Scientific photography in El Salvador is only safe with armed guards.
The whole aim of my dome effort was the visualization of geoscientific data. This has been done for many years in the field of astronomy, so now at the same time many people are stretching their hands towards other sciences. My thesis is not actually about building a dome. If you are really really into these things, you can read the proposal here.
Calderas
I went to El Salvador to study the older deposits of a caldera volcano called Ilopango. Caldera volcanoes are basically large holes in the country that formed after the magma chamber beneath them had emptied. At some point, the cavity collapses and you have a ditch of some km in diameter, usually with a lake. Caldera volcanoes are important since they take some time between eruptions, but when they do erupt they are usually quite disastrous. My task was to explore the history of eruptions of this volcano and to create a geological map - the usual geology business, so to speak.
This is as close as I ever got to Ilopango. You can see the lake at 2 o’clock.
On top of that, I wanted to visualize all the data I was gathering in a dome. In order to make this particularly interesting, I already applied some interesting techniques in the field:
allsky photography
An allsky is essentially a panorama which is not a stripe but a circle where the horizon runs around the perimeter and the zenith is in the middle. Think of it as a fisheye lens that you point right into the sky. These images are the preferred format of panoramic projection in planetariums and have developed into a standard of the fulldome video industry. In my case I used allskies to map the outcrops that I visited. allskies help you connect the spatial distribution of rock walls much better than a panorama or a single image would do. Since I could always see the sun, I aligned the images knowing my position and the time of the day.
This is an allsky of my outcrop, the same as depicted under “story”. This one is commented with geological structures.
allskies are shot by acquring six fisheye images in the field, at an angle of 60° to each other and a tilt of about 20° up from the horizon. If you took 3 different exposures each, you can even eliminate the effect of shadows which is particularly disturbing for geological interpretation. These images are later “stitched” using special panorama software. Back home, you can load the images into your dome and pan and zoom them like a normal image, only that you are in the center of the scene.
Photogrammetry
Photogrammetry uses stereo information contained in two images to calculate a three dimensional model: Take two photos of the same rock wall but about 3m apart. Then run the images through a computer program and end up with a complete 3D simulation of the rock wall that you can load into your dome!
To do this, I contacted the company 3G Software & Measurement in Graz, Austria. I had found their website before I left but it was only in El Salvador that I had the time to actually talk to them about my project - over my cell phone, El Salvador to Austria, for 20 minutes…
Fortunately, the friendly people granted me access to their software so that I could generate rock walls out of my photographs. Not only rock walls - the entire outcrops were suddenly available again, to an accuracy where I could actually take measurements of the model in parts where I would never have reached the actual walls.
It is a great tool for a geoscientist: You can revisit, you can measure, plan, communicate your findings to other scientists and students. Just like on a map that you draw, all the meta data is visible and browsable, plus the models can be exported. The only problem is there is only so much that you can capture on a single photograph (or a pair, in this case). Therefore I ended up stitching large panoramas along my rock walls, only this time I did not use Photoshop but the powerful 3dsmax9 software. The difference between retouching a 3d image and a 2d image is not so great, after all. It just takes more time.
In the end I had a huge, very precise model of my rock walls that I could load into my dome environment. To be true, I only did so in the form of a movie because I simply did not have the time to persuade all pieces of software to accept the data as a real time model. It is possible, though.
GIS data
Next to allskies and photogrammetry, I had received the complete country of El Salvador on a digital geological map in the ArcView format. This allowed me to directly translate the data into Google’s KML format, the same language that Google Earth is using. While they have nothing to do with Google, the UNIVIEW software too supports this language so I could load El Salvaldor into Uniview and browse the map on the dome.
Geological map of Ilopango with shaded relief. The red line is the Panamerican Highway with my outcrops.
More data
Just when the dome was finally starting to work, my supervisors and I speculated that it might be possible to use this dome as some sort of high level entertainment during a very important evaluation of our institute by the German Research Foundation. That is to say, they were going to decide on all our jobs.
