This tutorial will show you how to make stereo pictures - the kind that are viewed using special glasses. It starts by explaining the workings of the two most common systems, and goes on to describe methods for producing the pictures yourself.
view or download Composite.jpg (61 kb)
I have produced the examples using Softimage 3D and Adobe Photoshop. You can use the 3D package of your choice, and any graphics program that can work with layers.
You can skip this part if you want - you don't need to know it to produce the images.
This is by far the most common method of distributing stereo images. It relies on the fact that a red filter will let red light through, but will block cyan light, and a cyan filter will do the opposite, letting through cyan light and blocking red. Using two slightly offset cameras, one for each eye, two views of the scene are produced. They are converted to black and white, and then one is coloured red, the other cyan.
Then the two are superimposed.
You can see that where the resulting image has both red and cyan information, the result is shades of white which will pass through either coloured filter.
The viewer wears special glasses with one lens coloured red, and the other cyan. The lenses separate the image back into left and right eye parts, and pass their own colour through to the corresponding eye.
It is possible to produce colour images using the red-cyan system. In this case a third central camera position captures a colour picture, and the red and cyan versions are laid over the top. It is much harder to get a satisfactory result using this process however, as the colour layer can tend to interfere with red and cyan layers. There are some ways to help the situation, but they need human intervention and judgement, and are hard to automate. For good examples of colour pictures, see the 3dimages website. They also sell the 3D glasses you need.
The red-cyan system is relatively simple to use, produces images that can be can be printed, displayed on t.v., sent over the Internet, projected etc.
On the down side, colour images can be difficult to produce well, and the effect of depth is less marked than in many other systems, and some people have reported feeling unwell after using the coloured glasses for extended periods.
This system uses one of the wavelike properties of light. As light waves travel they vibrate, and the direction in which a wave vibrates is called its polarisation. Ordinarily light waves from the sun or a light bulb are polarised in random directions. It is possible to filter the light using special plastic, so that the light only contains waves polarised in one direction. If the two eye images are projected together onto a screen, but each is polarised in a different direction, the combined image can be split again using a pair of viewing glasses with identical filters, one over each eye.
The Polarising filters system has many advantages over the red-cyan system, the main one being that the depth effect is dramatically better. The results can be extraordinarily realistic. Many IMAX theatres show 3D films using this system - go and see one if you need convincing. The problem with colour pictures that plagues red-cyan does not apply to this system, as colour is not used for image separation, so colour pictures are just as good as black and white.
The main drawback is that the system can only be used with projectors. It cannot be used in print, or on TV or computer monitors, which limits it's usefulness a great deal.
You need to get two pictures of the same scene, each offset by a small amount, and ideally rotated about the centre of interest. Softimage has a lens shader which does exactly this called nStereo.
The shader looks after moving the camera to render both of the images. The Eye Separation value is the total distance between the two camera positions, the shader moving the camera half this distance each time from the starting position for each picture. This distance is important because the brain uses it as a yardstick to measure objects in the scene. Set the eye separation to the equivalent of 4 inches or so in your scene and objects will subjectively feel the right size.
As I mentioned before, you get the best results by rotating the camera around a focal position rather than just moving the camera to the left and right. The obvious centre point for the rotation is the camera interest null, but Mental Ray cameras don't have interests, just directions. The Focus Distance parameter is supposed to get around this by providing a point ahead of the camera around which to rotate. Unfortunately, I haven't been able to make it do anything useful, so I generally set to 0 which switches the rotation off :(
The shader can generate either left eye, right eye, or both pictures using the Eye Checkboxes. If both boxes are checked, the shader puts both images side by side in the output, so the image width should be doubled in the Matter module>Render Setup dialog.
For our purposes, produce each picture separately by checking one checkbox, rendering the image, the reopening the dialog and checking the other box. Remember to rename the output file before rendering again.
If you're not using Mental Ray you can rotate the camera a small amount by hand or move it slightly from side to side, If you want to produce an animated sequence you will probably have to write a script or use a pair of cameras joined in a hierarchy in order to make the rotation effect happen reliably.
Then:
If you want to try using colour, you will need to render a central third image - this time with neither eye checkbox checked in the nStereo shader dialog. When compositing the three images I have had good results pasting both eye images over the central one, using a Normal layer mode, and setting the transparency of both layers to 25%. You may need to brighten the composite up after flattening the layers by using Image>Adjust>Levels and moving the right hand arrow to the left.
Animation sequences work equally well using equivalent tools in After Effects or whatever compositor you use.
(C) Copyright Nanomation Limited 2000
This tutorial is free; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version. This tutorial is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You can see the GNU General Public License here; if not, write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
SOFTIMAGE® is a registered trademark of Softimage Inc., a wholly owned subsidiary of Avid Technology, Inc., in the United States, Canada, and/or other countries. Microsoft® and Windows NT® are registered trademarks of Microsoft Corporation in the United States and/or other countries. All other trademarks belong to their respective owners and are hereby acknowledged.
What this means is, if as a result of following these instructions your computer explodes, all your clients desert you for the competition, your business collapses, and your husband leaves you taking the kids, you can't sue me.
