Adventures in Digital Projection

John Hart

Program in Atmospheric and Oceanic Sciences

University of Colorado

Boulder, CO 80302

hart@tack.colorado.edu

nimbus.colorado.edu/hart/science.htm

Feb. 26, 2005

INTRODUCTION

Can digital machines be used in the classroom for 3D projection?   Digital projectors are mysterious beasts.  There are literally hundreds of models.  However, it is very difficult to see them side by side, or even, in fact, to preview and test various units.  The main salesman at a large online company told me "You have to decide what you need by reading the specifications, then order one (unseen and untested)."  You usually cannot return a projector.  If you can, there is  typically a huge restocking fee, like 33% of a large purchase price.  There are a few reviews out there, but generally not of ones the that might appeal to those interested in 3D or "stereo" projection.  The only opportunity to see a lot of units at once is to travel to a trade show like CEDIA.  Since I couldn't do this,  with some great effort, and help from Projector Superstore and Maland Presentations, who were gracious enough to help set up demos, I was able to test a number of the latest and greatest digital projectors, specifically:

Ektagraphic 35mm slide projectors (for reference).

Boxlight CD726 DLP XGA, 2500 lumen projector (our current unit).

Canon Realis SX50, LCOS, SXGA+, 2500 lumen projector.

Projection Design F1+ , DLP, SXGA+, 2500 lumen projector.

Panasonic DT5500, DLP, XGA, 5500 lumen projector.

Panasonic DT7700, 3-chip DLP, SXGA+, 7000 lumen projector.

My measurements and impressions of these machines are summarized below.  First, it is useful to discuss some realities of stereo projection screens and projection geometry before evaluating the projectors themselves.

SCREEN GAIN AND SEATING

Most stereo projection uses the passive polarized method where two projectors, equipped with front-of-the-lens polarizers, send out Left and Right beams polarized at + and - 45 degrees off of vertical.  These beams are reflected off a polarization-preserving screen and are viewed by an audience wearing inexpensive polarized glasses with similar polarization axes.  Screen extinction must be high to avoid ghosting.  Since light losses in the polarizers are at least 2 stops (a factor of 4), high-gain screens are often used.  Such high-gain screens are more directional than low-gain screens, and this can lead to undesirable variation of the illumination level across the screen.  This is exacerbated if you sit in the wrong place.

A very good screen for 3D projection is the Stewart Silver 3D.  It is the best I have seen (high extinction, high gain, relatively wide viewing angle).  Below is the gain chart for this screen, off the Stewart website:

Gain is 3.2 on axis (i.e. perpendicular to the screen, the projectors are assumed to be far back, sending out essentially parallel light).  The "half-power" point, where the gain is 1.6, is at 22 degrees ( + and - angles are off the central perpendicular to seat A, as shown below).  The yellow lines are a mathematical linear interpolation to the gain curve that is accurate enough for our purposes, provided the angle theta is less than about + or - 30 degrees:

Gain G = 3.2 * (1 - | q | / 44 )                                                                 (1)

Here is a typical theatrical viewing geometry:

The screen has width W feet, a row of seats is D feet back, and q is the outboard offset of the most extreme seat (measured from the edge of the screen).  Some things are immediately obvious with respect to the location of the first row.  If we require that the perceived illumination not vary by more than 1 stop across the screen (a fairly perceptible amount actually):

(2)     q₁ should equal the half power angle.   Thus D/W = 1 / [2 tan (22^o )]  = 1.24 , gives the nearest position D of the central front row seat A in units of screen-width.  Or, seat A should be 1.24 screen widths back ( ~ 15 feet for a 12 foot screen, etc.).  For other screens just substitute its half-power angle for 22^o.

