The total solar eclipse on August 11:th 1999, an animation from photos taken south of lake Balaton in Hungary.

The four seasons of Djurgården, Stockholm

Hiking on Corsica 20070527 20070602.
Vacation in Nice Nov 2007.
Vacation on Hvar 2008-0701-0712
Brazil 2010-2011

Storm at dusk in Linköping 2008-07-27.


A Panomrama of Nice from the castle (2007-11-27).

When I worked with remote sensing, for fun I generated some vrml-datasets from digital terrain models using the C++ library above (dem2vrml, ElevationGrid nodes), something that software dinosaurs like Arc/Info can also do. Here is the Kebnekajse region draped with a Spot XS image, with simulated natural colours (vrml 2.0). Data courtesy of the Swedish national land survey and CNES/SpotImage. The easiest way to view the vrml-landscapes is to get a plug-in software associated with your favourite browser. Nowadays I use Cortona from parallelgraphics.com, since it handles both vrml 1.0 and 2.0 (you have to get a converter first). You can also use freewrl (see the links), which works nicely for linux/KDE (Konqueror, tested for older RedHat distributions and SuSE 9.2).

In 2004 NASA released an amazing software called Worldwind which allows you to zoom in on any part of the world seamlessly with satellite imagery (landsat) and experience the terrain like in a video game. More importantly, it is linked to meteorological and satellite data which allows for rapid visualization of events like storms, forest fires and earthquakes and more quiescent data like land use. It is now being used in disaster monitoring and is likely to find uses in archeology, geography, geology and many other areas. Google Earth is a similar software with 3D capability and addition it has high-resolution images over populated areas and chances are you can find your house! More importantly, the days are probably over when satellite images were prohibitly expensive and we will most likely see much more interesting applications in environmental inventory studies, monitoring, tourism to mention a few. With the addition of more densely placed and better placed spectral bands and atmospheric correction models, it will also be possible soon to make real physical measurements with optical sensors using spectrometry. Radar sensors have since long been used for altitude measurements, extraction of terrain models, change detection, earthquake displacement changes and wave height measurements.
Using modern high-precision navigation systems, remote sensing images from aircraft can be rectified without the use of traditional triangulation methods. With a high camera frame rate is possible to to achieve automatic and simultaneous texture and 3D mapping. The technique is still in its infancy but its usefulness makes it likely that it will deployed in many mapping application. One exponent of this technology is C3technologies, a spin-off company from Saab Saab Bofors Dynamics and used in the 3D-map application on hitta.se. A parallell development is automatic sparse 3D reconstruction from partly overlapping photos, using robust scale and rotation (partly) invariant feature detectors (such as SIRF and SIFT) for image corrspondances, and bundle adjustment for retrieval of 3D-information. See photosynth.net for a nice implementation.