I hope this is an appropriate place to mail this. Please flame away if not.
I help maintain an image processing library called VIPS. We've just released a new stable version and one of the new features is a Python binding. http://www.vips.ecs.soton.ac.uk The binding is a very simple one generated by SWIG from our C++ API, so it's a bit rough, not well tested, and not very Python-esque. And the new stable version of the library hasn't filtered into many linux distros yet (Debian and ArchLinux are the only two so far I think), so if you'd like to try it out, you'll probably have to compile from source. But it does have some interesting features (IMO): 1) It's demand-driven, so it needs very little memory to process even very large images. I made a tiny benchmark and it needed only 20 MB of RAM to load, crop, shrink sharpen and save a 10,000 by 10,000 pixel RGB image. 2) VIPS speeds up on multi-CPU machines. Wall clock time for the benchmark I made drops from about 12s to about 7s when I ask it to use the second CPU on my PC. If you use the VIPS file format (rather than the single-threaded TIFF library) scaling is more nearly linear: we've seen a 27x speedup using 32 CPUs on a machine at Princeton. Benchmarks and some notes here: http://www.vips.ecs.soton.ac.uk/index.php?title=Python John _______________________________________________ Image-SIG maillist - Image-SIG@python.org http://mail.python.org/mailman/listinfo/image-sig
