Back to Page 1:
Transferring Images into the QTVR application.
Briefly, images can be scanned from prints, imported from PhotoDisc, downloaded from digital camera, transferred by Firewire from DV Stills, or imported from other applications, including the web.
the Software
The volatility of the software industry is legendary. The brand of software I have used in the past, Panimation, is no longer available, though I will be showing some of its User Interface windows in this article for illustrative purposes. I too am looking at QTVR authoring applications for a replacement. You might want to check out the following URL's for software:
www.vrtoolbox.com/ToolVRWX.html
www.apple.com/quicktime/resources/qt4/us/help/pl/pgs/plVRMov.htm
Apple and VRToolbox make programs for Mac owners, VRToolbox also makes them for Windows users. The QTVR movies will play back fine with a standard Windows install. Often in a corporate environment, IS staff will uninstall the Quicktime components in order to keep systems lean and functional, and more easily managed, but with more than 20 million downloads of the QT-based Starwars Phantom Menace, most personal computers have the QT player. And it's been a built-in helper in Internet Explorer and Netscape since version 3.0 of each.
the Workflow
The first step is to setup the project parameters-how big the final viewing window will be, how much overlapping of images will be required, how many images will be used, and other setup information. This particular program uses a tabbed interface to make it easy to move through the process.
We will be looking at a view from a lookout tower in a public park along the Mystic river in Medford, Massachusetts, with the Boston skyline visible. The images were captured with a Canon Optura DVCam, at the native 720 x 480 resolution, mounted vertically. They were captured with a Radius (now Digital Origin) DV card via firewire. This same card can be used for still capture, or for editing video using DVedit software.
The next step is to import the image files. Before I do this, I often do touch-ups in PhotoShop, my image edit tool of preference. Many times, I will do another round of touchups after I have made a first pass at "stitching" the images together. This gives me a better idea of how well the images mesh. The stitching algorithms are quite sophisticated, but not infallible, and sometimes need help, especially where objects or people fall into the overlap area. In the Copley Square pano you saw, several people have been painted out of the original 18 frames.
The next step is correlation, or stitching the images together,
intelligently combining pixels at the overlap.
When the correlation is done, we can see the uncorrected wide angle perspective shifts that the player program processes to give the final illusion of depth and dimension.
This first stitch or correlation can be edited, by defining the accuracy of the overlap where joints or items like the hand rails might not butt together properly. Sometimes it is necessary to move an image left or right, up or down in order for the correlation to work. This is a process of trial and error, but is usually quite easy and fast. With experience in shooting and pre-visualizing, it becomes less necessary.
The next step is to render the images into 1 large single image. It is still possible to edit the image at this point. The built in cloning tool might be used to reduce or eliminate the grafitti for example.
At this point, we are ready to make output selection choices. Be sure you use formats that can be played back on the target machines-JPEG photo compression is the safest and most universal. You will need to experiment with details like resolution, color depth, and sharpening or blurring. Each has specific uses.
Now we've reached the point of composing the final ouput. This simply means taking the decisions we just made about size, viewing window, starting point, resolution, color depth, and others, and turning that into a final movie composition. If I was making a QTVR for a Cd and for the web, I would do this twice, once in high and once in a lower resolution.
Don't get confused here, the program now defines the composition by how many tiles are in a complete circular panorama. This has nothing to do with how many images were used to create the panorama. If you were showing a piece of jewelry, it might be desireable not only to rotate around the object, but to be able to view it from various heights. This would mean that you would shoot multiple rows of images, rather than the one used in a panorama. This is where the automated systems prove valuable, for speed and accuracy.
That's it. Once you hit the compose button, you get a finished file like this:
Boston
park - 179kIf you would like to see some work I have done on the web, please visit these sites for the Seaport Hotel Boston, and the World Trade Center Boston. Directions for viewing the QTVR's are on the home page of each.
Visit other links from the Apple, Peace River studios, VR Toolbox, and Kaidan sites. You will be amazed at how sophisticated the work can be. Jaguar and BMW let you circle the car, change paint color, and put the convertible tops up and down!
Check out this website of an organization for professionals using QTVR.
I encourage you to give this a try, thousands are making these panoramas and objects daily, some for profit, and some for pleasure. They can be both. If you have questions that aren't satisfied by any of these resources, email me.
Good Luck and happy stitching!
Paul IzbickiPaul Izbicki lives in Boston and likes to introduce the media community to the thrills of new digital technology. He's also insightful in the presence of guys in polar bear outfits, girls who have visions in the desert, and a nice Harley.