(3)     q₂ should also equal the half power angle for the seat B in the first row perpendicular to the screen edge.  But for this seat, D/W = 1 /  tan (22^o )  = 2.48.  Thus the first row edge seat B should be much further back than the center seat A!  For example, if the first row is straight and positioned at the closest distance compatible with (2), i.e. D/W = 1.24,  then for seat B, q₂ = arctan (W/D) = arctan(1/1.24) = 38^o, and the power loss from the close edge to the far edge will be about 2 whole stops!  The 3D image will also, of course, be significantly skewed, but that's another issue.

The obvious fix is to make the whole front row  ~ 2 screen-widths back so that all seats have about a 1 stop or less variation across the screen.  We shall see that this distance is consistent with minimizing pixel artifacts in common digital projection setups.

What about flare out, the widening of the rows as you go back?  If P is the acceptable power ratio across the screen (in the above P = 2, or 1 stop), then, using the gain model eqn.(1), it is easy to see that the ratio of gain at the right edge from seat C, GR, to the gain at the left edge from seat C, GL is:

GR/GL = P = (1 - q₄ / 44) / (1 - q₃ / 44 )                                                               (4)

Inserting the trig formulas, and rewriting gives:

P [44 - arctan ( (1 + X ) W/D )] = 44 - arctan( X W/D)                                          (5)

where

X  = q / W    is the ratio of column outset from the screen edge, in units of screen width.

Here is the approximate (quick numerical) solution of (5), for P = 2 :

Looks like a straight line starting with q = 0 at D/W = 2.48 (as we found in (3) above).  Math aside:  use of small angle approximations in (5) would give a straight line.  For the row that is 4 screen-widths back, the seats can extend about 1.5 screen-widths past the edge of the screen.  Or, the total row-width at D/W = 4 is about 4W, but the total row-width at D/W = 3 is only 2W .  OTOH, at D/W = 6, the row could be ~ 8W wide.  Anyway, this is a useful guide to seating an audience so that screen brightness is no obnoxiously non-uniform.  My guess is that such a seating will also prevent abnormal 3D image skew as well.

3D PROJECTION AND PROJECTOR GEOMETRY

Typically, 3D projection is accomplished by stacking the projectors vertically, like this:

A pair of Boxlight CD726's used for polarized 3D projection in the classroom.

Polarizers at +45 and -45 degrees off of vertical are placed over the lenses, as shown.  The vertical stacking leads to several errors, compared with the ideal of having the two lenses superimposed.  The geometry is shown below:

Here, in this perspective sketch,  we have two projectors 1 and 2, a height H apart, located distances D and D' to the screen.  Alpha is the tilt angle, required to get the images to line up vertically on the screen (assuming the projectors do not have the "lens shift function").  Theta is the lateral (in and out of the paper) beam-spread half-angle, determined by the focal length of the projection lens (which may be variable if the lens is a zoom).  The base widths of the two projected images are W and W' , respectively.  Because α is non-zero and the distance D' is not equal to D when the projectors are stacked vertically, there are several errors, the size of which are easily worked out using trigonometry and liberal use of small-angle approximations based on α ~ H / D << 1 :

Base Magnification error   M  = W' / W - 1 = D' / D - 1 =  sqrt( 1 + H² / D² ) - 1    ,   as a fractional error.   This should be corrected by moving the tilted #2 projector an amount deltaD ~ H * H/(2 *D) closer to the screen, or by using its zoom lens if the projector is so equipped (most are).

Keystone error    K  = 1 - W'' / W  ~  W  H / [A D² ], as a fractional error w.r.t. screen width W.  Here A is the aspect ratio (width/height) of the presentation (e.g. 4/3).  This error comes from the fact that the top of  projector-2's image will be narrower than projector-1's, even if W is made the same as W' at the bottom.    To minimize keystoning, one should use projectors that have small heights and stack them close together as possible vertically in order to keep H small.  Since K scales with the square of D, keep D large (i.e. put the projectors as far back from the screen as possible, consistent with its lens).