We subsequently loooked at my boss Steffen’s data and chose data sets that looked promising or meaningful for dome projection. In addition, we asked other scientists as well to contribute data to our project.
In the end, we put together a 16 minute show consisting of a number ov very different scenes:
At allsky.de we have spent the last 3 years developing a camera that can actually film in fulldome video. Nothing is rendered, insted the scene is captured on 35mm film analog to an allsky image, but 24 times per second. I took some of this test footage shot on a German research vessel two years earlier, to show the possiblilties of recreating a place in the best possible way on a dome.
Compare the original dome master (left) with an allsky of the projected image (right)!
Next, we used the software Fledermaus to render flights over extremely highly detailed terrain offshore Central America. Here, our group worked out the details of the process in which the oceanic plate dives into the Earth’s mantle beneath Central America, forming a trench close to the shore. The Fledermaus views were rendered as stripes which were subsequently wrapped as panoramas around 160° of the dome circumference.
Fledermaus view of a flight towards Jaco Scarp off Nicaragua.
A colleague of my supervisors had provided us with biological data of some sub marine faunal communities on the slope of the continental plate off Nicaragua which had been visited by a diving robot (ROV). After recreating the regional terrain and the dive paths across it, I lotted the observations in the form of a biological map of many different species. Since there was a lot of layered information within these maps, I arranged them in several layers above each other. Seen from the top, they would add up to the original two dimensional map while it was possible to look at each layer in the context of all others as well.
Point cloud of the fauna around Jaco Scarp, surrounded by video windows.
In addition this dataset contained short video clips of individual sites along the dive path which could be identified using the dive log. These clips were positioned accordingly as a sort of virtual movie screen and could be watched as the camera paused in front of them. While this scene was rendered as a movie for lack of time, it would be perfectly possible to browse it in real time using Digistar3 and its extensive video playback capacities.
The main part of the show consisted of the terrestrial data that my supervisors had gathered, complemented by offshore sediment core records. The resulting database is documented in maps of so called isopachs, i.e. the mapped extend of certain ash layer (tephra) thicknesses on land and in the marine sediments. Normally these maps are just drawn in two dimensions such as the height data on a topographical map. In this case, though, we estimated the rough ash cloud height for the individual tephra thicknesses, ending up with a sumwhat approximated geometry of the cloud as it had been distributed across Central America. Watching the rendered animation on the dome, we could look at this data set in every possible angle.
Isopach data in a dome master and in a zoomed view. Individual eruptions are color coded.
The isopach data was followed by geophysical models of the response function for a seismoligical study offshore Nicaragua. Normally these models are calculated in 3D but rendered as two dimensional sections. Stacking 15 of these semi transparent sections behind each other, we recreated the volumetric impression of the original data set. Interestingly, the two dimensional nature of our original data led to a polarized transparency of the dataset in the direction of the slice planes.
Out of this data, there had been studies in which structures had been identified in the slices which had been assembled across sections to a 3d model that could now be loaded into the dome simulation. Looking along the section planes, we could easily make out the shape of the subducting plane as it was diving down into the mantle beneath Central America.
The show commenced with a film that I had created from my outcrop data. We flew through one of my outcrops, the walls passing to the left and to the right, with the fault planes marked as yellow polygons. Next, I showed some macro photgraphs of my rock samples as well as a view through a microscope.
The show ended with another film clip of the allsky.de camera.
Feedback
While I did not have time to collect a lot of differenciated feedback, it was widely stated that much of the content had been presented too fast. This was obvious and I had expected it since I had only been alotted a time window of 15 minutes for a very diverse range of content. Most people acknowledged this fact.
On the other hand, there were only few scientists who mentioned some sort of scientific value by themselves. Most emphsized the enormous impact for public outreach. In my opinion this is an ironic development since for the last eighty years people working in the planetarium profession have gathered a tremendous expertise regarding public outreach using domed displays. Now that it is time to reach out the scientists themselves, they are heading the other way but we zoom past each other…
I will definitely try to assess this point a bit more precisely in my upcoming thesis paper.
