Sony EX1/EX3 Mirror rig DYI

[Bruce Schultz]

Thanks for the link to the video, it is easy to understand how to use it from the video.

I'm interested in the cross hatch pattern that Bwe shows in his two pics from the post above;

http://www.naluproductions.com/downl...ie_laser01.jpg
http://www.naluproductions.com/downl...ie_laser02.jpg

What laser was used, how did he achieve the cross hatch, and where and how was the laser mounted in front of the rig?

[Bwe Weston]

Hey Bruce,

Sorry for the delay, was enjoying a weekend off work for a change!

Here is a picture (pic 1) of the laser level box, the tech guys at work sorted it for me, so I can't really give you specifics on it, hopefully this helps in tracking one down in the US. They just got it from a hardware store (Bunnings in Australia) over here.

http://www.naluproductions.com/downl...aser_level.jpg
http://www.naluproductions.com/downloads/IMG_0079A.jpg

The real beauty is it auto levels, so just mount it in front of the rig (see 2nd pic) and it really helps initial alignment and checking the rig. You also get a nice flare in the lens which you can line up in the viewfinders too.

If you need any more specifics let me know :)

Brendan~

[Bruce Schultz]

Thanks for the additional picture. I found a cross hatch laser at my local hardware store so I'll be experimenting with it later this week.

Bwe, if you want to compare 3DFF rig notes, contact me off list and I'll be happy to share my tips and tricks with you since I've been using the rig quite a lot since January of this year.


Bruce Schultz
herrschultz@usa.net

[Jim Arthurs]

Got brave enough to test the mirror in the frame, a very nice fit. Here's a couple shots of the 1/16" sponge neoprene to cushion the mirror...

http://ftp.datausa.com/imageshoppe/i...EX3_rig_31.jpg
http://ftp.datausa.com/imageshoppe/i...EX3_rig_32.jpg

And here's the cameras in position for the first time!

http://ftp.datausa.com/imageshoppe/i...EX3_rig_33.jpg
http://ftp.datausa.com/imageshoppe/i...EX3_rig_27.jpg

With a few pieces of black show card installed to block light around the mirror box, I think it's time for some tests...

Jim Arthurs

[Jim Arthurs]

Okay, I've got the worst talent in front of the rig possible; myself. Yet here is the very first series of tests from the rig. I've varied the inter-axial settings on a room sized scene, shot parallel, and converged in post (Sony Vegas 10) to test things out. At this point, not even a tripod mount, it's simply setting on a table.

YouTube - STEREO MIRROR RIG "FIRST LIGHT"

Actually, a pretty useful tests for anybody to do or see... we could use more stereo examples like this posted. Hint, hint...

Wait a minute! I forgot the most important part... the darn thing actually works!

Jim Arthurs

[Alister Chapman]

Looking good, but your camera heights are miss matched. While you have matched L/R images on the screen plane, distant and near objects are shifting either up or down giving you some vertical disparity, particularly noticeable with objects in the background. This problem is a difficult one (and common one) as you can adjust out the error using mirror or camera tilt for any given distance, but at other distances you will have errors. Getting everything equal and exactly on the optical axis is tricky. Even some of the high end rigs have trouble with this.

There also appears to be a rotational difference between the left and right cameras. as objects on the left of frame are aligned better than the right.

I strongly recommend the use of a scene with much greater depth for alignment tests as errors become much more obvious when there is more depth. When the errors are easier to see, they are easier to resolve. I think part of your problem is because the mirror pivot point is not aligned with the cameras optical axis so when you tilt the mirror you also move the virtual height of the reflected camera so you end up in a vicious circle where you adjust the tilt to get zero vertical disparity at the screen plane but the tilt moves the virtual hight with throws out near and distant disparity. Ideally you need the cameras at exactly the same hight, centred on the mirrors optical and rotational axis, then changing the mirror tilt will not affect near and distant disparity.

Still a good job, building a mirror rig is one thing, getting everything centred on the optical axis correctly is another. You look to be pretty close. The really hard bit comes when you start to converge the cameras as any tilt or roll of the converging camera, it's mounting, the lens or the sensor within the camera will translate to differing amounts of camera tilt for every convergence setting. It's enough to drive you mad trying to work out where the problem is.

My Hurricane-Rig design had a problem where the optical axis was changing by around 0.2 degrees as you wound the camera from zero IA and 120mm IA. It took an age to find what was causing the problem. In the end we found that machining out a small area of metal to make the rigs reference plate lighter was causing the 12mm thick plate to distort very slightly in the opposite direction to the machined area. We only discovered this when we noticed the machined plate "ringed" when placed on a surface plate (perfectly flat 5" thick slab of marble) as opposed to making a flat sound. We are now skimming the reference plates after machining and every reference plate will be checked for flatness. Even anodising parts has to be done in a special way to ensure we don't get any distortion.

I'm not sure about neoprene for the mirror surround. Its a tricky one and different people have different views on this. One issue you do need to consider is how the mirror will respond to sound. If the mirror is not mounted rigidly enough it will act as a large microphone and vibrate in noisy environments. Mount it too tightly and distortions will be introduced as the frame expands and contracts with temperature. It needs to be fairly solidly mounted, with no pinch points (which introduce all sorts of weird distortions to the light passing through the glass) but the frame needs to be able to expand and contract without stressing the mirror.

[Adam Stanislav]

What I'd like to know is how you manage to hold the EX-3 pointing down. The EX-3 has its tripod mount attached to the camera on a thin plate held to the body of the camera by four tiny screws. I would be so nervous that gravity would break the camera away from the plate and that I would lose a very expensive (to me) camera as well as the mirror rig, that I could not even concentrate on filming.