Depth of Field Requirement DOF ~ W  H / [A D cos(2θ)] .  This gives the required depth of field that the projector lens must have to adequately keep the top and bottom of the keystoned image from the top projector in focus.  This required DOF is linear in H.  Still, small H is better than big.   The projector lens typically produces a DOF ~  W * W * a / (P * S), where a is the aperture of the projection lens (typically a ~ 1.5, i.e.  f 1.5),  P is the resolution of the imaging chip (typically <1500 pixels), and S is the size of the imaging chip (typically 1 inch).  Thus for a 6 foot wide image on-screen, a typical digital projector lens will have about  5 inches of DOF.

The above formulas show that to minimize errors you need to keep H small and make D large.  Sometimes making D large is impractical (room size, available lenses, etc.)  As an example, consider typical projection using Kodak Ektagraphic slide projectors stacked vertically, with 4" focal-length f2 lenses.  For this system H ~ 8".  The DOF of the projectors is fairly small  (because film has more resolution), maybe 3 inches.   If W = 6 feet, and D = 18 feet (typically):    M ~ .07% ,  deltaD ~ 1/6 inch, K ~   0.9%  (~ 0.7"), and the DOF required is  ~ 3 inches.  Thus, for slide projection using twin Kodak slide projectors stacked as close as possible, the DOF is just barely OK (f3.5 lenses would be more OK).  The keystone error will be noticeable by anyone sitting 3 or so screen widths back, or closer.  Going to 7" projection lenses and making D bigger  would reduce the keystone error by a factor (4/7)² , in other words diminished about a factor of 3.  In our example, it then becomes just 1/4 inch.  Projection setups with this latter error are commonly used by stereo professionals such as the National Stereographic Association (at their conventions, where people watch slides for hours without complaining too much).  Thus, it is perhaps fair to say that a keystone error of .25% or so is tolerable.

THE PROJECTOR SHOOTOUT

I was able test four "just shipping"  top-line projectors, the Canon Realis SX50, and the Projection Design F1+ (a.k.a. Christie DS+25, DigitalProjection SX+), along with two super-bright Panasonics, the DT5500 and the DT7700.  Except for the XGA DT5500, these all sport SXGA+ = 1400x1050 resolution, which is a good improvement over the now standard XGA (1024x768) class machine.  In this technote, these projectors are compared with impressions of film projection using Kodak Ektagraphics carrying high quality lenses, and with another good XGA DLP machine, the relatively inexpensive Boxlight CD726 (seen above in the stacked-projector picture).  All these tests were done at 4:3 (Width:Height) aspect ratio, which is typical of stereo slide projection where slide mounts that are 32mm wide x 24mm high are employed.   The links just given supply fairly complete manufacturers' specifications.  But what do photographic images look like out of these machines?  Can digital machines be used for polarized 3D projection by placing linear polarizers oriented at plus and minus 45 degrees over the lenses of two projectors without getting serious color errors?

Various projectors:  Canon Realis SX50 (left, top), Projection Design F1+ (right, top), Panasonic DT5500 and 7700 (L-R, bottom).

The Realis and F1+ are  ~ 4 inches high, 12 wide and 12 deep) and weigh about 7 or 8 pounds each.  The Boxlight is a little taller (H ~ 6").  The Panasonics are bigger.  The DT5500 is about like two slide projectors including carousels and tips in at 33 pounds.  The DT7700 is a 50 pound monster probably suitable only for fixed installation.   Though H is so large for these Panny's that keystoning would be a big problem, these guys have lens-shift, whereby keystoning can be eliminated completely without inducing any digital artifacts.  All these machines use arc-lamps that have a color temperature much more like sunlight (6500K) than the yellowish light of a halogen-lamp slide projector.   Both the Panasonics have dual arc-lamps and are extremely bright.

