Has anyone attempted 3D pano?  

All things panoramic.
An open forum for every topic on photography, panoramas, IT or related.
no avatar
Phil Howard
Member
 
Topic author
Posts: 69
Likes: 0 post
Liked in: 0 post
Joined: Sat Oct 04, 2008 6:19 pm
Info

Has anyone attempted 3D pano?

by Phil Howard » Tue Oct 21, 2008 7:58 pm

Has anyone attempted to create 3D panoramic, or at least very very wide, photos? I can see many technical challenges in it (and that gets my interest). I would envision a double-mount pano head being rigged together to do it, using 2 cameras operated together under a common remote control. The image sets might have to be processed separately with great care to ensure they are "stitched alike". Viewing a 3D pano might be interesting. It can't be a wide display, so it might have to be limited to a synchronized panning animation of 2 closely spaced images that can be viewed properly for 3D.

no avatar
mediavets
Moderator
 
Posts: 16415
Likes: 2 posts
Liked in: 130 posts
Joined: Wed Nov 14, 2007 2:12 pm
Location: Isleham, Cambridgeshire, UK.
Info

by mediavets » Tue Oct 21, 2008 8:07 pm

Andrew Stephens
Many different Nodal Ninja and Agnos pano heads. Merlin/Panogear mount with Papywizard on Nokia Internet tablets.
Nikon D5100 and D40, Sigma 8mm f3.5 FE, Nikon 10.5mm FE, 35mm, 50mm, 18-55mm, 70-210mm. Promote control.

no avatar
Phil Howard
Member
 
Topic author
Posts: 69
Likes: 0 post
Liked in: 0 post
Joined: Sat Oct 04, 2008 6:19 pm
Info

by Phil Howard » Wed Oct 22, 2008 4:32 am

That first one has some nice panos. When I click on one (I tried a few) I get a double image set. But there is no animation to it. It says it runs using JavaScript, the the script is absent and there are no references to it in the HTML. 3D viewing works, but there is no panning.

The 2nd one gave some more ways to view 3D, but no more panos.

I'm working on a JavaScript based panner. I'll see what I can do to make sure it can also do a 3D pair for parallel or cross-eye viewing (it will at least need to syncronize the panning).

User avatar
DrSlony
Moderator
 
Posts: 1893
Likes: 0 post
Liked in: 0 post
Joined: Sat Nov 03, 2007 6:30 pm
Location: Sweden
Info

by DrSlony » Wed Oct 22, 2008 9:52 am

Those java(script) viewers dont work well, most of the time the two halves were way out of synch since they loaded one after the other, not at once.

I like stereograms, I made one or two stereogram panos but they were not 360°, just 4 or 6 shots each and rendered rectilinear iirc. Here is one: http://www.autopano.net/forum/t4329-poland,zielona-gora-a-sunny-and-rainy-day

Stereogram 360°x180° panos sound like a wonderful thing, but they would need a solid player and an expensive camera setup. Those compact camera panos work, but they are capable of no more than a demonstration of the possibility of making stereo panoramas, the quality they offer is simply #### compared to two DSLRs. Perhaps some day when people grasp what a normal panorama is and what a normal stereogram is I will invest in two identical DSLRs :)

One thing I don't understand is how you could overcome parallax errors when using two cameras side by side... I imagine they are mounted on an arm and the arm is mounted to the tripod in the center, like this:
Code: Select all
X_____X
   |

Turning this whole thing around will result in huge parallax errors and therefore stitching errors in each pano. To fix parallax errors, I imagine the cameras could be mounted on little swivels and those attached to the arm, like this:
Code: Select all
X     X
|_____|
   |

and instead of turning the whole thing, you would turn each camera around its own nodal point axis individually, but this would make it impossible to shoot when both cameras are pointed to the left and right (one would be in the way of the other)... so how is it done? Or are parallax errors unavoidable and this only works for places where there are no objects anywhere near the camera?
Last edited by DrSlony on Wed Oct 22, 2008 9:55 am, edited 1 time in total.

no avatar
Phil Howard
Member
 
Topic author
Posts: 69
Likes: 0 post
Liked in: 0 post
Joined: Sat Oct 04, 2008 6:19 pm
Info

by Phil Howard » Thu Oct 23, 2008 8:44 am

DrSlony wrote:Those java(script) viewers dont work well, most of the time the two halves were way out of synch since they loaded one after the other, not at once.

