The purpose of this review is to provide hard data as well as subjective opinion of the DLA-RS20’s performance. The projector was evaluated as a stand alone unit and also when mated with a Lumagen Radiance video processor. This review is broken down into two parts; the first is a review with the Lumagen Radiance installed, and the second part is extended technical analysis of the projector its self.
The JVC RS20 is a nice looking unit even without calibration. The preset THX mode offers the viewer a somewhat accurate image out of box. Also, there are custom user settings which have been suggested in the forums that can be used to improve the imagery of the projector. However, the RS20 is fundamentally limited due to internal design restrictions. Because of this, the RS20 can not be calibrated alone as accurately as an enthusiast may desire. To mitigate the RS20 design limitations a Lumagen Radiance video processor can be used to nearly fully correct the RS20 and get it close to ideal performance in all areas including color gamut and color luminance.
Test Equipment Used:
- PhotoResearch PR-650 Spectroradiometer
- Sencore VP-403 VideoPro Multimedia Video Generator
- ColorFacts 7.5 Professional Suit
- Digital Video Essentials BluRay Disc
Video Equipment Used:
- JVC DLA-RS20
- Lumagen Radiance XE Beta
- Draper M1300
- Denon DVD-3800BDCI
- Toshiba HD-DVD
- Dish ViP722 Satellite Receiver
Initial Impressions and Visual Analysis
I was pleased to be given this job of combining the new JVC DLA-RS20 with the new Lumagen Radiance XE Beta video processor. This video gear is on the cutting edge of what you might find in a typical high end home theater. Also, the fellow for whom I worked has been a regular and courteous customer of mine for several years now. It was a pleasure to see him again and to play with the new equipment. After all, the Radiance XE isn’t even released yet. This was however the first of many RS20’s and RS10’s that I have on my schedule over the next six months.
I am an authorized Lumagen dealer and was fortunate enough to procure the first two Radiance XE beta processors produced. The first went on a Marantz VP11S1, but that’s a story for another day. This RS20 set up was the second I have been able to do with the new XE unit. The Lumagen Radiance XE brings to the table HDMI 1.3 support including all uncompressed audio formats, along with the other excellent features of the earlier Radiance models. Included in the Radiance processors is a fully functional CMS (color management system) with a spectacular installer interface that makes setting the gamut a breeze if you know what you are doing. Special thanks to Jim Peterson of Lumagen for getting me these units early and with very short notice.
Before I even took any measurements I started by viewing the 5th Element on BD with the Denon DVD-3800BDCI connected directly to the RS20. With the projector set to THX mode I thought the image was pretty respectable for a preset out of the box state. To me the image looked good, but there was a definite magenta tone to the entire picture. Flesh tones were decent, but looked on the gray or green side; somewhat natural, but not really believable.
This customer I would consider to have an excellent eye for video quality. I hope he doesn’t mind me saying, but I would also consider him to be one of the more picky customers I have. Fortunately he’s also been keeping up with the RS20 calibration thread on the AVS forum. From the thread, my customer had inserted CMS settings into one of the JVC user modes that have been dubbed the LDVD settings. The LDVD settings seem to be very popular based on what I have had time to read of the thread on AVS. With these LDVD settings in the CMS, we again looked at the 5th Element and compared LDVD settings with the native JVC THX preset. To both our eyes, the LDVD settings were a definite improvement visually over the THX preset. LDVD CMS settings did improved and reduce the magenta nature of the image over THX mode. LDVD CMS settings had also improved flesh tones making them appear more natural when compared to THX mode with less gray green looking skin. However, even with the LDVD settings there were still visual problems. People remained a little gray or green and their skin looked a little under saturated. To me the entire image seemed under saturated looking as well. Also, colors were not consistent depending on the contents of an image.
So for both THX mode and for the LDVD CMS user mode the image lacked the full depth and realism that I would have expected from a projector in the league of the JVC DLA-RS20. There was not a deep 3D quality to the image. It was also clear to me that the 5th Element lacked the realism in colors that I expect from a high end video projector. Like I said, THX was good for out of box, but it is fair to expect far more with proper calibration.