Here are important observables:

Brightness:  Digital projectors claim high brightness.  But the posted specs are usually for the data-graphics mode, which is useless in projecting images.  To successfully project images without burning out highlights and muddling the shadows, you have to adjust gamma, contrast, colors, and brightness of the digital projectors to get to good film-like rendering with good dynamic range.   All the units except the Boxlight have single-click settings that give a good start towards excellent film-like projection.  Such adjustments invariably decrease the brightness, often significantly.

Contrast:  Early digital projectors claimed contrasts of 300:1 or so.  A good computer LCD monitor sports 500:1.   Now-a-days digital projectors often claim 1000:1, and an elite few go up to 6000:1 or more.  In a digital projector the contrast specification gives an indication of the light leakage in the black level, as it is the ratio of full-on to full-off (255 vs. 0 inputs into all three RGB colors).  But contrast also gives a rough measure of the range of brightness that can hold detail.  The higher the projector contrast, the greater the tonal range, so along with deeper blacks and better color saturation, you get more details in the shadows.  One can easily see the difference between a 300:1 and a 1000:1 projector, but at the high end (1000 vs 2000 vs 4000 etc.) other issues such as presentation technology also affect the image quality.

Resolution:   Here is a most controversial topic.  Film resolution is often measured by seeing when closely-spaced black and white lines on a resolution grid meld together into a gray mush.  The "digital extinction resolution" is half the lateral pixel count (in line pairs / per image width).  The absolute (or graying together) resolution is often found to be about 80 to 85% of the extinction resolution (ref:  Phil Askey's digital camera reviews).  Digital and film projection are different animals.  Digital is better at displaying sharp edges, while film is more effective at displaying muted textures at small scale.  Photographic images generally have both features, but the eye first notices the former, and if allowed to study a picture carefully then the textures come into play.

Pixels:  Digital projectors beam out individual cells or pixels.  There are three popular technologies.   Transmissive LCD panel pixels historically have had small apertures (where each element is surrounded by a large black border - leading to the infamous screen door effect, or SCE).  Recent implementations in LCD home theatre projectors like the Panasonic AE700 have very little SCE.   Reflective DLP pixels have a little SCD. which is only noticeable a couple feet from a 6 foot wide screen.   The new LCOS reflective pixels of the Canon Realis have essentially NO SCE at all, each light element butts right up against the next. 

MEASUREMENTS

We measured brightness and contrast and took notes on a set of test images (two are shown below) after the projectors were adjusted to send out high quality film-like pictures.  All tests were done in a partially, but not completely darkened room.  Quantitative level measurements were made with a Minolta Autometer IVF digital light meter in the incident mode (i.e. we looked at the projected beam, not what is reflected off whatever screen was being used).   Qualitative distances are simply my impressions (with 20/20 vision) and should be considered with a good dose of "plus or minus".

(1)  Need 4 slide projectors to dissolve for 3D projection, w/ stands, lenses, and dissolve units.  Need two digital projectors for polarized 3D projection.

(2)  Pixel = line (extinction resolution).  For film, we estimate "gone-to-gray" by projecting a test pattern.

(3)  ANSI on-off manufacturers specification.

(4)  As claimed in specification, but see relative brightness data.

(5)  Lumens ~  2^EV, or 1.0 EV = a factor of 2 in brightness.  Only differences in the numbers shown are relevant.  Example:  8.4 vs. 7 => 2 ^ 1.4 brighter.

(6)  Difference of meter reading of highlight retaining detail and darkest scene elements retaining detail.  Not the same as full-screen on-off.

(7)  Distance where film and digital start to look similar (eye cannot see major detail differences).  6 foot wide screen.

(8)  Distance back from a 6 foot wide screen where the image looks smooth (no grain, pixels, or artifacts can be resolved).  In feet.

(9)  In feet. This is the worst case - a highly detailed scenic.  A scene with sharp edges against a uniform background may hold up 30% closer under digital projection.

