I've been watching hours of video from my dual camcorder rig and a thought occurred to me.
My camcorders have 15 power zoom which causes major window violations and major eye strain when going from min to max zoom.
I have been fixing this in post but it's a hassle.
I wonder if my controller were to adjust the interaxial spacing in combination with zooming if that could solve this issue.

Yes. On automated rigs we tend to decrease the interaxial as we zoom because the magnification causes larger and larger divergence in the background. However, you should consider the object-flattening effect caused by longer focal lengths when choosing to zoom.

How about a combination of variable zoom,interaxial and toe in?

How about thinking about moving the camera instead of zooming? With proper follow focus I've found this to be the best solution of all. Zooming introduces way too many problems from lens divergences to cardboarding.

What I am basically trying to do is operate the camera for wildlife monitoring by remote control.
I am already doing this successfully except for the above problem.

It should be possible to move the interaxial servo motor and toe motor together with the LANC controlled zoom to maintain the proper parallax should it not?

I am beginning to see that in order to maintain proper perspective from main to max zoom, I would have to increase the IA and TI proportionally with zoom.

I know that you are using consumer cameras, but you probably need to built a rig with the similar capabilities as Cameron did here: YouTube - Avatar's Cameron-Pace 3D Camera Rig Review (http://www.youtube.com/watch?v=142gTbBDzWM) . He can control everything remotely, which can be fairly easily done with wifi these days (or RC model controllers). Otherwise, since you don't need to be portable, you seem to need a remote controlled scaled down Hurricane rig.

I did some experiments this morning and have discovered that all I need to do to obtain satisfactory results is toe in the cameras in sync with the zoom. I already have a servo controlled toe in and I was using that but can't synchronize them with my current method.
This will definitely get incorporated into my new controller.

I did no looking into this at all, but I know that RC helicopter controllers can combine several axes with one control motion. Maybe there is some blending useful for your need. Of course you cannot zoom while shooting and get really good results. You must zoom - setup - shoot. The optics generally "dances" all over the place throughout the zoom range. This was discussed on this forum in fair detail in the past.

"Of course you cannot zoom while shooting and get really good results."
I'm going to try it anyway but thanks for the info.

Pavel is correct. There's a couple of problems when you zoom using consumer cameras with build in zooms.

First, they may not zoom at the same speed- they're not built to that type of precise tolerance.

Second, the optical center of the lens will change as you zoom.

Third, you'll need to adjust your alignment in post, probably using several keyframes through the zoom.

But I encourage you to try it, even if you only learn that you don't want to do it anymore!

Pavel is correct. There's a couple of problems when you zoom using consumer cameras with build in zooms.

"First, they may not zoom at the same speed- they're not built to that type of precise tolerance.

Second, the optical center of the lens will change as you zoom.

Third, you'll need to adjust your alignment in post, probably using several keyframes through the zoom.

But I encourage you to try it, even if you only learn that you don't want to do it anymore! "
Thanks for the feedback."


A short explanation of what I am doing might be in order.
I have experimented with several different camcorders and finally settled on the Canon HF21s. They have problems with sync which I am able to work around for the most part.
I have done numerous tests with zooming in and out and am having excellent results. I have obtained virtually perfect alignment throughout the zoom range of fifteen to one.
I am controlling and monitoring the cameras in a somewhat unique way. I am viewing the live 3D video in a window on my desktop computer using the checkerboard mode of my 3DTV.
The live 3D video is of surprisingly good quality.
I will put some samples on line and let you all see for yourself soon.

There are people spending 100's of thousands of dollars on matched zooms, motorised rigs and controllers that still end up having to correct dynamically with something like the Sony MPE200 3D processor or in post production. It is extremely difficult to get any two lenses to track correctly without shifting the optical axis. You might get away with it on a small window, but on a big screen it's a different matter.

To elaborate on what Tim said even if you get the electronic and mechanical parts all in perfect sync there is still the problem that a stereoscopic zoom sends two conflicting messages to the brain. Two dimensionally, as you increase focal length you foreshorten the image, that is to say the near and far objects within the scene appear to get closer together, and all objects appear to move closer to the viewer. At the same time the magnification of the image increases the disparity (left/right difference) between those same objects, so stereoscopically they all appear to move away from the viewer and away from each other. This is a serious conflict for the brain, one message, the scale depth cue says everything is getting closer while the stereoscopic cue says everything is moving further away.

You can mitigate this by decreasing the interaxial in sync with the increase in focal length, but now your image becomes flatter and less rounded as the focal length changes, which is again an un-natural effect and objects appear to become a series of flat cutouts at differing depths.

Long focal lengths and S3D are very difficult to get to work right, without cardboarding or scale issues. Zooms even harder to make convincing.

Good luck with your experiments!

Thanks very much for the information. It might prove helpful.
Let me point out that I had cataract surgery on both eyes years ago
and can't focus, in addition to retina surgery on both eyes.
Therefore I'm not qualified to judge the quality.
I have my grandson for that.

This, ironically, would help with the vergence-accomodation problem. Once you correct your vision with eyewear to focus to the screen, you are not re-focusing with convergence. At least it sounds like it.

