Film-Tech Forum ARCHIVE: Superimposing 2 Data Projector Images

This is topic Superimposing 2 Data Projector Images in forum Film Handlers' Forum at Film-Tech Forum ARCHIVE.

To visit this topic, use this URL:
https://ft-forum.com/ft/cgi-bin/ubb/ultimatebb.cgi?ubb=get_topic;f=1;t=005480

Posted by Matthew Peters (Member # 1495) on 10-19-2003, 09:53 AM:

I’m setting up two digital projectors and superimposing the projected image to increase brightness and contrast in a location where unavoidable factors are causing and inferior presentation. This is a once off short-term solution.

There is too much ambient light in the room and consumer model projectors make the problem even worse. However stacking two projectors on top of each other and correctly aligning the images so they superimpose upon each other have significantly increased the quality.

I project an alignment image I created, and then adjust the keystoning and frame position until they line up perfectly, increasing the overall brightness and contrast.

This took close to half an hour of fiddling to align the projectors correctly,
Does anyone have any ideas on:

a) How to align them faster?
b) A better aligning image

My alignment image

Thanks
Matthew [thumbsup]

Posted by Matthew Nock (Member # 1575) on 10-19-2003, 10:28 AM:

Hi Matthew,

the Panasonic LT-6xxx series projector we are using in our largerst screen for Video Advertising talks about doing exactly what you are doing, using two projectors for making the image brighter.

Unfortunately, they make it sound very simple! (they describe it in a paragraph).

If both projectors are exactly the same, I would talk to the manufacturer, and see if they have a process (the panasonic units have a bracket, and an interconnect cable I beleive).

This should then allow them both to work as one, well thats what I understand from the jinglish manual that we have.

Cheers,

Matt

Posted by John Anastasio (Member # 607) on 10-19-2003, 12:53 PM:

You're running into the same problem we used to have when converging color picture tubes. Getting all three colors to track exactly the same way on a screen used to be a real pain. What you're doing is really quite common practice. The Museum of Broadcasting in NYC used a pair of projectors just like you are to increase brightnes and contrast. Use a video signal test generator and set it on the convergence pattern. It's a series of white lines equally spaced horizontally and vertically on a black background with a white dot in the middle of each square that's produced on the screen. It's very easy to see the pattern from each machine with this arrangement. You could also easily generate the pattern with any computer.

Posted by Steve Guttag (Member # 268) on 10-19-2003, 01:01 PM:

There are honest to goodness stacking rigs out there (Try Premier Mounts).

As to a test pattern...I would use a crosshatch. It will let you see the convergence on all areas of the screen.

I'd be surprised if your Contrast went up...your overall light level will go up, blacks as well as whites.

Steve

Posted by Paul Mayer (Member # 355) on 10-19-2003, 01:54 PM:

A half hour is not bad for doing a good overlay. Some of the things I do to make things go faster when setting up to do one:

Ensure both machines are level to each other--I bubble level both machines for front-back and side-side tilt as a starting point. Use of a stacker can help here.
Ensure both machines' projection axes are parallel--no physical toe-in, at least as a starting point. Again, use of a stacker can help here.
Obviously you want the same lens in both machines.
Ensure that any electronic keystone, linearity, or skew correction is dialed out as a starting point.
I also zero out any previous picture adjustments (gains, black levels, chroma levels and phase, clock phase).
I especially zero out sharpness. On most machines any added sharpness greatly mucks up the quality of the edges of vertical lines, making it tougher to see when you've got a good overlay.

With the machines physically positioned this way, most of the mechanical part of doing an overlay will be straightforward, i.e. size, centering (H/V lens shift/H timing), and focus should be all you need. One should try to position the machines such that little or no electronic keystone correction will be needed. Many machines introduce visible moire to the image once the image pixel array no longer maps to the imager pixal array. Also one may need to touch up H/V linearity or skew (rotation) on one machine in order to achieve perfect alignment. Linearity and skew, if available at all, are not usually available on user menus--you may have to access a service menu to get to it.
As for geometry patterns, I use a center cross to start, then switch to a circle hatch. The cross lets you get the basics of size, centering, and focus. The circle hatch lets you evaluate overall geometry (H/V linearity and/or skew) and corner quality.

I also bring along a piece of cardboard to quickly douse one machine--by rapidly flagging one machine you can see small overlay errors pretty easily. This also works for comparing picture adjustments.

That's it for overlay and geometry--the rest is picture adjustments.

I set picture black and white levels next, using a needle pulse or white window for the white level, and a pluge with APL for the black level. Obviously the room lighting must be set before doing this. The SMPTE bar pattern will work for this in a pinch.

I take a quick look at grey scale using a chip chart only to see if there are any really gross RGB mistrackings going on--ideally all of the chips should look neutral grey. If things are really ugly you could try to adjust RGB gains and blacks by eye to get close. To do a real grey scale you have to have a 6500K reference and a color temperature meter and more than likely a lot of time.

Next I look at chroma level (and phase if using an NTSC source) using SMPTE bars. The reverse color chips make setting chroma level (and phase) very easy compared to the old EIA bars.

Finally I look at the actual show pictures and check sharpness level. I try to use as little sharpness as possible, since most sharpness circuits are really noise generators and enhancers, fattening up vertical edges and adding ringing to the picture.

Here's some frame grabs from the Avia and Video Essentials DVDs. Note that in the actual cross and hatch patterns the H and V lines will all be the same width and amplitude--the variances you see here are from JPEGing these frame grabs for posting here:

100IRE center cross (Avia):

100IRE circle hatch (Avia):

Needle pulse plus PLUGE (Avia):

PLUGE with low APL (VE):

PLUGE with high APL (VE):

Grey scale (Avia):

SMPTE bars (VE):