I recently had the privilege of hosting a Q&A with Ben Miller, from double act Armstrong & Miller, at the Brighton Science Festival. There was a man there called Mark Bennett taking 3D pictures of us. He has very kindly let me have the originals so I thought I'd write a blog post about 3D photography using these images as examples. Here is a previous post about how 3D cinema works.
How do we see in 3D
Being able to see in 3D just means being able to perceive how far way things are. We do this by interpreting the little differences between the images we get from our left eye and the images we get from our right eye. The calculations are made subconsciously. All we're aware of consciously is the ability to gauge distance.
So how can you make 3D images? All you need is some way of taking a picture for your left eye and a second picture for your right eye and then somehow getting the correct image into the correct eye. First of all, how do you take the pictures?
3D cameras
The easiest way is to strap two cameras together! You need to make sure that the camera lenses are the same distance apart as the average pair of eyes. That way the little differences between the two images you get will be the same as the little differences your eyes get in real life. You also need to make sure they take a picture simultaneously for live action shots which can be tricky. Some cameras come with a socket for connecting an external trigger. With a bit of rewiring you can make an external trigger operate 2 cameras at the same tame. Here's a guide to making a 3D camera from 2 cameras.
This is pretty much what Mark Bennett had I think (with superior cameras) but his trigger wasn't working on the day so some of the images will be a little peculiar when viewed in 3D. Here's a collection of his work that shows off the technique a bit better (follow the Next link to see more).
Here's an example of a pair of images from the Q&A:
So now you've got your images, how do you view them in 3D?
Turning two images into a 3D experience
All you have to do is get one eye seeing one of your images and the other eye seeing the other. There are a few ways of doing this. In a 3D cinema for example they project the two images using two different kinds of polarised light. The glasses you wear filter these two kinds so you get the correct image in the correct eye. Complicated setups exist for viewing 3D still images too, like this ViewMaster from my childhood:
But I want you to be able to experience me and Ben Miller in glorious 3D without having to buy anything or set anything up and we can do that by tricking your brain...
3D pictures are a trick of the mind
Before your brain is able to work out depth from the two images sent from your eyes it has to work out which bits of the two images match up (this cat my left eye is seeing must be the same cat my right eye is seeing, and so on). If you've got a really varied field of view, it's impossible to go wrong. But if you present your eyes with two very similar things side-by-side you can trick brain into thinking they are one and the same thing. Take a look at the two images below:
To trick your brain, you need to get the two images to overlap and in this case you do that by going cross eyed. If you can control your eyes easily then just go crossed eyed until the two images exactly overlap. You might need to tilt your head slightly from side to side until you have them lined up but at that point they should snap into place and suddenly you won't need to be actively crossing you eyes anymore; your brain will be happy with the deception, it will have matched up one image in one eye with the other image in the other and the trick is complete.
If you have trouble going cross eyed try putting your finger on the image on the screen then slowly moving your finger towards your nose. Keep focused on your finger but be aware of the picture behind it. At some point they will overlap. Again you might need to tilt you head a little one way or the other to get them overlapping exactly but at that point you should be able to focus on the picture instead of your finger without your eyes going back to normal.
Here's what your eyes are doing:
Some people find it easier to do the opposite of going cross eyes, to relax their eyes so they move apart. This is sometimes called cock eyed. For those people, it's better to have the two pictures the other way round like they are below:
To view this pair correctly you need to be looking through the image, beyond your computer screen. If you have a reflective screen, try looking at your reflection while focusing on the image. You're always trying to get the two pictures to exactly overlap so remember to try tilting you head a bit if one looks slightly above the other.
In this set up your eyes are doing this:
It might also help to put a sheet of paper between the two images and look either side of it (nose resting on the edge of the paper).
There is another way to get these two images into the correct eyes. It requires that you have a pair of those old 3D glasses to hand, the ones with one red filter and one blue (or sometimes green) filter. I'm guessing that a lot of you might have a pair so for the lucky few, here we are again in red/blue 3D:
This was really easy to make. First you make the two images black and white. Then, you put a red filter on the left eye image and a blue filter on the right eye image. You then add the two images together. Now when you put the 3D glasses on, the blue lens will filter out all the light from the red image so you only get the blue in that eye and the red lens will filter out all the blue light so you only see the red in that eye. Here's what's happening in this setup:
I'd love to be able to render these images as an autostereogram, or Magic Eye picture, because I was obsessed with them when they first came out. It's why I can see cross eye and cock eye stereograms so easily now, I had loads of practice with Magic Eyes. Unfortunately I can't figure out how to turn a two-image stereogram into an autostereogram. All I need to be able to do is extract the depth information from the two images. If anyone can point me in the right direction let me know in the comments.
One final way of rendering these pictures in 3D relies not on getting one image in one eye and the other in the other but exploits another way that our brains perceive 3D.
As we move around and our viewpoint changes we gain useful information about the relative positions of objects in our field of vision. That's because they move over each other as we move and our brains can calculate depth information from that.
To exploit this I just need to create an animation that moves back and forth between the two images:
Do you know any other ways to render images in 3D? Let me know in the comments.
You might also like these recursive images and this post about 3D cinema.