I'm now including stereo capability in the panner animation JavaScript I'm writing. It will pan both photos in sync by forcing them to always have the same offset in the panning. Then it is up to the photographer to be sure they are geometrically in sync.

DrSlony wrote:Stereogram 360°x180° panos sound like a wonderful thing, but they would need a solid player and an expensive camera setup.

That's what I hope my panner code will be, solid enough to use with stereo panos.

DrSlony wrote:Or are parallax errors unavoidable and this only works for places where there are no objects anywhere near the camera?

Stereo is a parallax "error" anyway. If you are looking for a perfect solution, then shoot 3600 stereo shots with a rotating camera pair (rotating the assembly of 2 cameras) at 0.1 degree increments and make a movie. The challenge will be displaying 2 movies in sync. That would be fixed by editing them into a single bi-view movie (or whatever 3D viewing technology is used).

Consider a lens design where 2 lenses actually cross each other so their entrance pupils are at the same point. Either you have to play optical tricks to get the entrance pupil to be optically/apparently ahead of the lens (and that would result in a very narrow field of view and a very small aperture) or build it as a single lens assembly, for 2 cameras, where they share the same nodal point (think of 2 lens barrels crossing through each other). So with this kind of lens (if it is even practical at all, which I doubt), you really don't get the correct parallax error that makes stereoscopic viewing work.

So you must have entrance pupils at offsets from each other, orthogonal to the direction of view, to make stereo work. For a parallax-free pano, you want to have each camera individually rotate its entrance pupil at exactly the same point. These two requirements are in conflict. So at best it will have to be a compromise. For panos that are not too wide, this is not much of a problem (though a wide picture sure makes cross-eye viewing hard to do). For 360 degree panos, I believe you're going to have to live with parallax error and stitch around it somehow. What I worry about is the stitching compromises will impact the stereoscopic perspective.

My guess is the best starting point is to rotate around a point exactly half way between the entrance pupils (a stereo head on top of a pano head). Then you have to decide how much or how little offset to use. More offset enhances the stereo view at the expense of stitching problems. Less offset makes stitching easier at the expense of the stereo effect. I suspect the narrower field of view lenses will work best with this (let wider angle be vertical), along with a lot of shots (72x2 at 5 degree intervals).

I need to get a sacrificial wired remote for my 450D and take it apart to see just exactly how it works, and how I might rig it to operate 2 cameras at the same time. That or see if someone has documented how it works. I'm hoping it is just a simple contact closure for each of 2 contacts (half press and full press) that I could wire to the cameras in parallel through a "Y" adapter.

User avatar
John_Sauter
Member
 
Posts: 173
Likes: 0 post
Liked in: 0 post
Joined: Sat Jul 28, 2007 4:59 pm
Location: New Hampshire, USA
Info

by John_Sauter » Thu Oct 23, 2008 1:26 pm

To duplicate human stereo vision, you will need two kinds of rotation for your stereo camera. To simulate head movement, the entire assembly rotates around a point well behind the nodal point, and the parallax errors are part of how our brains judge distance. To simulate eye movement, the two cameras each rotate about their nodal points, but the rotation is limited to plus or minus 90 degrees from the head direction, and at large left and right deflections the view of one eye is blocked by the nose.

