.jpg http://www.cir-engineering.com/bin/mitsxx833/ciepostcal.jpg
Furthermore, color gamut is only one aspect that needs attention with respect to color. The luminance (brightness) of the colors also needs to be properly adjusted and not all displays have this ability. Below is a table of the luminace of each color with respect to an ideal luminance. Readings for this table were taken at 75%. The left column is the calculated ideal luminance with respect to the actual white luminance measured. The second column is the measured luminance for each color as seen by my PR-650.
http://www.cir-engineering.com/bin/mitsxx833/colorluma.jpg
It has been stated that the Mits xx833 models are not capable of accurately displaying proper luminance for all colors. As can be observed this is not correct. The data gathered shows that these sets do have excellent color luma response when compared to the ideal luma level.
After color, I moved to greyscale the results of which can be seen below.
http://www.cir-engineering.com/bin/mitsxx833/greytrack.jpg
As is common these days, the Mits 833 was extremely cool (blue) out of the box with an average color temperature of over 13,500 K. At some points on the output the grey was actually reading over 17,000K. This level of inaccuracy is just terrible and will result in a blue tone to everything and colors will be very washed out and inaccurate looking. Also, different color temperatures with respect to luminace will result in black and white images having varying degrees of "color" depending on the output. At lower light levels black and white will look more blue than at higher light levels. The chart only goes to 12,000K so what you are seeing in the precal chart is that the greyscale is literally "off the chart."
The good news is that after calibration, the greyscale came right into line with an average temp of right around 6,500 K, just where it should be.
http://www.cir-engineering.com/bin/mitsxx833/cordinates.jpg
And above are the results for delta "E" before and after calibration. DE is essentially a more precise way to measure error (RMS error), and in this case error to grey tracking.
As can be observed, before calibration the DE was running around 65 across the entire luminance range. DE of 10 or less is considered ideal for a digital display so obviously 65 is way too high. But after calibration the DE came out really well averaging less than 3 at all luminance levels. You may observe that at 10 IRE and 20 IRE IRE the DE was measured at 35 and 15 respectively, but the windows looked grey to me and there was a little ambient light in the room. At 10 and 20 IRE it can be difficult to get a good reading with any ambient light because those luma levels are pretty darn dim. Therefore, the DE readings for 10 and 20 IRE should be taken with a grain of salt because they probably aren't accurate anyway.
And finally for greyscale here are the plots for color tracking. As can be seen, out of box there was so much blue that it was off the chart. There was also not enough red in the grey. After calibration all three colors combined brilliantly to make D65 grey.
http://www.cir-engineering.com/bin/mitsxx833/colortrack.jpg
I went back and touched up grayscale, black level, white level, gamma, and color and found that there was a lot of interactions between settings. This required several more calibration iterations and was time consuming... but in the end it was worth it.
Finally we got to view some material. BluRay and HD cable looked phenomenal. Colors were realistic and very accurate. Black level was excellent, about as good as I've seen on any digital display. I did not watch any 480i as the cable box on hand could only output one resolution and was set to 1080i.
I have watched 5th Element on every display I've calibrated for the past seven plus years and I will say it looked great on this Mits. Parts of this movie are like test patterns for me because I have seen them on so many units. This digital really looks fantastic. Casino Royal is a new addition to my reference material and I have seen it on a hundred or so units. And again, the Mits kicked!
So all in all I think Mits is moving in the right direction. These DLP units with color management are great and I'd be pleased to own one myself.
One final note, all displays require unique service and user menu settings for accurate imagery. It is a well established fact that simply coping settings from one unit to another will not improve accuracy. Therefore, any emails or inquiries asking for service menu settings will be deleted and not read.
Happy viewing :)
craigr