The digitals were from about  ~1/3 to more than 2 stops brighter than an Ektagraphic equipped with a  f2.4 lens and an EXR lamp beaming through a "high-bright condenser module" with an empty Gepe glass mount in the slide gate.  In practice a "clear" film sheet will induce even more attenuation.  The dual-lamp Panasonics appeared quite in tune with their specs, even after adjustment to get a good film-like image.  In addition to the raw brightness data, the digital projector images look snazzier,  because of the much higher color temperature of their arc lamps, compared with the dull yellowish halogen light of the slide projector.  Skies are really blue, whites are really white, etc.  It is worth noting that using an f3.5 lens on the slide projector causes a loss of about another 1/3 to 1/2 stop of light compared with our test system.  Thus the digital projectors can appear significantly to blindingly (on a 6 foot screen) brighter in practice.

Dynamic range for these high contrast projectors seem comparable to film projection.  I wasn't able to measure this carefully.   Take an image with very bright parts and some deep dark shadows.  Use the digital exposure meter to determine the difference between those highlights retaining detail and those shadows still having features rising just above the mud.  A Provia slide and the Boxlight projecting a scan of it had about 8 and 7.7 stops of range respectively.  The Canon had about 7.5 stops.  All projection methods retained some information in the darkest parts of the images (the test images are shown below).

Resolution, or IMAGE DETAIL, should always be referred to the field of view of the audience (measured in distance to the screen divided by screen width).  There are so many nuances here....   But if I move back, where does it start to look really nice?  Or, beyond what distance would increases in the pixel count not be significantly noticeable?  Or, How close can I get before pixel grids and digital artifacts like jaggies can be routinely detected by the human eye?  Or, for film, where does glass-mount anti-Newton dimpling or film grain become apparent as I move closer to the screen?  We looked to two test images that stress both dynamic range and detail (reproduced below in low quality).  There is a textural scene (left) where you have to move back to 18' (from a 6' wide screem) in order that an XGA projection looks similar to the projected slide.  The right scene has sharp edges in the flying water, set against a dark wetsuit.  For this one, IMO, the XGA projection is better than film, until you get so close that the pixel grid "noise" can be seen in the water.  This happens at about 13 feet for the XGA Boxlight using this particular image.

• The Boxlight looks smooth and film like at about 3 screen widths back (SWB),  i.e. with a 6 foot screen, sit 18 feet back and the slides and digital images look similar w.r.t. detail that the human eye can resolve.  However this varies with the scene.  Some scenes with lots of sharp edges look great in digital projection, even quite close, as discussed above.

• The SXGA+ machines look smooth and film-like at about 2 screeen widths back,  i.e. with a 6 foot screen sit about 12 feet back and you will have a good experience.

• Extrapolating, a 1600x1200 Realis would look film-like about 1.7 screen widths back, or about 10 feet for a 6 foot screen.  The incremental decrease in comfortable viewing is probably not worth the premium price increase one would see in going from 1400x1050 to 1600x1200.  A 2048x1556 would allow seating at 8 feet or about 1.33 screen widths back.  This latter "QXGA" projector will cost a fortune, and recall the problems with screen illumination when you sit this close.

• Pixel noise (i.e. the grid lines or regular structure of pixels, or other artifacts) becomes annoying at about 14' for the XGA and at about 8' to 9' for the SXGA+ machines (again w.r.t. a 6' screen).  If you sit too close to the screen, these grids start to be seen in individual image features.   The Canon is better because of its lack of SCE, but you still see artifacts like jaggies from a anomalously close front row.  SCE pixel annoyance is more prominent in uniform color sections of an image.  If a scene does not have these, SCE artifacts can be less noticeable.  However, very finely detailed images can have internal pixelization that becomes apparent as well.

