libpng and new volume rendering methods

Hi everyone, (if you're on yt-1.6, feel free to ignore this message!) I've backported the multivariate volume rendering (for instance: http://yt.enzotools.org/attachment/wiki/Screenshots/hiiregion_rgb.png http://yt.enzotools.org/attachment/wiki/Screenshots/Reion512_600px.png http://yt.enzotools.org/attachment/wiki/Screenshots/SimB-BH.png ) to the yt trunk; this also comes with the ability to project through a simulation along a non-orthogonal vector, so you can create images that, for instance, peer down the L vector of a gas cloud but include contributions from all the intervening gas -- it's much faster and higher quality than stacking cutting planes, which was the old method for doing this. (Example: http://yt.enzotools.org/attachment/wiki/Screenshots/halo_0029.png ) The new backend for volume rendering should be API compatible with the old, but if you update your installation you will have to rebuild the C extensions either by doing: $ yt instinfo -u or by re-installing. The new multivariate volume rendering as well as the off-axis projecting have not yet been documented. This also comes with a direct PNG writer, which is much faster than using matplotlib for outputting raw images; it also comes with a couple colormaps. This requires that libpng be installed. yt will do its best to locate libpng, but if it cannot find it, you will have to supply it in png.cfg, much like for hdf5 -- however, the installation script has been updating to install libpng if requested, so if you run into any problems, rerun a new version of the installation script. (Running it over top of an old installation should be fine, as it will just conduct the necessary upgrades.) For the next release, both the new volume rendering mechanism and the image writer will be documented in detail; the release is probably about a week or two away. Let me know if you're eager to get your hands on the API docs for either of these things and I'll shoot you some sample scripts. If you run into any problems, let us know! Best, Matt
participants (1)
-
Matthew Turk