Toronto's Saturday Star made headlines the other day when they published most of their photographs and advertisements in 3d. After reading through the paper I began to wonder how to make one of these 3d pictures. I hit the net running. One site was a little helpful -- they suggest that you use Photoshop (or your other favourite photo editor) and simply shift the red-channel of your photograph to make it 3d. I don't know if they look at their examples, but the photographs do little more than suggest 3d.
What is an anaglyph?
Anaglyph photography fools each eye into thinking that it sees a separate view of the same scene -- like what happens in the 3d world. Your brain combines these two separate views to form a 3d scene in your mind. To give your eyes two separate perspectives of a single scene we need a way to control what each eye sees.
To control what each eye sees, we print both scenes on the same piece of paper (or monitor). Then, we view the scene through filters so that each eye only sees one of the scenes. In anaglyph photography, the filters are red and blue tinted glasses: the left (red) eye sees all colours but red, and the right (blue) eye sees all colours but blue.
Toss the theories
Theory aside, it all comes down to one fact: the farther right you shift the red channel of an object in a 2d photograph, the closer your eyes brain will think the object is.
Now, there are good and bad ways to do this.
The bad way
The worst way (as I found out) is to simply shift the red channel of the whole photograph. In this case, you have a scene that suggests 3d -- but really nothing more.
The best way
The best way to create an anaglyph is to take two pictures of a scene -- from two different angles -- and merge them together. However, we're not talking about creating an anaglyph from scratch, we're talking about creating one from a 2d photograph!
The good way
I've found the best way to create an anaglyph from a 2d photograph is by hacking the photo's red channel. I will give my example as though you use Adobe Photoshop, but feel free to use any other editing app that has these tools. Here we go:
1) Use File | Open to import a picture.
2) Use Image | Mode | Grayscale to convert your picture to grayscale. It's possible make a colour picture into a 3d one, but it's quite a bit less effective. What happens is that you get a lot of "sorta-reds" and "sorta- blues" that confuse your eyes and brain. When Adobe asks, Of course you want to discard the colour information!
3) Use Image | Mode | RGB Color to really confuse the gremlins inside Photoshop. We do this so that we can get the red channel back to play with.
4) Near the bottom-right of your screen, click on the "channels" tab right beside your "layers" tab. Next, click the "Red" channel.
5) Now, figure out what you want to stick the farthest out of your picture. Use the lasso tool (or the magic-want if you're lucky to have it be that smart) to select this object.
6) To make this object jump out at you, select the "move" tool and nudge your selection to the right. If you go too far, your eyes will have a hard time focusing -- so press the right-arrow key about 8-10 times.
7) You might notice an unsightly gap that your object has left behind. This gap will look strange when you view the whole picture again, so you should fill it with a colour similar to its surroundings and use the smudge tool to even it out. Accuracy isn't hugely important -- you just don't want any glaring differences between the red channel and all of the others.
If this is the only object you want to add depth to, you're done! Select the RGB channel, throw on your tres-cool glasses, and enjoy. If that's not all then repeat steps 5-7 until you've given depth to a few more objects. In my foosball example, I gave depth to each of the rods (and connected men) in the picture.
Make your photos even better
I have three reccomendations to help make your photos even more effective:
1) Keep a non-shifted reference. If you red-shift every object in your photograph, your eyes have trouble establishing a reference point. To combat this, keep parts of your picture completely untouched. In my foosball example, I leave the ground untouched.
2) Use photos that naturally suggest perspective. Some photos are better suited for this conversion than others. If a picture already has a lot of 3d visual cues (definite foreground, mid-ground, and background,) it probably makes a good 3d anaglyph.
3) Use variable red-shifting. This is mildly advanced, but stick it through and we'll be all-right.
As I mentioned earlier, the more you shift an object's red channel to the right, the closer it appears. Our current technique is great if we want an object to sit a uniform distance away from us, but what if we want it to come toward us like a star-wars scroller?
Photoshop has a great tool called the "Skew" to help us with this. After you lasso your object, select Edit | Transform | Skew. If you want the bottom of your object to appear closest, drag the two bottom corners to the right -- depending on how drastic you want the effect to be. If you want the right-hand side of your object to appear closest, drag the right-two corners even more to the right.
When you press "Enter" to apply the transformation, Photoshop effectively red-shifts the bottom of the object more than the top -- so the object's bottom appears closer than its top.
Of course, don't forget to fill and smudge the part that photoshop clears.
Don't forget your teacher :)
I'd love to hear your feedback. Even more so, I'd love to see what you've done! If you create pictures that you're proud of, feel free to email them to me.