The Robot
I have been interested in panoramic photography for quite a while. As with most aspects of photography my interest in panos began with collecting them. I have hundreds. Some of the most spectacular photos of the St. Francis Dam Disaster are panos taken by Bernie Isensee in 1928. A man named Dingman took some in the Ventura County area. My collection also includes smaller ones taken with a Kodak pano camera in the early part of the last century. Several of them were exhibited at the San Francisco Museum of Modern Art for their snapshot show about 10 years ago.
I took the pano on the front page of my blog and that image will also appear on the front page of my newly designed web site. That pano was done with my Canon 5D and the 70 to 180 zoon on a tripod and stitched in Photoshop CS3. It does a good job of stitching. The Strawberry Field was made up of only 9 shots. It prints about 5 feet long and looks impressive.
I think it was in the New York Times, must have been. It’s my favorite newspaper and I read the hard copy every day. There was a mention of a new device called the Gigapan. That was about 6 months ago. I visited the Gigapan.org site and saw amazing panoramas. I could zoom into them just like I was doing with Google Earth. This was way better than what I was doing with only 9 images stitched.
I found my way to the link about the Gigapan robot. This device is what the conventional camera is mounted on. Then the robot takes a hundred or so photos and they are all stitched into one. The robot was not for sale. Only a few scientists and insiders were using it. I put my name on a waiting list. Eventually I was contacted and offered one. I became a Beta tester and received a $100 discount. The Robot is $279.00. I thought that was an amazing bargain.
If you want to look at a couple of my panos taken with the Gigapan go to www.gigapan.org and search Santa Paula. You can search my user name “Sespe” but you will also bring up photos of the actual place Sespe. My first GP Pano, that’s short for Gigapan Panoramic, was to contain 100 images. When I got home there were only 60 usable files on the card to download. I learned that I had set the interval between exposures at 5 seconds. The robot finger pushed the shutter every 5 seconds. That did not leave enough time for the camera to write to the card. For my next pano I set the duration between shots at 10 seconds. That did the trick.
It took me almost one month of experimentation and failure to get all the elements working. Now I think I’m having pretty good success and just need to work on the details and learn to see panos better. In the interest of science I’ll discuss my equipment and system There is a forum for all these discussions on the gigapan site but I find it difficult to navigate. Too many meaningless comments clog the system.
The Robot from the front
I’m using a Canon G9 as the camera to mount on the Gigapan robot. I have a Lensmate hood that I just put on to help shield the lens from stray light. I did have to remove one screw from the robot to allow the lens hood to fit on the camera holder. I had attempted to mount a quick release plate on the camera and then mount the quick release unit onto the robot but that did not work. It raised the camera up too much and was out of the range of the robot. I did make a release plate for the bottom of the robot to mount to my tripod.
My initial problem was learning to use all the manual controls on the G9. I read the manual and reviewed and learned to set the white balance to something other than AWB. Since the camera is taking so many exposures over such a broad range you don’t want the white balance shifting from exposure to exposure. For out doors it is set to sun or something appropriate. Then the manual focus is locked in. I then manual focus depending on the scene. I assume most landscapes will be at infinity. Next I set the exposure to Manual and select the combination that works for my time of day. I take a test exposure at that setting and check the histogram.
Part of the programing of the robot is to set the zoom all the way out and use the robot to find out the field of view. I pushed the up botton and the camera moved up and then down to show something like 12 degrees of coverage. The robot uses this to figure out how far to move up and down between exposures.
Rather than read all about if from me you need to go to You Tube and there are 4 excellent tutorials that explain it all.
What is not explained in the tutorials is how to process the images. With the Canon G9 I’m shooting RAW. The GP stitcher will not process RAW so I convert them. I put them all into one folder. Then I open them all in the Adobe RAW converter. I select one with a lot of detail and foreground (not a sky area) and adjust the exposure for maximum detail in shadow and highlights, just like for a regular photo. I then adjust saturation and what ever else I want. With that single image adjusted I then sync those settings with all the other images in the pano folder. Then I save them as a tiff and move on to stitching.
The Gigapan program that comes with the robot does the stitching. Just tell it how many rows of images and they all line up in a mosaic grid. I need to remember to write that down in the field and take better notes. It sometimes takes hours for all the stitching to get done. I notice that many people on the Gigapan web site just upload what comes out of the GP stitcher. That is lazy in my opinion. There are usually very ragged edges. What’s up with that?
Coming out of the stitcher I save the single image made up of all the individual shots taken by the robot. I save it as a tiff. I name it with the date and place. Pretty soon I have a folder with all the RAW images out of the camera, all the tiff conversions and the output from the GP stitcher. In the case of the pano I did of the Ventura Westisde the file size was about 2.6 Gigabytes. The minimum file size that can be uploaded to the GP site is 50 megabytes. The maximum file size that I can get Photoshop to save in tiff format is 2 GB. I had to downsize the pano to under 2 GB before Photoshop would save it.
I then treated it like a regular photo I would want to print. On the screen the pano was filling up my screen at only 2% of actual size. To sharpen it i zoomed and zoomed up to 100%. I found that it needed a little more than usual sharpening at that size. I could also crop it. Cropping was done back down at a small size. Adjust curves and Levels and anything else you want at this time. The file is ready to upload.
The upload for my 2MB file to the GP site took all day. By that I mean about 6 hours. The stitching took at least a couple of hours. This is the part where I need to get more experience. What am I going to do when I take a pano with more than the 144 that made up the Ventura Westside pano?
Once uploaded with keywords and title I could hook it up to Google Earth. That’s fun. Now I have two panos that Earth users can discover and look at.
I have converted one old camera bag for my system. It holds the Robot, the camera, batteries, charger, instruction manuals, flash cards and a Wolverine storage drive. The Canon G9 will take 4GB flash cards. It takes about one of those to hold one pano. If I had more time I would be out there filling up all the flash cards I have. My plan is to download them to the Wolverine after each pano is taken. All these images are going to fill up my hard drive too. My plan is to dedicate one DVD for each pano and burn all the raw, tiff, final upload and notes and data to individual DVDs.
The part I like about this process is that I can make huge prints. With a 2GB file I was able to easily print something 13 inches by 5 feet on canvas and pop it in a frame. Everybody loves it. My 24 inch printer was not working at the time but my next print will be something 24 inches wide by about 8 feet long. Printing on canvas and using stretcher bars means no glass or Plexi to deal with. I’ll blog on this part of the experience later.
The Camera/Robot/Wolverine Bag