Once we agreed LDVD CMS settings looked best, we viewed other demo material including the Matrix BD, an Eagles concert on HD-DVD, a Cheryl Crow concert stored on the DVR, as well as some other random things. We viewed these with the LDVD settings because we both thought that they provided the best looking image at that point.
Calibration of the DLA-RS20
The next step was connecting my Sencore VP-403 directly to the projector. I set up my PhotoResearch PR-650 and started the ColorFacts software suit.
With the JVC grayscale set to the 6500K preset, I did a full sweep from 10-100 IRE in 10 IRE steps. Unfortunately the data set for these measurements was somehow lost during the calibration process. I do remember precisely that the grayscale stayed between about 5800K and 6100K across the entire luma range and that dE numbers were between 11 and 13 through the entire IRE range. Ideally after calibration a projector should remain very close to D65 and have a dE of five or less across the entire luma range. Obviously there was room for much improvement in gray tracking through calibration.
So after taking preliminary readings, I went into the service mode of the RS20 and adjusted the bias and gain controls for the 6500K preset to make it match D65 as closely as possible. Below are the charts after calibrating D65 in the JVC service mode.
As can be seen, the RS20 tracked to D65 extremely well after calibration. dE numbers fell to below 2 or better across the entire luma range. Also, the temperature tracks very closely to 6500k across the entire luma range with a deviation of just +75/-83 or better. Below is the color tracking chart to reaffirm these findings.
For the purpose of the review section of this article I will now discuss the results with the Lumagen Radiance processor installed. Please check out the technical analysis section of this article for data and discussion regarding the native CMS of the RS20, THX mode analysis, LDVD CMS settings, linearity of the color gamut, and more.
Before adjusting color gamut, I wanted to verify the color decoder. I was pleased to observe all primary and secondary colors seemed to be set correctly in the color decoder domain. Blue looked perfect as did red. Green was very close looking as well, in fact close enough for my liking. I was extremely pleased to see there was no red push.
Errors in color decoding are frustrating because when they exist, even after calibrating the color gamut, the colors will still look wrong. Having correct color decoding is a big deal and will really make for a better looking image. Often times, a projector will have improper color decoding with no adjustments except for the master color and tint controls. The most common color decoding error is when a manufacturer decides to push red. With red push, people’s faces will look red even when the color gamut is perfect.
Fortunately the RS20 has no such errors. With color and tint set to zero in the JVC color decoding is correct. This projector has no red push or any major errors in its color decoding when feeding RGB or 4:2:2 HDMI into the RS20.
It is worth noting that the Lumagen Radiance has both a full CMS system for the color gamut and luma levels, but that the Radiance also has full color decoder controls for master color and tint as well as independent controls for red and green hue and saturation. Had there been any error in the color decoder, I knew before hand that it would have been possible to fix both the color decoding and the gamut through the Radiance VP.
All that being said, some projectors also have non linearity in the color decoder as well as in the color gamut. I am pleased to report that I looked at the color decoder with 50%, 80%, and 100% stimulus and saw a flat color decoder with the JVC CMS settings all nullified. The color decoder is linear with the RS20 in Cinema 2 mode.
Rather than using the RS20 CMS, for this installation the CMS system of a Lumagen Radiance XE video processor was implemented to correct the color gamut. I connected the Lumagen to the RS20, and then plugged my VP-403 pattern generator into an HDMI on the Radiance. I set the Radiance to output component 4:2:2 over HDMI to avoid unnecessary digital conversions back and forth between RGB and YCbCr in the Lumagen Radiance and inside the RS20. Doing this avoids conversion errors and results in the most pristine image at the end.
Earlier in the day I had determined that the Cinema 2 mode in the RS20 user menu was nearly perfectly linear with respect to xyY values of the colors across the entire stimulus range. When using a Radiance to correct the color gamut it is highly advantageous to try and make the projector perform in a linear and predictable manner so that the colors remain true through the entire stimulus range. With Cinema 2 enabled, I also set color, tint, contrast and brightness all to zero in the JVC. I returned the RS20 gamma control to “normal” and grayscale to the 6500K preset that I had setup earlier in the RS20 service menu.