At 2 SWB, SXGA+ resolution seems adequate.  It is somewhat uncomfortable at 2 SWB with the Boxlight or DT5500, and it's better to be 15' or 16' back from a 6 foot screen when using these units.  Please note that this review concentrates on projecting quasi-still images - which is a most difficult test.  For moving video or still-in-motion (a la Ken Burns) slide shows, there is mitigation of the pixelization and artifact problems by motion and eye-averaging over noise, and a 2 SWB XGA presentation can look pretty good at 2SWB.

IMPRESSIONS and COMMENTS on the VARIOUS DIGITALS

• The Realis and F1+ are clearly superior to the Boxlight in image quality.  Color saturation and resolution are much better. 

• Canon has a real winner with its new reflective LCOS design, especially considering its much lower price point!!

• The Realis appears sharper.= than the single chip DLP F1+.   SCE is virtually non-existent.  Although its overall contrast spec is lower than the F1+, on subtle textures there appears more substance with the Realis.

• Colors are slightly more saturated, appearing more vivid with the Realis.  Reds and blues are vibrant, compared with being a little muted on the F1+.  This is a typical problem with single-chip DLP's.

• The Realis lens is more centered (slightly) on the screen.  F1+ has the center line from the lens matched to the lower edge of the screen, meaning the projector must be tipped more, unless the screen is mounted high.

• The F1+ seems to have slightly better dynamic range.   Dark areas hold detail perhaps just  little better than on the Realis.  But Realis overall seemed better.

• Having higher contrast (depending on settings), the DT7700 can display an image easily the equal of the Realis.  The DT7700's picture is much brighter, and has no alignment artifacts or lens distortions (since the DT700 has an expensive high quality optic, costing almost as much as the Realis itself).

PROBLEMS FOR 3D PROJECTION WITH THE REALIS and DT7700 (color shifts with external polarization at plus or minus 45 degrees).

• Polarization of the F1+, Boxlight, and DT5500 single-chip DLP beams leads to even white (i.e. external polarization does not affect the beam). 

• Polarizing the Realis by placing a glass linear polarizer (a Tiffen camera filter) at 45 CW degrees leads to a non-uniform color error.   The top center part of a white frame shows magenta and the bottom and upper sides become slightly tinted green (see the sketch below). Not much, but noticeable.  Polarizing 45 CCW inverts the color error (green tint on top center).  Combining the two should get back to more or less pure white (I think), but in stereo projection you might get perceptible color errors, particularly in images where objects should obviously be pure - like a snow field or a cloudy-sky extending top to bottom in the background, or perhaps a full frame portrait of a human, where everyone will know what the skin should look like.    Without setting up a pair of projectors and trying it, it is hard to say how significant this will be, but it is a potential show stopper.  Can we fix the problem?  Possibly, with preprocessing of images using a color mask that corrects it.  For example, take the 45 CCW image and add green to the top-center, etc., to recover a uniform sort of white balance.  You could probably make the mask by photographing the projected white image with a digital camera.

My rendering (approximate, from memory, do not base a purchase decision on this sketch) of what 45 degree CCW  linearly polarized white looks like out of a pre-production Realis.  A pure white bar has been added for comparison purposes.  45 Degree CW is inverted with yellow-green on the top and a magenta tint on the bottom.

• Polarization of the 3-chip DLP Panasonic DT7700 produced a slight, but uniform, color shift.  In one filter orientation white was tinted a tad blue, while for the other angle, it was just perceptibly warmer.  However, the effects were uniform in space, small, and probably could be fixed simply by adjusting each projector's color balance.

PROJECTION TIPS

• Use of digital keystone correction on these projectors seems to cause a significant loss of sharpness!  Thus,   digital keystone correction is NOT RECOMMENDED.  It may even be better to set the lower projector up perpendicular to the screen, and live with the small keystone effect you might get.  Be sure to set the projectors as far from the screen as their zooms will allow.  For example:   The Realis is ~3.8" high, and projects a 6 foot wide image from 14.7 feet away.  Thus the keystone error (using the formula from above) is 0.5%  or so, or .36" on a 6' wide 4.5' tall screen.  This seems a little too big (it's about 7 pixels), and one might be forced to make a digital correction, thus lowering the quality of the image.  With a taller shorter-throw projector like the Boxlight, the keystone error is more like 1.3%, and will certainly require digital keystone correction.   The Panasonic projectors have "LENS SHIFT" in which the image can be moved horizontally and vertically using motorized adjustments on the lenses.  With lens shift, you can easily and effectively  remove most all projection geometry errors!

