Synchronization, the degree to which the cameras actually fire together when the preset button, then the trigger button, is pulsed, was tested with the spinning wheel.  The chart below shows the results of a 15 minute test run.  As compared with consumer cams like the Nikon CP5000 that  have a predictable drift in and out of sync, the DSLR does not appear to exhibit such periodic sync drift.  Periodic drift  is usually associated with video output.  The Rebel only has post-shot video.  It’s a "shooter-priority" camera.  Nonetheless, there are misfires.  82% of the time the degree of synchronization is better than 5 milliseconds, good enough for most types of photography (excepting action or motions that cross a substantial part, like 3 or 4%, of the field of view in such a time).  90% of the time sync is better than 10 milliseconds.  On the other hand, in general use there will be some mis-syncs.  One philosophy is, "it’s digital, blast away, discard the bad ones."  However, this simple parallel wired remote trigger is much more effective than dual-fingering the shutter button.

Sync errors when shooting in the M exposure mode with MF (manual focus). Preset is depressed for 3 seconds, then the cameras are triggered.

It is interesting that such a diagram is about what I found for a Pentax Film SLR (the ZX-5N), and is about what I got for a Pentax *istD DSLR.  My guess is that this behavior is typical of this grade of DSLR.  A lot has to happen (mirror goes up, etc.), and some electromechanical variations are perhaps to be expected.

A more difficult test is flash.  In this test both cameras are triggered using the remote.  Only the "master’s" built-in flash is up (open), the "slave’s" flash is down (closed). Since the master’s flash is fired when its curtain opens, if the slave’s curtain is not simultaneously fully open, it will appear in the slave’s field of view. How often does this happen? It depends on the shutter speed.  In these tests cameras were set to Manual (M) mode, the shutter speed and aperture were set, focus was set manually, then the pair was triggered.  For exposure times as shown:

1/30’th: Flash worked ~80% of shots

1/50’th: Flash worked ~65% of shots

1/100’th: Flash worked ~ 30% of shots

Thus flash operation in stereo mode is somewhat problematic.  Similar behavior, maybe slightly better, was found with the Pentax *stD and the Pentax film SLR (ZX-5N) twins.

In conclusion, fairly good sync is obtained for regular shooting. The use of flash will be relatively unreliable. 

SO:   Why would anyone want to try this route, when other solutions like the LANC controller attached to a good consumer-cam like the SonyV1 or 717 can do non-flash sync quite a bit better and more predictably, as was shown earlier in my TechNote on the  Sony V1 Twin.

Why use the Canon 300D?  The image quality out of the Rebel is superior.  Here’s an example comparing the 6Mpix Rebel with the 5Mpix Sony V1, one of the highest rated of the 5Mpix consumer-cams.  There is no post-processing of these files (other than resizing to make the thumbnails below):

Canon 300D (L), SonyV1 (R), tripod, f8, ISO100, largest jpeg output.   Click to download full original (LARGE FILES).

300D crop view (showing actual pixels).

SonyV1 crop view (showing actual pixels).

OTHER QUESTIONS:

1. How does exposure lock ("preset") work?

The preset only applies to one exposure.  If you want to shoot a bunch of pictures at the same exposure, you will need to use the Manual Mode (exposure and focus).  In Manual Mode, you can just press the trigger (exposure) button.

2. For stereo, how do you set the zooms to the same value?

The zoom ring is manual.  There is no readout of the setting.  One way is to calibrate by shooting a ruler and marking the rings on the lens.  Slight differences in zoom settings are easily corrected using the resize function in a program like StereoPhotoMaker.

3. Can you make a series of images at different foci in which the image magnification does not change?

Not easy to do with the kit lens.  Manual focus is rough and imprecise.  Focus cannot be controlled by camera (i.e. focus cannot be set electronically).

4. Will the camera work on a microscope or telescope?

Yes.  I put a Kalt Canon EOS adapter on the Rebel and attached it to a microscope. All the exposure functions work fine.  For a fixed ISO the camera picks a shutter speed.  Exposures from several seconds to 1/500 sec were all pretty accurate.

5. How is the viewfinder?

Small, but about as bright as a SLR (Pentax ZX-5N being my reference).  The view is about 2/3 the width of that found in such a film-cam.  The focus screen is decent (depends on lighting and contrast of the scene of course).  AF seemed to work well.  The viewfinder is not a highlight of this camera.

6. Can you focus it easily on a microscope by using the viewfinder?

Depends on the magnification.  With 2X or 4X objectives, yes.  With high magnification, it is difficult.  For the latter, better to use TV or remote capture to mount the camera co-focal with the oculars.  It took me about 10 shots to get the focus right with a 10X objective (and a 3X relay lens) by using the camera's TV-out to display each focus sample.

7. Does the Twain Driver work with PaintshopPro and Photoshop?  Yes, you can download images from the camera directly into both these apps.  Not fast:  20 sec per picture (see below).

8. How is the remote capture software program from Canon that comes with the camera?  How about   http:breezesys.com ?

Both work and do similar things.  You can take pictures remotely from your computer (tethered via USB1.1 cable), and the images get dumped to your computer (in about 20sec for a highest rez jpeg). You can adjust shutter, aperture, ISO, WB, etc., but not focus or zoom.  I don't think either of these programs are set up to fire twins.

9. What is file upload time?

USB1.1 is not that fast.  20+ images (full jpeg), or about 64MBytes, takes about 6 minutes (or ~20 sec per pix).  Probably need to use a CF card reader running USB2.0 if you need faster transfer times.

10. How is the video out?

Very good quality.  It’s not motion video, and only comes out after you take a picture.  Stable and clear.  Can be used to focus, and is faster than the remote capture, because the video frame appears right after the image is taken.  But remote capture is even more accurate of course because the video frame out resolution is limited to NTSC (or PAL).

BOTTOM LINE:   Really good quality images (better than SonyV1, Sony F828, and the like).  Slightly unpredictable sync errors (> 1/200 about 20% of shots, but never > ~1/30).   Unpredictable flash sync:  shutter curtains appear in the slave's image a little less than half the time with a 1/50'th sec shutter opening.  This performance is similar to what I found for Pentax *istD.  Based on a sample of two DSLR's, I speculate that this is the norm for the typical low-end (~$1K each) digital single lens reflex.

Would I use it for 3D scientific photography?  Probably not.  Can’t afford the data dropouts.

Would I use it for (non high-speed-action)  art photography?   Yes!  The image quality is much better than consumer digicams. 

Would a pair of Canon 1DMk2's do any better sync-wise?  Maybe, because the mirror and shutter design is quite different (to enable 8fps….), and the camera is highly optimized for speed.  But at $4.5K per pop, I’m not in a big hurry to find out……