Patent: Canon stereoscopic lens for the RF mount

Mar 15, 2019
3
5
This would be a dream for me. The sensor may be used in halves, so that you have 2 portrait oriented shots (these could be cropped to landscape with a large 45 mp sensor like the R5). One shutter release so that you have same metering, shutter speed etc. in one frame. I have also photographed with 2 cameras on a rig that try to get same settings. It is way too finicky and problematic. It must be perfect. If you have a single leaf or small subject blowing in the breeze and out of sync the mind cannot process it and it makes you nautious. I have used a lens that splits the image via mirrors but it was cheaply made and results were poor. A true quality stereo lens would definitely get me back into stereo photography which I left because it was.too finicky and current stereo cameras were essentially point and shoot. Stereo photo on a great quality on my R5 would be phenomenal.
 
  • Like
Reactions: SwissFrank

Kit.

EOS 5D Mark IV
Apr 25, 2011
2,011
1,355
I suppose if both cameras start at the same time (via a split manual release) then a video with time-code could keep them in sync later for post combination.

Of if the cameras were as close together as possible on a nice tripod then you could get stills or video with moving subjects just like you would with a regular camera on a tripod.
I'm concerned about depth perception. If the subject moves between its "left" and "right" exposures, a distance to it will appear wrong and potentially inconsistent.

If you project both the "left" and the "right" images on the left and the right halves of the same sensor with a vertically moving shutter, that won't happen.

Otherwise you would need to synchronize the shutters of two cameras very precisely.
 
  • Like
Reactions: usern4cr

SwissFrank

from EOS 1N to R
Dec 9, 2018
450
200
I'm concerned about depth perception. If the subject moves between its "left" and "right" exposures, a distance to it will appear wrong and potentially inconsistent.

If you project both the "left" and the "right" images on the left and the right halves of the same sensor with a vertically moving shutter, that won't happen.

Otherwise you would need to synchronize the shutters of two cameras very precisely.
I would guess each lens casts a small image circle, so you have two 90-degree fisheyes in your one 45MP exposure.
 

SwissFrank

from EOS 1N to R
Dec 9, 2018
450
200
I don't know why my takeaway seems different from many others here.

This is clear to me to be a VR lens. Therefore the two input lenses have to be a distance apart of two normal eyes, not wider.

The two lenses will have the same angle of view, and should have as near as possible all other features like bokeh, coma, any other distortion you can think of, with the difference that one of them in effect is longer (enabling it to jut far farther out to the left, then the right lens does to to the right). That's probably not a easy trick to do in lens design.

It is clear to me that it will make two circular images in one frame at one time. You will then post-process to peel those images out into left and right images, and I'm guessing, also to choose your "35mm film equivalent" effective focal length, say 35mm or 50mm. By having 35mm, you can actually go as wide as about (I think) 21mm, with some loss of resolution.
 
  • Like
Reactions: usern4cr

usern4cr

EOS RP
CR Pro
Sep 2, 2018
444
411
Kentucky, USA
I'm concerned about depth perception. If the subject moves between its "left" and "right" exposures, a distance to it will appear wrong and potentially inconsistent.

If you project both the "left" and the "right" images on the left and the right halves of the same sensor with a vertically moving shutter, that won't happen.

Otherwise you would need to synchronize the shutters of two cameras very precisely.
Again, having a manual remote switch (Canon RS-80N3 which I have) and then electrically (and safely) spliting into 2 plugs for the 2 cameras, would start the exposure at the *same exact time* and that remote press can be mechanically held "on" until you release it (again happening to both cameras simultaneously)!
 

usern4cr

EOS RP
CR Pro
Sep 2, 2018
444
411
Kentucky, USA
baseline would be far wider than human's eyes and thust not realistic. I imagine these two lenses are an average eye distance apart.
Not really. Hold your R5 to your face and look into a mirror. If the center of your eyes were at the center of 2 lenses then they'd entirely fit in the width of the R5. Now you can make them slightly more separated as needed to allow for the desired max. width of the lenses you have (either fixed on the camera or interchangeable). But it can definitely work.

Now selling it profitably is another issue! But you have an entire planet of people who've never had access to a "binocular camera" before, so look at the market potential! Ca - ching!
 

usern4cr

EOS RP
CR Pro
Sep 2, 2018
444
411
Kentucky, USA
Some people here think that you have to have the 2 lens centers the same distance apart as the human eye for you to comfortably see the result. This is not at all true! Grab a big size pair of binoculars and you will instantly see that the centers of the 2 lenses are far wider apart than the human eyes. Heck, I've owned a pair of Fujinon 150mm diameter occular, 25x power binoculars (which are HUGE! and must be mounted on a heavy-duty telescope tripod) and those centers are waaaay further apart than my eyes, and everything looks *perfect*. ;)
 
  • Like
Reactions: TominNJ and RaPhoto

SteveC

R5
CR Pro
Sep 3, 2019
1,719
1,589
Some people here think that you have to have the 2 lens centers the same distance apart as the human eye for you to comfortably see the result. This is not at all true! Grab a big size pair of binoculars and you will instantly see that the centers of the 2 lenses are far wider apart than the human eyes. Heck, I've owned a pair of Fujinon 150mm diameter occular, 25x power binoculars (which are HUGE! and must be mounted on a heavy-duty telescope tripod) and those centers are waaaay further apart than my eyes, and everything looks *perfect*. ;)
I will guess that the more you zoom, the farther apart the "eyes" should be, otherwise it will just look flat.
 