If you want a single turn control on your display, have it move the eyes up to 90 degrees in 200 milliseconds and for motions of more than 45 degrees, or if the look direction remains steady for more than 500 milliseconds, move the head at a rate of 180 degrees in 500 milliseconds, using eye motion to maintain the look direction.
Last edited by John_Sauter on Thu Oct 23, 2008 1:28 pm, edited 1 time in total.

no avatar
GURL
Member
 
Posts: 2943
Likes: 0 post
Liked in: 0 post
Joined: Tue Dec 06, 2005 1:57 pm
Location: Grenoble
Info

by GURL » Thu Oct 23, 2008 2:49 pm

John_Sauter wrote:the parallax errors are part of how our brains judge distance.

I would raher say "paralax changes" to make clear that this helps our brain to understand what is around us. I suppose that babies are spending a lot of time to learn how to put that to good use when they are in their cradle :)

Parallax changes occur when we move ou head slightly forward/backward or walk. A movie camera can render that using a dolly.
Last edited by GURL on Thu Oct 23, 2008 2:53 pm, edited 1 time in total.
Georges

no avatar
BeeZed
Member
 
Posts: 43
Likes: 0 post
Liked in: 0 post
Joined: Thu Jun 07, 2007 10:38 pm
Location: New Zealand
Info

by BeeZed » Thu Oct 23, 2008 8:06 pm

Phil wrote:I need to get a sacrificial wired remote for my 450D and take it apart to see just exactly how it works, and how I might rig it to operate 2 cameras at the same time. That or see if someone has documented how it works. I'm hoping it is just a simple contact closure for each of 2 contacts (half press and full press) that I could wire to the cameras in parallel through a "Y" adapter.

I am pretty sure the 450D is the same as the 400D etc, so you can definitely use a 2.5mm plug and a Y connector.

This page has the basic wiring schematics for a single camera. Worked on my 400D.

no avatar
Phil Howard
Member
 
Topic author
Posts: 69
Likes: 0 post
Liked in: 0 post
Joined: Sat Oct 04, 2008 6:19 pm
Info

by Phil Howard » Fri Oct 24, 2008 5:41 am

BeeZed wrote:I am pretty sure the 450D is the same as the 400D etc, so you can definitely use a 2.5mm plug and a Y connector.

This page has the basic wiring schematics for a single camera. Worked on my 400D.

They all use the Canon RS-60E3, so that seems to be the way it is wired. My next curiosity is if the connector is merely wired in parallel with the actual shutter button. If it is, then cross connecting 2 cameras on that 2.5mm port would let either fire both.

User avatar
fma38
Moderator
 
Posts: 5850
Likes: 2 posts
Liked in: 2 posts
Joined: Wed Dec 07, 2005 6:21 pm
Location: Grenoble, France
Info

by fma38 » Fri Oct 24, 2008 7:31 am

Mmmm, there is a risk connecting 2 cameras, we don't know how it is done inside. I suggest you use a relay with 2 sets of contacts, to avoid cross-wiring shutters...
Frédéric

no avatar
Phil Howard
Member
 
Topic author
Posts: 69
Likes: 0 post
Liked in: 0 post
Joined: Sat Oct 04, 2008 6:19 pm
Info

by Phil Howard » Fri Oct 24, 2008 11:52 pm

True, we don't know what is done inside. But we do have it narrowed down to contact closure allowing current to flow that prepares the shot and/or takes the shot. Obviously some voltage would have to be applied. I would also suspect that even when the switch is closed, that voltage is current limited very significantly, so it would not be putting the two batteries in parallel (the most dangerous possible risk I see). Possibly, however, the voltages may be balanced across a voltage drop circuit that, when 2 are connected in parallel, results in more residual current than usual, resulting in both cameras believing the connection to be closed. If I were designing a camera like this, I'd be sure the current levels involved would be as low as possible, so the power efficiency is overall better. But that could mean a very tiny current that would trigger the shot. So it could be that instant you connect both cameras together, they both start firing off shots as if you held down the button all the way.