Or is it all holding together stronger than it looks?

[Jim Arthurs]

Originally Posted by Adam Stanislav

What I'd like to know is how you manage to hold the EX-3 pointing down. The EX-3 has its tripod mount attached to the camera on a thin plate held to the body of the camera by four tiny screws. I would be so nervous that gravity would break the camera away from the plate and that I would lose a very expensive (to me) camera as well as the mirror rig, that I could not even concentrate on filming.

Or is it all holding together stronger than it looks?

Hi Adam, this was a major design feature in my rig... check out the early posts in this thread. I've made custom resin saddles that fit the curved bottoms of these cameras, relieving all the rotational and other stresses on the screws. They also account for the optical axis differences between the EX1 and the EX3, so the lens centers are at the same height. I spent MUCH time on this. The cameras are held absolutely solid and rigid.

I now use these cradles even when using the cameras as normal on a tripod, it's that big of an improvement over simply screwing them down. A manufacturer should really jump on this and provide them for the EX cameras.

FYI, I've got the EX1 (smaller and lighter) hanging down not the EX3...

[Jim Arthurs]

Originally Posted by Alister Chapman

Looking good, but your camera heights are miss matched.

Thanks Alister, I really appreciate your input. Your rig will largely make these sort of mirror-rig DIY projects moot, and I look forward to seeing your rig at NAB in person. I've got a great amount of respect for anyone attempting to not only build a one-off, but to actually mass-produce their design. Also it gives great hope hearing your own personal trials and tribulations with the process.

It's important to know that my rig is a one trick pony, designed to use the EX1/EX3 exclusively, probably living its life at full wide on the zoom, acting as a test bed for HIT and other image processing functions in-house with Convergent Design.

I'm aware of these alignment issues, and I've got some tweaking ahead, so perhaps posting anything image-wise was premature. Call it the excitement of the moment in seeing any footage with tiny inter-axial!

The cameras are on posts that are machined to custom heights for each camera. I haven't had the time to machine in the exact height yet. And the rig had to be dis-mantled for anodizing, and I've left all the adjustments until after it's put back together. Your tips are wonderful and will aid in that alignment. Thank you.

Originally Posted by Alister Chapman

I'm not sure about neoprene for the mirror surround. Its a tricky one and different people have different views on this.

I appreciate knowing these views, as this is simply my best guest approach without benefit of any industry wisdom! I simply have no idea of how people are securing the mirror in the real world. I really need to take one of those weekend seminars so popular now and get up to speed with how the real rigs work.

Anyway, It's hard to display the exact "fit", but here's my description of the fit; The mirror slides into the frame with no binding, but no real amount of play either. The frame channel is 1/4", with 1/16" neoprene above, below and to the side of the 1/8" thick mirror.

I've not had the chance to lay eyes on a "real" rig, but I think I'm pretty close to "low budget ideal" with this design. The mirror is cushioned, but not compressed, and floats in the frame, with hopefully any external forces acting on the frame and not directly on the mirror. Of course testing is required, and will reveal all the design flaws.

Regards,

Jim A.

[Adam Stanislav]

Wow, you're really good! A very impressive solution. Thanks for sharing it, even if it is way over my head (I am a software guy, not a hardware guy). And yes, I hope someone will start manufacturing your solution and even pays you for the idea!

[Alister Chapman]

When you set your camera heights be aware that the orientation of the camera (ie vertical or horizontal) causes the image stabiliser element in the lens to shift (even when "Off") which moves the optical axis of the camera up and down. It's one of the many difficulties added when you use lower cost cameras. The zooms also wander up and down a lot at different focal lengths.

Find a top quality anodiser. The preparation process can cause warping and any variation in the thickness of the anodising can cause the parts to distort by small amounts.

[Jim Arthurs]

Originally Posted by Alister Chapman

When you set your camera heights be aware that the orientation of the camera (ie vertical or horizontal) causes the image stabiliser element in the lens to shift (even when "Off") which moves the optical axis of the camera up and down. It's one of the many difficulties added when you use lower cost cameras. The zooms also wander up and down a lot at different focal lengths.

Yikes, I did not know that about the image stabilizer. Nor would I have ever found out in this particular case, as the measurements for the milling adjustments to the standoff height would have taken place while the EX1 was in vertical position pointing down.

I had previously charted the the zoom center wandering of the EX3 compared to the EX1 (at least for our samples). Here's what that looks like over the full zoom range (note, the zoom speed is not matched between cameras)...

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

I was going to plot the zoom range offsets out at various focal lengths, but that became moot when I decided that the two cameras were going to spend their time at full wide on the rig, with all the adjustments and tweaks aimed at making that focal length the most perfect match I can get.

Thanks again, Alistar! I'll soon discover how much damage was done by the anodizing, as the parts are all finished and prepped, in a pile, waiting to be put back together this weekend...

[Alister Chapman]

Your video is a good example of zoom centring errors. The EX's are particularly bad, which is why at this point in time we are not doing a dual zoom control. It gets worse when the camera orientation is different. The most troublesome bit is that initial dip that many of them seem to do around 10mm to 20mm which is in the "useful" part of the zoom range from a 3D perspective.

Have fun building up your rig. I look forward to the next installment.

[Sareesh Sudhakaran]

Thank you Jim for taking the time to give us a detailed DIY guide. Would really love to know how the final rig turned out!