Optical axis shift with EX1/3
 

I have been experimenting with the EX1/EX3 combination, and have noticed a shift of the optical axis when changing the zoom. For example, set to a tighter field of view, align the cameras for the desired convergence plane, and then when I zoom out the convergence plane shifts. Curious if anyone else has come across this?

Brian

It is a fairly common thing with camcorders and was very apparent with the Sony PD/VX family to the point where I hypothesised that Sony might have allowed for a wider 16:9 sensor to be subsequently accommodated with no modification in the design.

With small sensors and the minute precision involved, there is very little leeway for a sensor's position or the centre axis of the lens to be mispositioned for it to be noticeable. Additionally, with the EX3, there is a detachable lens and its mount as another variable.

I think the manfacturers do very well getting it as good as it is.

I am using this combo. You need to re-adjust the cameras to each other at 0 convergence for each and every focal length you use. It's a pain, but that's the way it works. I haven't found the focal length disparities of each camera to each other to be that great that it couldn't be easily re-adjusted with X, Y, Z axis rig alignments.

Thats what I have been finding, having to readjust at each focal length. I was mainly curious if it was a factor of my cameras or if others also were experiencing the issue. Thanks for the replies.

Does make you wonder if a zoom control for both cameras is very useful.

This can happen with any lens. I was using Zeiss Master Primes the other day and the optical axis was way off between different focal lengths. No two lenses are ever the same. Even the Fujinon zooms that are supposedly matched for 3D are not as perfect as advertised.

Tolerances in the EX-1/3 are simply very sloppy on all counts - the placement of the chip itself, individual lens disparities, and the lens mount. It all adds up to the fact that you need to readjust for every shot. Not a good solution and of course it makes it impossible to zoom.
I recently used 2 EX-3's with a pair of matched Fujinon lenses hoping the lenses would solve the problem but there was so much slop in the mounts and especially the B4 adapters that it was all over the place and probably was shifting during air travel as well. Too bad.

Lenny Levy

I guess from this discussion one can see why 3ality rigs can rent for up to $50,000 per day (with crew) as they have a proprietary lens adjustment algorithms which are computer controlled to adjust for lens disparities. This is what is used on sports broadcasts to enable live zooming, and even Pace 3D rigs utilize their technology.

While many of the sophisticated rigs can compensate for horizontal lens shifts it s very complex to do mechanically, compensating for vertical shifts however is much harder, I have not seen a rig that can dynamically compensate for vertical shifts. The majority (all I think) of the 3D OB trucks are using the Sony MPE-200 3D processor to provide zoom centring compensation. As you zoom, the newer Fujinon and Canon digital zoom lenses pass data about there zoom position to the camera, this is then encoded in the cameras HDSDi output. The HDSDi goes in to the MPE-200 which then know where the zooms are. You can then programme keyframes into the MPE-200 throughout the zoom range to apply horizontal and vertical image translations to the images to compensate for the lens shifts.

The MPE-200 can also compensate for image foreshortening as you zoom making the need to have a fully motorised ring less critical.

Hi Brian,

This might be useful, and it's something I did a while back to test out the EX1/EX3 combo... (a note that the zoom rates are different between the cameras).

http://ftp.datausa.com/imageshoppe/o...oom_center.mov

As you can see, the EX1 tracks almost perfectly straight, while the EX3 really goes for a ride in both axis, to finally land centered at full telephoto. I'm sure this varies from lens to lens for all the reasons mentioned, I'm sorry I wasn't aware of this sub-forum and would have posted this sooner.

I've built a side-by-side rig for the EX1/EX3 and finishing a beamsplitter version right now, with the cameras super-charged via nano3D's.

Many interesting and fun challenges and will share them here.

Regards,

Jim Arthurs

Nice video Jim. Ive been playing with a homebuilt side by side rig myself. A couple years ago I designed a small microprocessor board that has a LANC interface on it. Id imagine through the LANC the zoom percentage could be read out (and controlled) for both cameras. Then the uP could drive a couple of stepper motors to adjust for the axis shift. Hardest part would be prototyping the mechanical assembly, well that and and having spare time to work on it!

Ill pull out the uP board and see if I can read the zoom percentages out of the EX1/3.

Brian

A Lanc controller won't work with an EX. The only lens control is via the remote connector on the lens grip. I'm working on a dual focus and zoom controller that has programmable key frames. Focus is working just fine with the focus ring being driven by a servo motor. I have got zoom working as well, but the lens centering is so variable I'm not sure how practical it really is for EX cameras. As a result I generally work with a couple of different focal lengths programmed into the cameras using the A/B shot transition memories, plus full wide. I know the offsets I need to put in to my rig for each setting so I can switch focal length very quickly.

I can't wait to try out a pair of Canon XF105's as they use the OIS to compensate for lens shifts... very clever.

Thats a great idea on using the A/B settings and knowing what shifts to expect.

Yes, thanks Alistar! Great suggestions and ideas.

Up to now, we've laughingly considered our EX1/EX3 rig a $15,000 fixed focus, fixed zoom camera when configured for stereo work. At least when full wide (5.98mm/64.71 degree HFV) and at a good stop of 5.6 or so, it's pretty much infinite focus. Here's my custom side by side rig in a nice pro photo Convergent Design took of it with the nano3D unit attached...

http://ftp.datausa.com/imageshoppe/o...2015_72dpi.jpg

For doing side by side, I've made a custom spacer to raise the EX1 to the optical axis of the EX3, but I'm handling it differently for the mirror rig I'm putting together, by building the spacer and a custom molded "saddle" for the bottoms of the two cameras. Since the cameras are truly only secured by little 1" by 1.25" metal pads with 4 tiny screws, you have to do something... extra...

Regards,

Jim Arthurs

Interesting, I have ordered some sections of the T-Slot extrusion and similar "cars" to ride on them. One issue you mention that I have been thinking about was how to better secure the cameras with just the EX1 single mounting screw (if I remember correctly the EX3 has two but they're fairly close to one another).

In earlier test rigs, Ive noticed negative effects of one camera "vibrating" differently than the other during pans. When the cameras are stationary its not an issue. Im pretty sure its due to one of the two (or both) cameras not being secured well. I was thinking along the same lines, making a cradle, using some foam, or even putting a strap over the cameras to help reduce vibration. I wish there was a cheese plate that could be securely mounted to the body.

Hi Brian, I will share pictures and details of the methods I'm using to secure the cameras.... I'm going to very elaborate lengths, but I think it's necessary. I will start a thread on the EX1/EX3 mirror rig and include the details.

Both of the cameras are ultimately secured ONLY by those tiny screws in the metal pad. In fact, we had the EX3 get a bit sloppy when those screws got loose, and couldn't figure out where the slop was. Also I noticed about a 1 degree off of perpendicular slant to the EX3 when secured normally... grrr...

Regards,

Jim Arthurs