When people see my VR panoramas for the first time their comment is usually something like "Wow, incredible, awesome!" followed immediately by "How do you make them?".
The panoramas are made by piecing together a number of photographs, taken rotating around in a circle. The resulting image can be a cylinder (its left and right edges match up) or a sphere (360° around by 180° up and down).
QuickTime software from Apple allows you to view the image from "inside", making it seem as if you were there. That's why we call it virtual reality (VR).
Many people assume the panoramas are some sort of video, made by panning a video camera around in a circle. Actually it is much more sophisticated than that. What you are seeing is a single image, a cylinder or a cube, viewed from the inside and geometrically corrected in real time as you look around.
The first generation of VR panoramas were cylindrical - they had a limited view up and down. I still make cylindrical panoramas some of the time, and they are in many ways easier to shoot than spehricals, and hence a better way to get started.
Five Steps to Make a Cylindrical VR Panorama
Step One: Take the Pictures
Using a wide angle lens, take a series of from 12 to 16 pictures from a single point, rotating around in a circle. There should be 1/3 to 1/2 overlap between the frames, and the camera must be held level.
For best results the pictures should be taken very precisely, so a tripod and special camera mount are usually used.
Step Two: Get the Pictures on to a Computer
In the early days of this technology we worked with film and the negatives had to be scanned to provide digital images. Now a digital camera is almost always used and the images are merely uploaded to a computer.
Step Three: Stitch the Images Together
Special software (such as PTGui) is used to to process the individual frames into a single image.
It first warps each image, to allow the geometry of overlapping images to match. Then it correlates features in the overlap area of each pair of images to get a good match, and blends the overlapping area to produce a smooth transition. It continues with the following frames to create a single stitched image.
The result is a seamless panoramic image, where the left and right ends match up, essentially a cylinder. Perspective in the panorama is somehwhat distorted, similar to the effect of a fisheye camera lens.
Step Four: Create the QuickTime Movie
The panoramic image is typically very large, so it is reduced in size. Retouching and resizing are done in Adobe PhotoShop. Then another program (such as Cubic Converter) is used to compress the image and package it up in a special VR movie format. In the process you can set the beginning view's pan, tilt and zoom.
I routinely make two sizes: standard and fullscreen.
Step Five: View the Image
For the movie to play as a VR panorama it is necessary to use QuickTime software from Apple, either QuickTime Player or a web browser plug-in. This corrects to a more natural looking perspective, and allows panning up, down, and all the way around, as well as zooming in and out.
Spherical (also called Cubic) VR Panoramas
In the second generation of QTVR Apple added the ability to make panoramas that allow you to look not only all the way around, but straight up and down also.
Spherical panoramas are made with a similar technique to cylinders. You can either shoot more images (tiers pointing up and down as well as level) or use a fisheye lens that captures more per shot. The result is an equirectangular image - twice as wide as it is high.