First Attempt With the Lumagen Radiance
Most people prefer colors on a digital projector to be slightly over saturated rather than slightly under saturated. Because of this, I initially targeted xy values using the Radiance that slightly exceeded the Rec 709 gamut at 80% stimulus. I also decided to put the Y luma value slightly on the brighter side of the option. Once I had an 80% color gamut that slightly exceeded the 709 spec I took a look at 100% and 50% stimulus. At 100% stimulus I was elated to observe xyY values remained nearly linear when compared to those prescribed by 80% stimulus. This was looking good already. At 50% stimulus I was again pleased to see pretty decent response as well. With 50% the xy values of all colors had moved slightly more outside the gamut than they had been at 80%, but the degree of non linearity was pretty small and no where near as egregious as when using the RS20 native CMS. The values for Y had also migrated slightly away from ideal for each color, some higher and some lower, but they were all still within a very acceptable margin of error that was much improved over using the JVC CMS. I decided it was again time to evaluate real world material.
We popped the 5th Element back in the BD player and immediately my customer and I were floored with the dramatic improvement. Now instead of the RS20 looking under saturated, it actually seemed a tad over saturated. The magenta tint to the entire image that is obvious with THX and LDVD settings was eliminated, flesh tones were much improved with people no longer looking gray green. The image now had excellent depth and contained 3D realism that was lacking before. Pleased with the results at this point, we continued and looked at the Matrix again and also other demo stuff.
Second Attempt With the Lumagen Radiance
At the end of the test viewing we both decided with the Radiance calibration the image was far superior compared to how it looked before. However, since we also both thought the picture seemed slightly over saturated looking, I decided to redo the gamut and luma calibration again. My goal was to further refine the image accuracy and improve visual quality. The approach I took this iteration was to set 80% stimulus as precisely as possible for all colors’ xyY values. I figured this might work since the image appeared over saturated with real world material which was consistent to the slightly larger than 709 color gamut that I had set up on the first attempt.
I painstakingly adjusted the colors at 80% stimulus nailing down xyY to an insane degree of precision. I then followed up taking readings at 100% and 50% to verify linearity. I was happy to observe even better color tracking with the new CMS settings. Below are the charts and graphs with the final Lumagen Radiance CMS calibration.
RS20 with Radiance XE 100% Stimulus
RS20 with Radiance XE 80% Stimulus
RS20 with Radiance XE 50% Stimulus
We again viewed the 5th Element and the improved Radiance gamut again was easily observable. Over saturation was gone and the image looked spectacular everywhere. Flesh tones looked real, saturation looked perfect, and the image seemed even more transparent and rich then it ever had. We checked out all the demo material again including the Cheryl Crow concert. Before calibration, my customer had noted Crow’s skin looked green in many shots. After the calibration we both agreed that here skin looked real.
Once I felt good about the color gamut I moved on to check black, white, and gamma response. With the JVC contrast and brightness set to zero I saw no clipping of whites and the black level was impressive. With the gamma set to default I got an average response of 2.27 across the luma range. This customer prefers a slightly lower gamma so I decided 2.27 would be well suited for him. I made no further adjustments to brightness, contrast, or gamma in the JVC. Below is the chart for gamma response.
Low IRE Gamma Adjustment
One other adjustment I made was in the Lumagen parametric grayscale menu to enhance the low end gamma response. This customer had previously been setting the JVC brightness to +1 because he tends to like to see details in the very low end of the luma range (even though I tell him he’s not supposed to see everything). The trouble with raising the JVC’s brightness to +1 is that it squashes the gamma response a little, but it also raises the black level hurting the contrast ration and blackness of black. What I did to satisfy the requirement to see dark image details was to put several points in the Radiance parametric grayscale close to zero and to use them to raise the 0.5 – 3 IRE luma a little bit. I personally wouldn’t run my projector this way, but I am here to satisfy each of my customer’s own needs and he likes the response this way. We had actually done the same thing on his earlier CRT and Lumagen Vision processor combination and he was happy with that too. On the Vision line of Lumagen processors tweaking the low end gamma response this way had more deleterious effects than it seemed to with the new Radiance processor. I honestly couldn’t detect any adverse effects from this adjustment and dark details were enhanced quite a bit.
