On this page I am talking about some of the ideas that I had during the design process. The same story has been told a little differently in the March issue of the Planetarian, the journal of the International Planetarium Society. You can download the article here.
This dome was to have as many of the following features as possible:
- large (Smaller domes easily impair your perception of geometry.)
- light weight (Save money during transport!)
- easy setup (Save money for the construction crew!)
- inexpensive manufacture (Save money - period.)
- no airlock (Experience shows that visitors tend to hate them…)
- semi permanent (You don’t want it to collapse over your projector in the evening!)
- more than 180° field of view (Astronomers look up. Geologists look down.)
- tiltable ( … in case you need even more than 20° below the horizon…)
- a superb canvas (Mobile domes tend to show wrinkles along the seams.)
- easy canvas change (Reflectivity is critical in fulldome video - just change the canvas color!)
- rear projection canvas (Ok - this is where we have to start building two domes.)
- more than 1500px resolution (… critical for showing vector graphics and point clouds.)
- only two projectors (Projector alignment is a killer task!)
- real time graphics and fulldome video playback (Just what I was used to…)
Altogether, it was a dome meant for science and teaching, not for pleasure (other than mine). For the most part, the above criteria could be satisfied, although it is obvious that this is a prototype. There is still a lot to work on to give it a factory look.
This is a picture of the dome in its current state of completion, set up in the Lithothek at IFM-GEOMAR.
Influences
For some years I thought about hard shell domes, but they always come down to a large size of parts during transport. Therefore I soon settled for a design involving air pressure, and I came across the work of Prof. Jeffrey Shaw, now at the University of New South Wales in Sydney. His cupola project left a great impression on me.
Parallel to this discovery I learned about the German company Zendome and their (patented) geodesic dome tents. The combination of both ideas would bring together most of the features I wanted my dome to have, while I wanted to be careful not to copy anobody’s work. If I did, I apologize.
Dome Structure
I am really bad at math, but I found a tutorial on how to build geodesic domes for the Burning Man Fesitival which was detailed to the point where I bought a large quantity of PVC tubing and started following the instructions in my parents’ garage. I built a 4th frequency dome with simple but sturdy joints using nothing but screws. In addition, I designed a segment covering the span between 180° and 220° which accounts for the fact that the outer and the inner canvas have to meet at the spring line.
Four people can lift the stucture and carry it away.
I promise - that is all. It fits into a convertible.
My father and I after the construction. Two hours later the grass was cleared.
The evacuation nozzle near the spring line.
Canvas
The gores of a dome (or a parachute) are not great circles but follow a cosine function. In order to learn and make sure I knew the math, I had made a small (2.5m diameter) dome from cheap cloth which I sewed together during easter break 2007 which was never set up but is fully operational.
This is a screenshot of my excel sheet for the canvas hull. No guarantees!
The cheapest way to get a light and air tight membrane to project on seemed to be pvc foil, of which I acquired a large quantity. Ebay only provided yellow foil at that time, but that was not a problem since my intention was to coat the foil with flock. Earlier tests had shown that the diffuse surface promised good results as a curved projection surface.
My experiments with yellow foil were… instructive. Want the gores?
Unfortunately there is no supply of flocked pvc foil in Europe. Also, there was no way I could weld, clue or clamp together the foil gores using garage technology. I should not make this public but I am showing you these images to illustrate that the optimization of such a large undertaking under a nonexistant budget is bound to be seconded by failure. Outsiders may call it tinkering if they wish. At this point I decided for a quality dome over a cheap dome and bought high quality cloth for the outer and the inner dome canvas which was professionally crafted into the hulls by a local company.
The first negative pressure test succeeded because of duct tape. Now we have big Velcro stripes.
Projection
Since 2005 large digital planetariums employ a two projector projection technology based on the 4k Sony SXRD projectors and special optics. This technology is intriguing since there is only one large seam running across the zenith - the rest of the image is geometrically flawless. Through a fortunate coincidence I learned about the genereal geometrical shape that needs to be rendered for one of these channels (The geometry is the same for both channels. They just look in opposite directions.).
The projectors can only be tilted by 15°. The rest is done by a strong lens shift.
Since part of my work at allsky.de is the construction of optical relays for fulldome cinematography, I knew about an optical adaptor that could modify a high end 2k video projector to be usable for the dome. During the course of my project, two Sony VPL-VW50 projectors could be acquired, and an adaption mechanism was designed and built at our intitutional machine shop.
The modified projector.
Projected image without the fisheye.
The same image with the fisheye optics.
Software
I was fortunate to gain the support of two of the leading vendors of dome visualization software: Evans & Sutherland Computer Corporation and SCISS AB, generously granting me access to their Digistar3 and UNIVIEW software platforms. UNIVIEW additionally required the use of the OmniMap geometry correction library developed by The Elumenati. While a blog like this certainly is not the appropriate place to discuss or even compare these products, I can add that the projection solution that had been developed for my dome was a first with both systems.
Setting up the Digistar3 system, I had invaluable help from the techical crew of the Mediendom at the University of Applied Sciences in Kiel, offering one of the finest educations in the international fulldome business. One funny detail is that they are operating a dome just 800m across the water from my institution, so we could wave to each other at lunch. It was only through their support that I could manage to finish the system in time, hours before a big presentation. Many thanks, Markus!!!
Meanwhile we also added an audio system as part of Digistar3 that will soon be upgraded to a 16.2 surround setup. Yes, this is not a typo!
The cluster used for UNIVIEW. There are two graphics units and one control computer.
UNIVIEW in the dome.
The alignment of the projectors was intentionally never perfected for this temporary setup.
My working area in El Salvador. Imagine Google Earth™ over your head and you get the idea.
The Digistar3 cluster. Again 3 computers plus one audio server. Nice to look at and to work with!
An outcrop scene from my field work in El Salvador. This allsky image is a still frame.
Volcanoes on Io. Fulldome video from an E&S show. The gore seams were left open for alignment at this stage.
The Digistar3 console. A monitor is displaying the image of one channel.
Time Frame
The first ideas towards data visualization in the geosciences were born in 2005. My diploma thesis had a jump start when I went to El Salvador in february/march 2007 with little more than a Spanish dicitonary. The work on the dome started in April with the first test dome. The geodesic structure was built during two weeks in August, but it took three more months to figure out the exact shape of the two hulls. The dome was operational running Uniview by the end of November.
By that time I was asked to give a presentation at an upcoming evaluation of our research group by the German Research Foundation, which left me the Christmas holidays to work on it. This meant that I did not just process my own data but also the data sets of other scientists of our group. Read more about this in the dataviz section! Digistar 3 was up and running on January 17th 2008, two hours before the presentation. During the course of the following two weeks, various presentations were given to a number of audiences.
In the meantime, I concluded my active research in immersive geoscience visualization, at least for the scope of this project. I gave a talk on these matters at the Immersive Cinema Workshop in Plymouth, taking place in March. The dome was set up during the meeting of the Arbeitsgemeinschaft Deutscher Planetarien and will remain operational for the following European Digistar Users Group Meeting.