This blog is a continuation of my previous blog (HERE) of December 14, 2011 on “3D Crosstalk/Ghosting – Part 1” and also is a follow-up to my earlier blogs discussing screens for 3D projection (i.e., blogs from August 19th, August 31st, and December 1st. For the continuing discussion from my most recent blog on 3D Crosstalk, I have now completed measurements of the 3D crosstalk level from my JVC DLA-RS40 projector and for the further discussion on screens for 3D projection I have taken a quick look at sample “silver screen” materials from both Stewart and Da-lite that are being marketed for use with passive 3D projection systems, that use polarization for their 3D separation, as well as suitable to 2D projection.
3D Crosstalk/Ghosting Resultsfor my JVC DLA-RS40 Projector
In Part 1 of my blog (December 14, 2011) I discussed the potential sources for 3D Crosstalk/Ghosting with 3D projection systems, including both active and passive 3D projection technologies. In order to test for 3D crosstalk I have created a Blu-ray 3D test disc that includes a number of 3D related test patterns. This disc provides full resolution 1080p/24 test patterns and is encoded using the Blu-ray Disc standard 3D “frame packing” format. My home theater projector is a one year old JVC model DLA-RS40. Note that the RS40 has just very recently been replaced in JVC product line by the new DLA-RS45 which has been reported by a few new owners to have similar 3D performance to the RS40 (in one case when both were viewed side-by-side). Of the 3D test patterns I have created the one most appropriate for testing of 3D crosstalk is where the left image is solid black while the right image has a 100% white vertical bar on a black background (see photos below). This generally represents the most demanding situation for 3D and with most 3D projection systems will be the most likely to produce visible 3D crosstalk/ghosting. Any visible crosstalk will result in a dim (very dark grey) vertical bar being visible through the left lens of the 3D active shutter glasses while if there were no crosstalk the image on the screen would appear fully black when viewed through the left lens of the 3D active shutter glasses.
To evaluate the level of crosstalk present when using my DLA-RS40 along with my JVC 3D active shutter glasses I first allowed the projector to run for approximately 60 minutes in high lamp mode. This was done since it has been observed that the JVC 3D projectors, with their DILA (LCoS) micro-display panels produce far less crosstalk after they heat up to full operating temperature. This typically requires the projector to be operated in high lamp mode for approximately 15 to 20 minutes after which the level of 3D crosstalk appears to be stable. This characteristic appears to be related to how micro displays using LCD or LCoS technology have a longer response time when cold than when hot and in the case of the LCoS panels used in the JVC projectors it appears they have too long a response time to adequately support 3D until they are adequately heated up.
Also as background, when the 3D input signal is in 1080p/24 frame packing signal format, as normal for Blu-ray 3D movies, the JVC 3D projectors displays the 3D at 96 Hz., with each the right and left eyes being presented video with a 48 Hz. refresh rate. When displaying the above described test pattern and observing the projected image I could easily see a modest level of 3D crosstalk with a very dark grey vertical bar visible to my left eye and the full intensity vertical white bar visible to the right eye. I then proceeded to take measurements of the light level for the vertical bar as viewed through the right lens vs. the left lens of the JVC 3D glasses. To perform this measurement I used two different meters/sensors and two different measurement techniques (i.e., directly measuring the light from the projector vs. measuring from the screen). The crosstalk results varied between approximately 0.6% and 0.8% with a mean value of crosstalk measured at approximately 0.7%. Note that there will likely be some unit-to-unit variation in the level of 3D crosstalk produced by LCoS (and LCD) projectors, even of the same model, since the response time of the micro-display chips used within these projectors may have some unit-to-unit variations.
As discussed in Part 1 of the 3D Crosstalk/Ghosting Blog, there is some published information that indicates that the 3D crosstalk level should be always kept to less than 1% of the peak white level of the primary image (i.e., the ghost image should be no more than 1% as bright as the main image). With a 1% crosstalk level, 3D ghosting will be visible to most viewers during some higher contrast scenes and some viewers may consider this level of crosstalk too high. A 0.5% crosstalk level has been suggested by some as the minimum target level for any consumer 3D TV or projection system as most viewers will find the visible 3D crosstalk acceptable for casual viewing. With 0.5% crosstalk some viewers may not notice 3D ghosting, but more critical viewers will likely see some ghosting at least occasionally in most programs. More ideally the crosstalk level would be 0.1% or less. In this case the peak white level of the primary image would be at least 1000 times brighter than any crosstalk/ghost image. By comparison it appears that the better commercial cinemas using either linear or circular polarization 3D passive technology can achieve crosstalk levels of approximately 0.5% (some are worse or even much worse mainly due to use of less-than-ideal projection screen materials).
Silver Screens for 3D
Passive 3D projection systems that use polarization of the projected light as the means to separate the right and left image streams require the use of projection screen material that retains a very high level of the polarization in order to avoid 3D crosstalk/ghosting. Traditionally a “silver screen” material is used for that purpose. Such screen materials do have a silver color, rather than white or grey, and tend to have fairly high gain due to their reflective nature. While such screens are necessary for use with passive polarized 3D projection systems, they tend to be less than ideal for viewing normal 2D video material. This is because such silver screens have a tendency to have hot spotting, narrow viewing angles, and introduce increased grain in the image. Recently both Stewart and Da-Lite have introduced new silver screen materials that are marketed as suitable for use with both passive 3D as well as 2D projection. The Da-Lite material is called “Silver Lite 2.5” and has a rated gain of 2.5, a half gain viewing angle of 25 degrees and is specified to maintain 99.3% of the polarization. The Stewart material is called “Silver 5D” (as in 2D + 3D = 5D), has a rated gain of 2.0 with a half gain viewing angle of 30 degrees and is specified to have an extinction ratio of 140:1 (reference: Stewart White Paper on the Silver 5D screen material). An extinction ratio of 140:1 corresponds to the screen maintaining 99.3% of the polarization. Thus both the new Stewart and Da-lite silver screen materials are rated to maintain 99.3% of linear polarization. Frequently silver screen materials do not maintain as much polarization when circular polarization is being used. While 99.3% is a fairly high value, that still leaves 0.7% of the reflected light coming from the screen as becoming randomly polarized. It is this unpolarized, or randomly polarized, light that becomes the source for 3D crosstalk/ghosting. Since polarized 3D glasses are never perfect in rejecting all light having the opposite polarization as compared to the orientation of the polarization of the glasses’ lens, the net result can be expected to be a crosstalk level of perhaps as much as 1% (perhaps higher if circular polarization is being used by the 3D projection system).
I used the screen samples of the “Silver Lite 3D” and the “Silver 5D” screen material to do a quick look at how they performed for 2D projection. I projected a 100% white test pattern on to the screen samples and also used a Stewart StudioTek 130 sample for reference. It appeared that both of the silver screen materials were equally bright on axis and based on a quick measurement both appeared to have a a similar peak gain of near 2.5 on axis. Also both of the silver screen materials displayed increased grain (or a fine coarse texture) in the image as compared to the either the Studiotek 130 sample and also as compared to my main screen which uses a white matte screen material with an actual gain of about 1.2. The following composite photo shows portions of the same projected 2D image taken with the matte white screen material across the top, the Da-Lite Silver-Lite material at the bottom left and the Stewart 5D screen material at the bottom right. The top image from this composite was taken with a longer exposure time so as to provide an image closer to the brightness of the lower photos, i.e., for higher gain silver screen samples. The level of grain introduced into the image with the silver screen materials is not shown very well by these photos as it was more obvious when viewing in person.