The DLA-RS20 is a fantastic projector and has a nice image right out of the box when set to THX mode. That being said, the quality of the picture can be vastly improved through the installation and calibration of a Lumagen Radiance video processor. The Radiance processor circumvents the internal shortcomings of the RS20 and allows for near perfect color performance. An RS20 and Lumagen Radiance combination can provide the pinnacle of performance longed for by the high end home theater enthusiast. The combo renders realistic and rich looking images that will not leave you wishing for more. Together the Radiance and RS20 is fantastic.
I’d be happy to install and calibrate as many Radiance processors to existing RS10 and RS20 projectors as possible. I regularly fly all over the USA for calibration work and Lumagen installations. I offer special deals when customers buy their Lumagen through me and also hire me to calibrate the system. For high end set ups like this I usually do not charge any fees for travel either.
On another note, I expect to get similar results from both the RS10 and RS20 projectors when they are paired with a Radiance processor. The main features that the RS20 has over the RS10 are the CMS controls and the advanced iris. The CMS on the RS20 has its problems, and the iris doesn’t provide much better black levels whether or not it’s full open of fully closed. For my money, if I hadn’t bought a projector yet and was considering the Radiance, I would probably opt for the RS10 instead. I have several RS10’s on the schedule and plan to write a review once I get my hands on them.
- Craig Rounds
- CIR Engineering
Expanded Technical Analysis
Firstly, here is some info I gathered when playing with the RS20 without the aid of the Lumagen Radiance video processor. I took measurements for THX mode, LDVD CMS, and for the Cinema 1 preset; all across a range of stimulus levels. Before arrival on the job I was told by my customer that others had found the RS20 could not maintain a flat color gamut across a range of stimulus values. What this means is that if one calibrates the RS20 with colors at 80% (or any other stimulus level), when subsequent measurements are taken at varying degrees of stimulus, the color gamut will not remain properly calibrated becoming nonlinear. This is a problem for many projectors being sold today.
THX Mode Preset
For THX mode I started with the Sencore set to 80% stimulus. On the RS20 the THX mode is one of two presets (Cinema 1 is the other) where you can not make changes to the CMS or grayscale with the user controls. So what you get with THX mode is what you get and you can not make any changes to it’s performance in the user mode, and only the grayscale can be adjusted in the service mode. Below are the charts and data I collected for THX mode at 80%.
I was pleased to see that the color gamut was reasonable close to the Rec 709 spec in the xy space and that the luma levels Y were also reasonably close to spec. I then moved forward and took readings for THX mode with 50% stimulus and 100% stimulus with all colors and white. I was pleasantly surprised to observe that all coordinates xyY remained reasonably flat and linear for all stimulus values measured. Yes, in THX mode the color gamut and luma are linear.
This was an early positive sign. However, these readings confirmed that I could not use THX mode in combination with the CMS controls of the Lumagen Radiance. The reason for this is because all colors (with the exception of yellow) were already under saturated and inside the color gamut for Rec 709. Also, the capa Y values for luminance were measuring below Rec 709 spec. When using a CMS system such as the one in the Radiance (gamut correction to the signal before the projector), the colors can only be made less saturated and there is no way to make them more saturated. Furthermore, while it is possible to increase the Y values for each color with an outboard video processor, the technique used to do this can be difficult because it provides mixed results and can negatively impact the gamma response of the projector. So I was already hoping for a better option than starting with the THX mode. I felt it was particularly important to start with a more saturated color response especially because I noted that on real world material the projector seemed under saturated with both THX and LDVD settings.
LDVD CMS Settings
So the next step was to implement the latest LDVD CMS settings. I started by measuring LDVD with 75% stimulus and immediately after looking at the gamut at 100% stimulus. Below are the charts and data from the measurements.
As can be observed from the color gamut charts, there is a fundamental technical problem with the LDVD settings; the color gamut is not anywhere close to linear with respect to the xy coordinates with any color except maybe magenta (and magenta is over saturated). Furthermore, at 75% stimulus the luma values Y were quite a bit out of spec for some colors, though they are quite close for others. Red, green, magenta, and yellow were probably close enough, but not perfect. Blue and cyan’s luma levels were pretty far away from ideal though. I also took luminance readings at 100% stimulus for LDVD, but I did not record them because the values moved in a nonlinear manner. Basically at that point I decided the LDVD settings weren’t a good starting point for a professional calibration. It was however interesting to observe that even though the LDVD settings were not as accurate technically as the THX mode of the JVC, the LDVD settings still looked better on real world images.