• VGA analog input leads to ghosting and loss of sharpness.  Always use DVI digital input - whence sharp lines are clean, etc.

• Make sure the projector base and the screen are perpendicular.  If you tilt the projector w.r.t. the screen, it is possible the lens will not focus the image at both the top and the bottom at the same time.  Digital projector lenses are low f-stop affairs (f1.5 to f2.0 are typical).  They have very little depth of field.  You cannot fix this problem with the digital keystone correction, since this only affects the image geometry, not the focus.  The lens shift function circumvents both DOF focus problems and eliminates keystoning (provided the screen and projector bases are perpendicular to begin with).  If you don't have lens shift, choose projectors with small height H, and position them as far from the screen as possible.

• Use a high gain screen to compensate for polarization losses.  Dull or dim images have no snap.  Put the first row of seats about 2 SWB, maybe a little closer for those who don't mind artifacts and non-uniformity of brightness.  However, advise people sitting near the front to be seated near the center.   Pay good attention to room darkening and reflection control.  This builds contrast and saturation and enhances the experience.

CONCLUSIONS

The author's Personal-Opinion Table (ranked 1'rst, 2'nd, 3'rd, etc.,  and Good, Marginal, Poor)

*  4 projectors, 2 dissolve units, 4 lenses, two stands, etc.

**  Really hard to call.  Depends on what you need to do.  For videos and motion - you have to have a digital projector.  Period.  For a really big screen, there's nothing like a bright snappy dual arc-lamp image (hence ? on DT7700).

If I were just into small venue 2D projection,  I would buy the Canon in an instant.  What a little beauty!   The polarization color-error with the Realis is unfortunate, but it may possibly be circumvented by pre-processing the images.  If possible, such image back-calibration (in order to cancel the projector's color shift) must be the LAST step in any slide show generation/player or movie rendering.  For example, if a movie or slide show pans and image, the color correction must be done afterwards, otherwise the correction and the projector error won't line up.  The keystone error, and small distortions of frame on the Realis may also not be negligible, forcing the use of digital-correction that softens the images a perceptible amount.  Unfortunately all this seems a bit tedious w.r.t. the Realis for stereo display.

For 3D polarized projection these projectors are all bright enough for a 6 foot wide screen.  In a totally dark environment, 2000 - 2500 lumen units such as the Infocus LP530 have been used successfully (though in my opinion a bit marginally) for polarized projection on screens up to about 12 feet wide.   In any case, such digitals are about 1/2 a stop brighter than typical 35mm f3.5 slide projectors and have more snap because of the color temperature of their arc-lamps.  For a fixed large-venue installation the DT7700 is an awesome machine.  It is large, heavy, and a bit noisy for a small living room (at full lamp power).  For smaller gigs it's overkill and is so bright you can hurt your eyes (unless you turn off one lamp and close down the iris).  It's too bad Panasonic has not put the SXGA+ DLP chip into the DT5500 (yet), because the 5500 is very quiet and reasonably movable.  Rumor is there will indeed be a HDTV type DT5500 (1280x720) soon.  A pair of DT5500's, at about 65 pounds total, would be about equivalent to the struggle of carrying around 4 slide projectors and associated dissolve gear (not to mention forklifts full of glass mounted slide reels).  In any case, if you want to project moving material, digital is the key.  If the machine you want isn't listed here, at least I hope to have supplied some tips on what to look for.  The quest goes on......

Return to Tech Note Page

Home