Home of Professional Cinematography since 1996

3D Mirror Testing

Published : 18th December 2010

So much of the final quality of a 3D image seems to come down to the quality of the mirror. As a test of quality I like to use a long zoom lens and then see at what focal length the mirror begins to adversely affect quality. I was wondering if anyone had any other possibly more scientific methods of testing mirrors? One issue is the partial polarising of one channel.

Can this differ between mirrors? Are there other properties I should be looking out for that can stuff up the pictures?

Tom Gleeson DOP
Sydney


Tom Gleeson wrote:

>>So much of the final quality of a 3D image seems to come down to the quality of the mirror. As a >>test of quality I like to use a long zoom lens and then see at what focal length the mirror begins to >>adversely affect quality.

The standard testing method nowadays for glass flatness and surfaces is  laser interferometry. Most
reputable mirror manufacturers will certify their mirrors, but assuming  it's a front surface mirror, I think it would have to be an obviously crappy mirror to be noticeable.

Amateur telescope enthusiasts often build their own laser  interferometers to check their mirrors.

Brian Heller
IA 600 DP


>>"One issue is the partial polarising of one channel. Can this differ between mirrors? "

There's an interesting question! I would imagine that having a standard setup that you could repeat easily might help to assess any differences as you check out mirrors .I can think of creating a "scene" with a reflection off of a known glossy surface at a certain angle to beamsplitter (mimicking a surface that would become polarized by the rig).Then you would study the signal on a waveform of the relative strength of the reflected images... not an elegant solution though.. really only practical if you have a number of rigs all in the same place at once.

Maybe one could illuminate a white card, full frame, and then somehow study/measure the polarization effect with a pola on the taking lens of the reflected camera? Simply comparing extinction levels , crossed and uncrossed or something...

Douglas Koch cscDPToronto


> > "One issue is the partial polarising of one channel. Can this differ between mirrors? "
>>There's an interesting question! I would imagine that having a standard setup that you could >>repeat easily might help to assess any differences as you check out mirrors

It is also worth noting that polarization from a reflective surface is not continuous across the spectrum of colour. Certain colours get polarized more than others. So you'd not just test overall
polarization but how the light is being polarized as it's frequency changes.

Or you can fix the polarization problem and not worry about it anymore.

Eric Deren
Dzignlight Studios
VFX & Animation Design
www.dzignlight.com
North America, currently Toronto (that's for Geoff)


This may too basic a question as I am new to 3D and am just heading into my first 3D shoot next month, but would there not also possibly be an exposure difference between the reflected camera and the straight camera, polarization issues aside?

John Tarver csc DP

Temporarily in Ottawa ....


 

The type of interferometer used for telescope mirrors will only tell how flat the mirror surface is. It won't take in to account distortions that may be introduced by imperfections within the glass that will affect the camera shooting through the mirror.

From a practical point of view using a pair of long lenses and comparing images is a good test as this reveals the real world performance for both the reflected image and straight through image.

Alister Chapman
Cameraman - Stereographer
UK
Office +44 1344 483748
Mobile +44 7711 152226
US Mobile 405-822-1764
www.ingenioustv.com


>>"One issue is the partial polarising of one channel. Can this differ between mirrors? "

If you illuminate a diffuse surface, (e.g. from behind) then cover it with linear polarizing sheet, you can control the polarization entering the camera rig. Rotating the sheet rotates the polarization. If the
beamsplitter/lens combination is also linearly polarizing, then you will see the image dim and brighten as the polarizing sheet is rotated thro 360 degrees. The sheet costs about $50 from Edmund Optics.

http://www.edmundoptics.com/onlinecatalog/disp;layproduct.,cfm?productID=2102

Ian Marshall
lens designer
Cooke Optics


John Tarver wrote:

>> "would there not also possibly be an exposure difference between the reflected camera and the >>straight camera"

It would depend on the design and quality mirror.

Max Penner
Stereographer
Paradisefx.com
310 864 5124 Cell
07999288849 EU Cell
Maxp3d skype


Tom Gleeson wrote:

>>As a test of quality I like to use a long zoom lens and then see at what focal length the mirror >>begins to adversely affect quality

Hello!,

Also shoot a grid chart at different focal lengths and check for geometrical distortion. (Perfect overlap of both images at 0 interaxial). The very short and very long focal length are most critical.

Eric Deren wrote:

>> Or you can fix the polarization problem and not worry about it anymore.

How would you do that Eric? I have been told about a "depolarising front glass" but have not yet seen that in real life, and I have been using pola filters on the cameras behind the mirror, successfully matching my images again on polarising subjects, but at the expense of losing extra stops.

Do you have any other method?