Another interesting phenomena with the LDVD settings was that at 85% stimulus or higher the color gamut was basically linear and tracked close to the Rec 709 spec. This may sound good, but what I expect was actually happening was that the colors were clipping above about 85% stimulus. This means that varying saturations above 85% would not be displayed on the screen properly.
In my professional opinion, it seems likely the reason LDVD settings look better than the THX preset is because of the manner of the non linearity with regard to the xy coordinates. At 85%~100% stimulus the xy coordinates place all colors at full saturation or more. As stimulus is decreased, the colors become more and more over saturated and the produced gamut becomes larger. This resulted in a relatively pleasing image. It did surprise me that while the colors were exceeding the Rec 709 gamut that the image still appeared under saturated to my eye. I have observed this phenomena before so I wasn’t surprised, but it is still a bit puzzling to me.
How I applied the Lumagen
I felt good because I already knew that the projector looked nice with the LDVD and THX settings, but my job is to improve the image and hopefully make the projector perform better from a purely technical perspective while also improving real world image quality.
To successfully improve the image with the Lumagen I needed to accomplish three things first. 1) I needed to find some sort of settings in the JVC where the xyY values can remain linear with respect to stimulus levels. 2) I needed to find a mode in the JVC where the colors were all at least equal, or greater than, the Rec 709 gamut with all three coordinates xyY. 3) I wanted to find the closest JVC preset that natively falls near the xyY values for 709. I knew that if I could find a setting in the JVC that met these criteria I could certainly use the Radiance CMS system to make the colors perform reasonably well. Generally speaking, the further out of spec the native gamut, the more difficult it can become to obtain an acceptable gamut even when using the Radiance CMS. Furthermore, I have found that if there are less errors in the native gamut of a projector there is less chance of introducing new nonlinear when applying the Radiance CMS.
Another concern is that I generally prefer to use any controls native to the projector for as many parameters as possible. If I have a choice between using a control on the Lumagen verses using a native control on the projector, I usually choose the control on the projector. Unless I find a reason not to use the projector’s native control. Therefore, I was also hoping that I could find a way to work with the RS20 native CMS system that would allow the projector to remain linear and also improve the color performance.
Following my line of thought I looked at the Cinema 1 JVC preset using 75% stimulus levels. Immediately I realized that this was a disaster. The color gamut was greatly exaggerated in the xy domain and hugely exceeded the Rec 709 standard for all colors. Furthermore, the Y luma values for all colors were about half of what they should be. I did not even bother to record the findings because Cinema 1 is total garbage and useless for realizing accurate imagery. This type of color preset has been applied by many manufacturers of digital projectors and is highly subjective (and looks terrible). Under some situations the colors actually appear fairly accurate, but under most circumstances the colors look highly over saturated. I realized that I did not want to use Cinema 1 in combination with the Lumagen Radiance either because the Y values were just too low and I wanted to avoid utilizing the CMS in the Lumagen to increase Y if at all possible.
Continuing on I tried the JVC Cinema 2 mode and turned the CMS off in the JVC. I set the color temp to the earlier calibrated 6500K preset, the gamma to default, and the color, brightness, and contrast all to zero. My goal here was to try and nullify any non linearity the JVC might be causing to the color response with the user controls. I started taking readings at 75% stimulus and was pleased to see that all six colors exceeded their specified coordinates on the 709 color gamut. I was also happy to observe all colors had more luminance than specified by Rec 709. I moved on and took readings at 100% and 50% stimulus and was elated to discover that the xyY coordinates all stayed linear with very little error! With the CMS turned off, Cinema 2 is linear.
I new at this point that even if the CMS system in the JVC didn’t work at all, that I could most likely make the projector look fantastic with the Radiance XE CMS system. The Cinema 2 setting had the things I needed; xy color coordinates that exceeded the 709 gamut, Y luma values that were close to or higher than the specification, and linearity across varying stimulus levels.
