MELTING METALS IN A DOMESTIC MICROWAVE

August 12th, 2010

By David Reid: Read more

Also see this article by Popular Mechanics based on his work.

Comparison of SVG and Canvas in OpenLayers

July 30th, 2010

Tobias Sauerwein recently emailed me about my mapping framework Cartagen, and linked to his fantastic analysis of SVG/Canvas suitability for map rendering. Thanks!

OpenLayers with Canvas

Oh, of course, it bears mentioning this earlier comparison I dug up some months ago, which focuses on SVG/Canvas different sized rendering areas and different #s of objects.

gdalwarp cannot perform perspectival distortion

July 26th, 2010

I’ve been banging my head against a wall for a few days on this one, first struggling with ImageMagick, then switching to gdal to get full-resolution geoTIFFs from Cartagen Knitter. I had a ‘duh’ moment just now when I realized that gdalwarp can only do polynomial or thin plate spline warps, neither of which are what I want – that is, perspectival warping. I want to map 4 corner ground control points or GCPs to four latitude/longitude positions. Back to ImageMagick…

I know there are 5 GCPs in the image above – it was the same deal with just 4… it can’t do more than a shear unless you either add the minimum 6 GCPs for a single polynomial warp, or go for a thin plate spline (TPS) distort. A good way to think about TPS is as if the image were a sheet of thin metal (the reason it’s called a TPS) and that you’re bending it in the z-dimension, aplanar. This causes funny curved edges and is not what I’m looking for.

OK, one more note for future reference: see this page for a discussion of different warping techniques and also for the minimum number of GCPs required for different-order polynomial warps.

Morozov: examples of ‘bad’ uses of internet and new media

July 24th, 2010