John Tarver wrote:

>>would there not also possibly be an exposure difference between the reflected camera and the >>straight camera

Yes, with the majority of the mirrors I worked with there is a significant difference in both exposure and colour. I have measured differences from anywhere between 1/6 and up to one full stop between left and right eye. In order to get the same depth of field (same physical f-stop) in both eyes
you may want to use a neutral density filter on one camera. Colour differences are often significant but generally small enough to be dealt within post or by the auto-white balance feature of video cameras. I have not yet needed to colour filter cameras separately on the shoot. Two rigs I
worked with have near perfect mirrors. Less than 1/6th stop difference and virtually no colour shift. So yes, it can be done.

I suppose more rigs will get good mirrors over time. For the time being: yes, you need to test, measure and deal with it.

Brian Heller wrote:

>>I think it would have to be an obviously crappy mirror to be noticeable.

Sadly enough, I have been confronted with average quality (deforming) mirrors in 3D rigs more than once. I have had excellent mirrors as well. Mileage varies quite a bit from one to another. The way the mirror is fixed / held also can play a role as glass is more flexible as one would expect.

Best regards!,

Kommer Kleijn SBC http://www.kommer.com
VFX Cinematographer Brussels, Belgium, Europe
Stereography, Motion Control, Digital Imaging, Large Format


>>Colour differences are often significant but generally small enough to be dealt with in post or by >>the auto-white balance feature of video cameras.

In my experience, colour shifts between eyes which are based on either polarizing effects or poor shading consistency in the mirror are localized in various areas of the frame, and highly dependent on what part of the mirror you're shooting through at the moment.

Neither of those is an "overall" colour shift, which is the only thing that can be fixed by a camera's auto-white balance.

>>Two rigs I worked with have near perfect mirrors. Less than 1/6th stop difference and virtually no >>colour shift.

Care to mention whose rigs those are?

Bob Kertesz
BlueScreen LLC
Hollywood, California


Bob Kertesz wrote:

>>colour shifts between eyes which are based on either polarizing effects or poor shading >>consistency in the mirror

I luckily have not yet been confronted with either of those problems. The errors I was reporting were general colour shifts that I have found quite often on mirrors and were general and present on the entire image and also present without polarizing subjects. They cleared quite well by making a white balance on both cameras on a white surface through the rig for shoots where I was using video cameras in video mode. On shoots not using cameras in video mode (SI2K, RED, film), the error was fixed in post and apparently with not too much trouble. These are of course my experiences and the effects may depend on the mirrors used and the subject matter filmed as well. So sure every ones' mileage may vary...

>>Care to mention whose rigs those are?

I have worked with Verduci, Peschke, Derobe, P+S, Tango, and Swiss Rigs. I hope I didn't forget anyone. Of those, I think Verduci and SwissRig had the better mirrors. All were workable.

Best regards!,

Kommer Kleijn SBC
VFX Cinematographer Brussels, Belgium, Europe
Stereography, Motion Control, Digital Imaging, Large Format


Kommer Kleijn wrote:

>> How would you do that Eric? I have been told about a "depolarising frontglass" but have not yet >>seen that in real life,

It works very well.
Eric, the cats been out of the bag for a while.

Graham D Clark,
phone: why-attempt, S3D work phone: why-I-stereo
http://www.linkedin.com/in/grahamclark


Hello,

following the discussion, I have two questions:

1. Where can I get the mentioned "depolarising frontglass" or is it a part of the mirror? Who did you use it specifically?

2. Is there a way to "measure" the mirror to obtain comparable data for a standardized evaluation of the quality of the mirror?

Warm regards

Christian Meyer, bvk
DIT & Stereographer
Munich, Germany
tel. +49.89.447 693 72
cel. +49.177.258 7294
skype: chrime73


Kommer,

Would you know what the mirror coating is on the mirror's you have found to be better i.e. no colour shift. It seems the formulae for the reflective coating for the specialist mirrors needed for 3d work to maintain 50/50 split with no colour shift may not be an off the shelf product.

Dave Blackham
Esprit Film and Television
UK


 

Dave Blackham wrote:

>>Would you know what the mirror coating is on the mirror's you have found to be better i.e. no colour >>shift.

No I don't know. I am not that much into making mirrors, I am more into using them. What I did note however is that these "better" mirrors do have a particular visual appearance: they appear slightly yellowish to the eye and they seem light - the mirrors look a bit more transparent to the eye than some others and also a bit more transparent than one would maybe expect for 50/50.

Best regards!,

Kommer Kleijn SBC http://www.kommer.com
VFX Cinematographer Brussels, Belgium, Europe
Stereography, Motion Control, Digital Imaging, Large Format