User Mode 1
The next thing I did was explore the user mode 1 with the same settings in the JVC. Color, tint, contrast, and brightness were all set to zero. Grayscale was set to the 6500K calibrated preset, gamma at default, and CMS turned off. I observed that user mode 1 (with identical settings to Cinema 2) performed virtually identically. So now it was a matter of testing the RS20 native CMS system. With the CMS turned off the user mode 1 is linear.
Evaluating the DLA-RS20 CMS System
So now I wanted to try and use the CMS system of the JVC and improve the color gamut. Immediately upon looking at the tables for the CMS in the RS20 I was disappointed. There are three controls for each of the six colors in the RS20 CMS. They are saturation, brightness, and hue… and these are not the adjustments that I would hope to find in a CMS.
To exemplify the problem with this nomenclature let’s think about the color red. If I am looking at red and it exceeds the color gamut, what I would like to do is be able to add both green and blue to it. By adding the other two primaries individually to the red, the red could then be forced closer to the reference point with the addition of green and blue. What I would also like to see in a CMS is a third control to remove red in order to decrease the luminance. A CMS system can not take any red out of red, but if you have a control to decrease the red it will actually be able to reduce the Y luminance value for red. So a good CMS system will have controls for each of the six colors to add or subtract primaries from each color.
The RS20 has the luma level control labeled brightness (subtraction of color being adjusted), but it also has these two other oddly named controls called saturation and hue. In our example for red, what I would expect the saturation control to do would be for it to add green and blue to red simultaneously. This would bring red closer to the reference point. But what then happens when the projector doesn’t require the predetermined amounts of green and blue added that the projector CMS mandates due to the nature of the control. Well, I guessed that the hue control could then be adjusted to manipulate the proportions of green and blue. I was skeptical of this approach because as the stimulus level changes, the relationship of red to blue required for correction will change and result in non linearity.
However, also think about these controls for not just a primary, but also for the secondary colors yellow, magenta, and cyan. Yellow, magenta, and cyan are actually made by combining all three primaries in a digital projector. Thus, with the secondary colors the JVC saturation controls would have to work using the tangential primary color across from the subject secondary on the color gamut as well as the adjacent colors. The hue control would have to add and simultaneously subtract the other primaries, and the brightness control would have to adjust luma levels using all three primaries simultaneously. But with a secondary color, the mixing of all three primaries defines the luma level… this is a disaster for a CMS that is laid out so poorly. Why not have controls for red, green, and blue luma levels for each secondary and then everything would work fine!
Poorly Implemented CMS
I looked at red first in Cinema 2 with gamma set to default, grayscale set to 6500K, and color, tint, brightness, contrast all set to zero. I wanted to test the controls provided with the RS20 CMS to see what they did, and to try and determine which controls might introduce non linearity issues with respect to the xyY and stimulus levels. I already had the knowledge that the gamut is linear with the CMS turned off (or with the CMS controls at zero positions for all colors). But I also had the knowledge through the analysis of the LDVD CMS settings that making changes to the JVC CMS will likely introduce non linearity to the response. I wanted to use each control provided by the RS20 individually to determine which might work well and which might adversely affect performance.
JVC CMS “Saturation” Control
The first control I used when looking at red was the one labeled “saturation.” As I predicted, the saturation control did add both blue and green to the red moving it closer to the target xy coordinate. I reduced the “saturation” control until red was almost on top of the required xy coordinates. Since there is no separate blue and green control, it was not possible to hit the xy target precisely with just the “saturation” control.
Once red saturation was roughed in, a new problem had manifest that I predicted. The trouble was that after adding green and blue to the red (saturation control), that the Y luma value of red had increased. This is expected and a normal response for a projector; the reason for it is easily explained. Essentially, with the saturation control given in the JVC CMS, you can not subtract any red from the red; you can only add blue and green to red. When you apply additional blue and green to the red (decreasing saturation), you are adding more colors (adding green light and blue light), so the luma response naturally must also increase with the addition of more light (decreasing saturation). This is why I was now seeing a higher Y value for red after decreasing saturation and getting close to xy.
