Sequential Capture 3D Camera Rig
Edelkrone doesn't market it's slider as a 3D rig, but it can be! The system shown consists of a SliderPlus Pro Large, video slider with an attached Action (Motion Control) Module and a Target Module. The system was designed as a lightweight dolly and pan device for video production. The Pro Large slider has a camera travel of 700mm or 2.3ft. This turns out to be much more than I really needed. In testing, it seems a distance of 200-250mm is sufficient for 3D capture for lenticular, much more than that and you begin to lose detail in foreground and background objects. The extra length does give me more options for timelapse and stop motion uses.
The release of the Target Module last Jan. was what sold me. With that addition, the system now did almost exactly what my own rig design would have done. After seeing it, I scrapped plans for my own rig and ordered one. I realized it could accomplish the same task for half what an engineer wanted just to build my rig. tThen there was software to be developed. While still not a true arc like the rig I designed, it should be close enough for most work.
A parallel track system, (like the one we initially used at Kodak), requires realignment of the images to a fixed point in the scene. The reason is because the camera is just pointed straight forward. You can only use the portions each image that overlap with all others, as a result you lose frame with and your rectangular image starts to become square. The Target module solves that issue by pointing the camera at a set spot in the scene and keeping it pointed at that spot during the entire sequence. The result, no realignment and no cropping of image width.
Some might tell you that this method causes distortion, while true, the amount is barely noticeable. I conducted a number of tests while with Kodak's Dynamic Imaging Division comparing pivoting camera head use versus "straight-ahead" shooting on a parallel track. Exactly this scenario. There was a slight bit of distortion on the wide angle lenses (improved with longer focal lengths), but usually not noticed single stimulus.
This is one of the first quicktime movies captured on the rig. Total travel was about 8 inches. Target was set to about 30" from camera.
Here are some new gifs from a shoot at Letchworth State Park.
One of the drawbacks of sequential capture. As you can see it was quite windy that day, however the movement isn't nearly as noticeable in the print, it kinda averaged out. The same was true for the second image. The foreground leaves don't appear to be moving in the print, the only things that appear to be moving are the clouds.
I learned something else important in this capture. You must be aware of the prevailing wind direction. The clouds in this scene are moving counter to the rest of the background. As a result, the clouds seem to appear in front of the trestle depthwise. I think I could have solved it simply by doing the capture in the reverse direction. I only did a left to right capture. From now on I think I'll do a capture in both directions just to be safe.
Both series of captures were done at a 200mm disparity (or travel distance), the pivot point (window plane, etc) was set to 12 feet for the first shot, and about 8ft for the falls shot. I captured 24 frames for the top image but only 23 for the second, I duplicated the first frame to give me 24 for interlacing.
Before the trip I did a series of time trials with the rig, measuring the time it took to travel over several set distances (400, 300, & 200mm) at different speed settings. The intervalometer on the Action Module can't do times less than 5 minutes, so I use an external remote/ intravalometer to trigger the camera. Knowing the time it takes allows me to divide it by 18 or 24 to determine the timer setting to capture either 18 or 24 frames during the slider travel. I created a chart which I taped to my Pelican case for the rig. I may now do one more timing test at a 250mm total distance and remove the 400 distance as it is way too much travel for my lenticular purposes.