However, a relay circuit to isolate the cameras would also require its own source of power. This would make a more complicated device than just a simple Y-adaptor. I'd hope that if some isolation is even needed, it could be done with a few diodes at most. I guess I should get a volt-current-ohm meter and take some measurements.

no avatar
BeeZed
Member
 
Posts: 43
Likes: 0 post
Liked in: 0 post
Joined: Thu Jun 07, 2007 10:38 pm
Location: New Zealand
Info

by BeeZed » Sat Oct 25, 2008 8:37 pm

Back with more third hand information: This guy says the camera circuit is 3.3v and 0.5 mA in a 350D.

I also had a link to a stereo rig description in the my archives which uses a hard wired "y" remote cable on a EOS.

Finally, I emailed a friend into 3D photography, and he has seen a demonstration where a single cable is used with 2.5 mm male at each end, and one camera triggers the other. Members of his 3D club regularly use this method without problems on at least 400D Canons.

Still, you might wish to take some measurements, or install some diodes to protect your camera.

no avatar
foundation
Member
 
Posts: 276
Likes: 3 posts
Liked in: 2 posts
Joined: Mon Jan 15, 2007 4:43 am
Info

by foundation » Mon Oct 27, 2008 4:45 pm

Another direction for simultaneous shutter release would be to use the usb ports and the sdk and some software. Software such as this http://www.sabsik.com/Cam2Com/
or this
http://www.breezesys.com/MultiCamera/index.htm
or
http://sourceforge.net/projects/multican/

Probably more out there of varying costs and complexity

For a simple 2 way adapter for a 3d rig the software is probably overkill but I thought I'd mention it

User avatar
Paul
Member
 
Posts: 778
Likes: 0 post
Liked in: 0 post
Joined: Sat Aug 30, 2008 4:46 pm
Location: Bonn, Germany
Info

by Paul » Mon Oct 27, 2008 6:14 pm

on sourceforge there is gphoto2 too ...

Simple GUI frontend for gphoto2 with features useful in scientific research - support for simultaneous shooting and post processing.
http://sourceforge.net/projects/grabui/
Paul

close, but no cigar ... ... ...

no avatar
mediavets
Moderator
 
Posts: 16415
Likes: 2 posts
Liked in: 130 posts
Joined: Wed Nov 14, 2007 2:12 pm
Location: Isleham, Cambridgeshire, UK.
Info

by mediavets » Mon Oct 27, 2008 6:24 pm

Phil Howard wrote:I need to get a sacrificial wired remote for my 450D and take it apart to see just exactly how it works, and how I might rig it to operate 2 cameras at the same time. That or see if someone has documented how it works. I'm hoping it is just a simple contact closure for each of 2 contacts (half press and full press) that I could wire to the cameras in parallel through a "Y" adapter.

How about using one IR remote to trigger both cameras?
Last edited by mediavets on Mon Oct 27, 2008 6:24 pm, edited 1 time in total.
Andrew Stephens
Many different Nodal Ninja and Agnos pano heads. Merlin/Panogear mount with Papywizard on Nokia Internet tablets.
Nikon D5100 and D40, Sigma 8mm f3.5 FE, Nikon 10.5mm FE, 35mm, 50mm, 18-55mm, 70-210mm. Promote control.

User avatar
fma38
Moderator
 
Posts: 5850
Likes: 2 posts
Liked in: 2 posts
Joined: Wed Dec 07, 2005 6:21 pm
Location: Grenoble, France
Info

by fma38 » Mon Oct 27, 2008 6:32 pm

What a great idea :)
Frédéric

no avatar
hankkarl
Member
 
Posts: 1284
Likes: 0 post
Liked in: 0 post
Joined: Tue Feb 21, 2006 5:32 pm
Location: Connecticut, USA
Info

by hankkarl » Tue Oct 28, 2008 4:21 pm

I haven't had much luck with IR--sometimes it works, sometimes it doesn't. If you are using a two-camera setup, Pocket Wizards would be better, or just search the net for a two-camera remote trigger hack--there are several out there.


Who is online

Users browsing this forum: No registered users and 2 guests