I did an experiment on zooming a long time ago, when I was really starting with learning about all the 3D pitfalls, and this is what I got: YouTube - 3d: Early 3D Zoom Sync Tests (http://www.youtube.com/watch?v=mUFACOmts1M) . I didn't bother with any precision alignment, just eyeballing it. It is not good enough, specially if you start looking at it on 100" or so screen TVs. I was also good (and still am) at window violations. This is here to make the masters feel good ;-)

Thanks for the YouTube link, Pavel.
I watched it in checkerboard mode on my desktop monitor and it was fun to look at.
All I have been able to do so far is zoom in small increments and adjust the toe in during the incremental pause in order to keep everything behind the screen. I am not able to synchronize the zoom with the toe in yet.
Also, my current setup only toes in the right camera. My new design uses one servo to toe in both at the same time.

I did an experiment on zooming a long time ago

It looked good with my nVidia glasses (with the exception of the vase or whatever it was at the right edge :-) ).

By the way, who was the singer? He sounds familiar but I cannot figure out who it was. Reminds me of Neckář but was much better than him.

Thanks Adam. I completely forgot about the pitcher, sitting on the table, but I didn't want to process it out. The singer is Pavel Novak, was part of a band called Synkopa - music of my youth. He was very popular at one time (~1967), but unfortunately flamed out fast.

Thanks. Unfortunately, I do not recognize his name, even though that was my youth as well. But he is good.

Yes, the pitcher was a textbook example of what not to do. :) It was seen by one eye several times, while the other eye either saw just a small piece of it, or even none of it. That made it look transparent. And it was confusing because it was impossible to decide where in the 3D space it was.

Of course it does not matter in your video because you were experimenting and the whole point of experimenting is to see what works and what does not. Your zooms certainly looked very nice as far as 3D is concerned.

I am starting to see that dual camcorders have a serious defect which is lack of optical stabilizer lock.
Not sure if there is a solution.

Yes, Alister talked about it on this forum in the past. I don't know how to lock the system either. Does anyone have a suggestion?

When the OIS is off on most camcorders it doesn't move about unless you shock the camera. What can happen though is that when you cycle the power the position of the stabiliser may change and you will often get a different position on the horizontal camera compared to the vertical on a beam splitter. This can normally be adjusted out on the rig.

What about if you zoom in and back, will that effect the floating stabilizing element, or will it stay put?

Frank, I just saw a footage here: YouTube - 3d walk yushimatenjin ( 3D?? ???? ) (http://www.youtube.com/watch?v=vlIDDvYd9vc) , which had a lot of zooming in it, and looked very nice. I asked the author, and he said that it was captured with the new JVC camera (GS-TD1), which was recently released in Japan. This is definitely way better for my needs, when taking closer shots, than my current rig. Zooming seems flawless, stabilized... I am impressed. I saw the camera but never any footage from it, until now. The depth bracket doesn't go too far, so it would be a good compliment with my current rig, that can be spaced wider with no problems. Both would be still lighter than mirror rigs.

Pavel the video was impressive showing the camera's ability to zoom, etc. It appeared to be an edited clip since it jumped from scene to scene so does this imply it records in a more usable format for editing than the Sony 3D camera? I haven't found anything showing the file format, and the JVC page only said you could edit the 3D still images. One other thing that seemed like a good feature was time lapse 3D, that could open up some more uses for 3D IMO.

The streams are encoded in MVC format, which is basically the same as the Sony. There are several "profiles" to all these formats, so it is hard to say whether they are exactly the same, but most likely they are. The problem I had with the Sony was that the rep told me that you cannot read the stream out of the camcorder, and it has fairly huge on board storage, but they should be able to output the stream over the USB port, so I don't know how to interpret that statement. I heard that more than once. Of course anything recorded on the removable media should be readable, so hopefully they don't scramble the data somehow. It is my understanding that the Sony Vegas Pro either can now, or soon, will be able to edit the MVC stream. There are some converters available that can unwind MVC to dual AVC streams, and other formats. Of course once you demux the stream, just about any editor should be able to handle it. The beauty is that everything is nicely aligned in time and space. I spend a long time to do that with my footage.

Sony Vegas 10 will edit most MVC formats.

I have been playing with a pair of Canon XF105's on my mirror rig and with the 105's you can use the optical image stabiliser to compensate for lens axis shifts. It's very clever and takes only a couple of minutes to calibrate the cameras using the built in routine. Once both cameras were set I was able to do sync zooms that tracked very well using the IR remote. Very impressive.

That is great news on both counts.

BTW., I have been watching some new YouTube contributions from the new JVC camcorder, and I have mixed feelings about the new batch of narrow stereobase camcorders. They will certainly be lot more error proof than current amateur "rigs", and useful for closeups, but as would make sense, the 3D bracket is very small and the videos are "cardboardy". Good for close shots, but the 3D becomes 2D very quickly. I'd love very portable down mirror rig...

I agree. A mirror rig is the only way to go. Anything else is too limiting.

I started building a portable mirror rig but had to stop because it got too cold to work in my garage. Will resume in the spring.
In the mean time I've been playing around with this contraption.
YouTube - RO3DCV (http://www.youtube.com/watch?v=saeEO2FxlgE&hd=1)
Needs to be watched in 3D.

This is cool Frank! You are improving R2D2; maybe this one will be R2D3...or 3D