I’m trying to run rpict on a sequence of octrees that have been generated using sunny skies in 10 min increments over the winter solstice.
At the moment I am generating a new octree for every sky file + rad file combination, which is taking a relatively long time.
Is there anyway to bypass this and run oconv once to generate an octree of the scene.rad files and then when I am ready to run rpict either select or substitute which sky file I would also like to run with this oct file.
There are two ways to speed this up. If your sky always has a sun in it, then you can just use the same file name and overwrite it without updating the octree. Since neither the sun nor the sky are actual geometry that need to go in the octree, you can treat them as you would material files. I.e.:
gensky {first date and time} > gensky_out.rad
oconv {scene files} gensky_out.rad sky.rad> full_scene.oct
rtrace, rpict, etc. using full_scene.oct
gensky {second date and time} > gensky_out.rad
rtrace, rpict, etc. using full_scene.oct
…
The second (and most fool-proof) method is to use the oconv -i option, which is designed to accelerate small scene changes such as this. I.e.:
oconv -f {scene files without sky} > skyless.oct
oconv -i skyless.oct sky1.rad > modelsky1.oct
rtrace, rpict, etc. using modelsky1.oct
oconv -i skyless.oct sky2.rad > modelsky2.oct
rtrace, rpict, etc. using modelsky2.oct
…
The oconv -f option makes the second step quicker by saving the scene data in the octree. Note that you must NOT use the -f option in the first approach. This method will work even if you completely change the sky description, such as using a cloudy sky in one case and sunny sky with sun in another. Although it still requires running oconv each time, the command is very fast using -i in this way.