So with the saturation of red lowered to somewhere around -25 in the JVC CMS I had a good xy value for red at 75% stimulus. Before making any other adjustments to the CMS, I wanted to verify that the xyY position of red remained linear for 50% and 100% stimulus. I am happy to report red was linear and remained at the same xy position while maintaining correct Y across the stimulus levels. Wow I thought, these reports of non linearity may be exaggerated and the CMS in the JVC might actually work pretty well.
JVC CMS “Brightness” Control
After setting the red “saturation,” I went on in an attempt to use the “brightness” control for the red CMS. I expected this control to decrease the now exaggerated Y luma value. I adjusted”bigness” at 75% and got nearly perfect results so I decided to check the linearity of the red xyY coordinates now. Here is where things got ugly.
Much to my disappointment, the red xy position on the gamut was no longer at all linear. Furthermore, the luma value did not remain consistent across the range of stimulus. I found that with increased stimulus to 100% the red primary was well inside the xy portion of the gamut (under saturated), and that luma was no longer on target. If I decreased the stimulus to 50% red would become highly over saturated on xy moving well outside the gamut. Bummer, but I had predicted this result after my earlier evaluation of the LDVD presets.
JVC CMS “Hue” Control
I then returned the “brightness” control for red back to zero and made adjustments to the “hue” of red with the JVC CMS. At 75% stimulus I was able to hit the xy coordinates exactly using both the “saturation” and “hue” controls in the red CMS. I proceeded to look at 50% and 100% with these settings while letting Y remain uncorrected and too bright. Again, much to my disappointment the xy coordinates were all over the place when the stimulus level was increased or decreased. This meant that the hue control in the JVC CMS also seemed to cause non linearity to the xyY.
So what I had discovered so far is that for red, the JVC CMS was capable of adding green and blue (decreasing saturation) while maintaining linearity of the xyY coordinates.”Brightness” and “hue” could not be used to obtain ideal results because they exhibit non linearity with respect to stimulus level. So basically only one control “saturation” worked in the CMS without adding more error and creating non linearity. But knowing that I was going to add the Lumagen Radiance to the video chain, I was pretty happy to know that I had at least one working CMS control inside the JVC. Had I not been planning the installation of the Radiance I would have begun on a strategy on a compromise already.
I next attempted the same experiments on green that I had done on red with the RS20 CMS system. Unfortunately, the results on green were even less satisfactory. Both the green “brightness” and “hue” controls caused severe nonlinearly to the xyY coordinates across a range of stimulus levels just as the same controls had done to red. However, the CMS for green was even less effective than it was on red because the “saturation” control is wholly insufficient. What I found with green unlike red, was that even with the green saturation set to the lowest level in the JVC CMS, green still far exceeded the xy values for Rec 709 so was still far outside the color gamut. By adjusting the green “hue” control I could add red to green and pull it closer to the target xy values at 75% stimulus. But as explained, with the use of the “hue” control comes the introduction of non linearity to the xyY values across the stimulus range. At this point I was fed up with the JVC CMS and decided to rethink things.
Generally I like to use as many of the native controls on board a projector for calibration as possible. However, I already knew that the CMS was mostly ineffective for green. The CMS for red had some positive benefit, but I was concerned that using the JVC CMS for some colors and not all could result in weirdness when the CMS from the Radiance was applied. So for this installation I decided to entirely scrap the use of the JVC CMS and attempt to correct the xyY values exclusively with the Radiance VP.
What I Would Try on an RS20 Without a Radiance
If I had not had the Radiance for this job I would have most likely first tried using just the “saturation” controls in the RS20 CMS to get the xy values as close as possible for each color. Next I would have tried using the master color control in the regular JVC user mode in an attempt to decrease the Y values for all the colors. Using the color control for this is less than ideal because it will effect all colors simultaneously necessitating the need for a compromise in Y values among all the colors. Also, the color control works in the color decoder domain and is not really meant to be a CMS control. Often using the color control to adjust the luminance of the colors will result in color weirdness across stimulus levels.
So that’s all for now. I will try and keep up as much as possible and answer questions on the RS20 calibration thread at the AVS and HomeTheaterSPoT forums.
- Craig Rounds
- CIR Engineering