Walkthrough: Producing a 3d Viewable Image

Who hasn't wanted to be able to shoot a 3D image - especially of something dimensional and visually interesting? As it happens, it's really easy - and it's a great way to explore making your photographic art a bit more interesting.

For this walkthrough, I will use a photo of an artists' model on a simple paper sweep background, lit by two softboxes to create an overall image with a sense of light and depth. That'll make our 3D effect even more apparent!

Tech and Tack

To do a 3D image like this, you don't need much in the way of special equipment. There are cameras that shoot "split" images but they are expensive and of limited use for non-3D applications. If you're shooting subjects that don't move around much (landscape, still life, architecture, artists' models) then you can get away with using what's called a "slider bar." The idea of a slider bar is to mount your camera so that it can be moved from side-to-side the same distance as the distance between your eyes. Without going into a long lecture about depth perception, I'll simply state that: The brain interprets how an object is from you, based on the intersecting angle between your eyes when they are focused on it. 3D depth perception is made up by your brain sampling a lot of these little distance measures and integrating them into a world-view that includes distance. So, we "hack" the brain by making the eyes focus on an apparent distance that is incorrect, and the brain sorts it out from there.

If you're interested in obtaining pre-build slider bars, they are available from a variety of sources on the web, including here. Building your own is fairly easy if you buy aluminum rail components from woodworkers' supply or a hobby store. Whatever you use, your slider bar should be strong enough to give a predictable, vibration-free movement.

The slider bar mounts on your tripod and your camera mounts on the slider bar so that the whole camera can move from side to side without changing angle. The slider must be level, or the resulting image will be hard/weird to view. (By the way, there are all kinds of fun possibilities for deliberately creating images that have contradictory 3D depth cues...)

(JVC video camera standing in for my S3, mounted on my tripod with slide bar)

Setting the stage

For every 3D photo you want to produce, you're going to shoot two images. One will be the left eye's image, and the other the right eye's image. If you're working with an artist's model, you'll need to make sure it's someone who understands "don't move!" When you're ready to start shooting your images, There are really only three things I think you need to worry about:

If you want to shoot images in portrait orientation you have two choices - buy a right-angle bracket mount for your camera, or shoot horizontal, and crop. Cropping images for 3D is a real pain unless you use the relatively simple trick I'll demonstrate in this walkthrough.

Compositing


(2 frames open in photoshop)

Pick your pair of images and open them both in photoshop. If you're like me, you'll frequently get confused about which one is the "right eye" frame and which is the "left eye" frame. De-confusing yourself is easy! Take a look at the left edge of the photo - if the stuff on the left edge is closer to the edge of the frame, it's your "right eye" frame. You can see how that works by closing your eyes alternately and watching your world-view shift left and right.

In our example, I picked an image where there is crud on the edges of the frame that I'd like to eliminate (although it'd make a good depth cue!) for artistic reasons. Rather than trying to crop both frames exactly the same way, I cheat and use layers.


(the two images superimposed as layers)

Click on one of the frames and select all, (Control-A) then copy it (Control-C). Close the image and click on the other frame. Then paste the first image on top of the second (Control-V). If you bring up your layers menu you'll see that both images are superimposed and you can toggle them back and forth as you see fit.


(Cropping both layers in register)

By cropping the image when the layers are superimposed you don't have to waste any time trying to get the cropping tool aligned pixel-perfect!


(Setting up the document size for the composite)

Next you want to create a working document that is large enough to fit both frames side by side. In the example above I just created a new document and doubled the number of pixels on the horizontal axis. I could just has easily have used "image canvas size 200%" and resized the canvas then moved the two frames around within it. Why did I do it this way? I don't know.


(Aligning the composite)

Now align the two frames side by side in your new document or expanded canvas. The right frame goes on the right and the left frame on the left. At this point you can test to make sure you have it correct by doing the cross-eyed look and seeing how it comes out.


(Flattening and b/w conversion)

Now you can start the clean-up stage of your work. I prefer not to work on the lighting of the image until I have it fully composited. If the two frames are "off" in terms of lighting levels you may wish to try to repair the before you flatten the image (good luck!). In this case I've decided to do a b/w conversion using monochrome channel mixer with 80% red 20% blue.


(levels adjustment)

Adjust the levels control to get the desired tonal range in the image. You'll notice that the original image as it came out of the camera was quite dark. I've taken to shooting my images on the dark side, so that they are less likely to blow out a highlight. When I get the image in photoshop it's easier to "stretch" the luminance range of the image to a desired value than it is to repair a blown-out highlight. The old Zone System mantra "Expose for the shadows, develop for the highlights" does not work for digital photography; in fact the exact opposite: "Expose for the highlight and develop for the shadows." If you think about it, you'll realize that this is because when you're working with film you're trying to capture the correct luminance on a negative which means that the luminance gets inverted during the printing process. Since digital is a "direct to image" process you don't need to compensate for problems that occur between the negative and the final print.


(Final Stage; clean-up and unspotting)

If I'm working on an image that I know is going to be black and white, I usually wait until the image is finished before I go through with healing brush and remove dust spots, loose hairs, backdrop wrinkles, and model zits. The reason I do that is because the conversion process I use (channel mixer -> Levels) often results in an increase in the image's contrast. So if I go through and fix dust spots before the image is "finished" I'll run the risk of popping them back out when I adjust the luminance levels.

For 3D images, be careful to make sure you don't accidentally alter one side more than the other. For example, if you leave a dust spot on one of the image pair and delete it from the other, the viewer will see a black spot "floating off" or "sunk into" the image. On the other hand, that could be a highly amusing effect if you used it deliberately!

Wrapping up


(click here for full-size version)

To view a 3D image like this, you put your face about a foot from the screen, and hold your eyes exactly level with the image. It helps if you can view the image in fullscreen mode so there is nothing else on the screen to distract your eyes. Cross your eyes slightly and you'll see a "ghost" image appear between the two. Adjust how crossed you've got your eyes until the central image is correctly aligned and in sharp focus and it'll appear to be 3D!

Once the final image is completed, I save it, then make a downsampled version for web use. I've found that it's hard for the eye to handle really large images and, since the whole image has to fit on the screen, I tend to keep my images in the 1000 pixel or smaller range (on the long axis). I do all my work on maximum resolution versions because the 3D composites work great if you print them out on a decent printer! I've printed the image above on art-grade rag paper with my Epson 2200 and it gets a great deal of attention at art shows.

mjr.
Bellwether Farm , Morrisdale, Pennsylvania Apr 9, 2006