  • Like
Reactions: usern4cr

Kit.

EOS 5D Mark IV
Apr 25, 2011
2,011
1,355
Again, having a manual remote switch (Canon RS-80N3 which I have) and then electrically (and safely) spliting into 2 plugs for the 2 cameras, would start the exposure at the *same exact time* and that remote press can be mechanically held "on" until you release it (again happening to both cameras simultaneously)!
In mirrorless, I would naturally expect shutter lag instability up to 1 EVF refresh cycle.
 
  • Like
Reactions: usern4cr

SteveC

R5
CR Pro
Sep 3, 2019
1,719
1,589
I expect dedicated hardware would have two lens mounts and two sensors--and ONE signal to trip both shutters.
 
  • Like
Reactions: usern4cr

melgross

EOS RP
Nov 2, 2016
614
401
I have to assume Canon knows what they’re doing. I imagine they have figured the problems out for whatever use this is intended for.
 

SwissFrank

from EOS 1N to R
Dec 9, 2018
450
200
This is not at all true! Grab a big size pair of binoculars and you will instantly see that the centers of the 2 lenses are far wider apart than the human eyes. Heck, I've owned a pair of Fujinon 150mm diameter occular, 25x power binoculars (which are HUGE! and must be mounted on a heavy-duty telescope tripod) and those centers are waaaay further apart than my eyes, and everything looks *perfect*.
Those are ALL telephoto applications. The lens in question is designed for normal angles, and my comment was needing a normal baseline was about normal angle lenses, like the one we'ere talking about. This isn't binoculars, clearly.
 
  • Like
Reactions: usern4cr

kten

EOS M6 Mark II
Oct 3, 2015
58
55
as swissfrank said for VR and near objects the further out from average IPD the cameras are the bigger the problems for viewers. Both myself and many I know have complained a lot of the VR video made to look impressive would be, until something comes in frame close to the camera and then it is headache time and just makes it pointless and shows up all weakness of that as a medium. No way I'd watch somethign featurelength like that. Problem is the double cam brackets generally can't get them close enough like with small cube bodies or vr dedicated cameras. To mount base to base on DSLR style bodies is not that simple either, although can work requires mounting solutions I've never seen and the heights would need staggering to keep the sensors level relative to each other. Also you would need to mess with signal flipping things although that isn't hard this could even theoretically be fed directly to a HMD. I'm guessing the output from this looks exactly what a valve index outputs to OBS (ie. side by side fisheye) and that can be fed direct in to most VR players which already have an option for horizontal split and do it for you. This solves all that and basically allows content producers to make stuff that looks correct like the VR only tools currently do.
 
  • Like
Reactions: SwissFrank

canonnews

EOS RP
CR Pro
Dec 27, 2017
793
1,240
Canada
www.canonnews.com
I would guess each lens casts a small image circle, so you have two 90-degree fisheyes in your one 45MP exposure.
yeah we showed this on our writeup for the patent.


it exactly fits a full frame with 8.75mm image heights.
stereostropic.jpg
 

Antono Refa

EOS R
Mar 26, 2014
1,100
281
I have to assume Canon knows what they’re doing. I imagine they have figured the problems out for whatever use this is intended for.
Why do you assume this patent is going to turn into a real lens, rather than just protect IP?
 

Daner

AE-1 Program
My favorite rock guitarist / Ph.D. astrophysicist might be able to tell us all a thing or two about stereo photography:

 

TominNJ

EOS M6 Mark II
Mar 14, 2015
72
62
Well, I didn't expect that much reaction to my post. In fact, I have seen people put 2 cameras on a bar and take high quality stereo photos in the past, I'm sure with a single shutter release hooked up to both of them.

I've always loved the View-Master slides & viewer, and wished there was a great mainstream stereo camera I could use, and envisioned it as a double lens & mount on the far left & right of a single body, and a separate viewer of some sort to view the result later. But maybe you could have both in a single body by putting the lenses at the top left & right and having 2 viewfinders with adjustable eye spacing and a cutout in the bottom center for your nose. That basically becomes a binocular camera, which I would dearly love to have!

It seems odd to try to pack 2 lenses into one lens body that funnels the light from 2 views which must be kept separate onto a single sensor, which I assume is done by making their images so small that they fit separately on the left & right sides of the sensor, with lots of wasted sensor space around both of them.

I guess the reason that this patent was done is that it is economically possible to have a single lens made (no matter how great the complexity and loss of IQ) to fit on a currently produced mainstream camera.

You know, Canon does make stellar image stabilized binoculars. Why don't they just put 2 sensor&EVFs on them and come out with a line of binocular cameras, with some having a wide angle of view or possibly with zoom ability. That's what I'd buy! :D
‘Years ago, Pentax made a stereo prism gizmo that screwed onto the front of a 50mm lens. It captured two side by side images on slide film. The slides could then be viewed in the viewer that they also made. I used the thing a couple times. It worked better on closer objects than on distant ones. I still have the stereo slides I took of our dog. They’re pretty cool.


I’m not sure how a similar device could be used with a digital camera other than using a VR headset or printing the images for viewing on one of those old stereoscopes.
 
  • Like
Reactions: usern4cr