I have experiencing an issue with my HoneyBee daylight simulation that the simulation stuck at CreateOctree task and CreateOctreeWithSuns task. Both task are taking forever to finish. I tried running the oconv command by itself and it finished with no issue with the enevlop.mat and rad files.
Have anyone experienced a similar issue? Is there anyway we can get more information regarding the status of each task? Thanks.
Best
Bing
@Bing ,
It sounds like your model is just huge and so it takes a longer time to run. In the honeybee radiance recipes, we have to create a frozen octree (-f option), which has all of the geometry and modifiers included in it. This is different than what the oconv command does by default, which is just to store references to the geometry and modifier files. This means that the creation of a “non-frozen” octree is much faster but, as soon as it is copied to another folder, it is essentially broken.
We have to create a frozen octree in the honeybee radiance recipes because each step of the recipe process runs in a separate folder that is unaware of the other steps. It’s this separation of tasks that makes the whole magic of recipe parellization possible. So trust me that, while it takes longer for this oconv step to run, the recipe as a whole will be a lot faster and more parallelized because of the independent execution of tasks.
I hope that clarifies what is going on.
Thanks Chris for sharing your thought. Normally I’m dealing with big model but in this case it is not. I was using the oconv with -f option when testing it. The envelop.rad file is about 60mb and the oconv is about 130mb as the outcome, which I do notice normally the octree file is the same size of rad file but it is not in this case.
I found the issue was caused by some geometry and rebuilding the geometry resolved the problem. I’m not sure why the make octree process in honeybee is performance differently than a oconv -f envelop.mat envelop.rad > scene.oct line but maybe I should take a look at what exact the command line in honeybee recipe is.
Best
Bing
Thanks for elaborating and what you describe sounds strange. You can always try running the Recipe from command line via the lbt-recipes CLI. If you use the --debug folder option on that command, you can have the recipe store the files of each recipe task in a separate folder along with a .bat file that has the command being run for that task. That would allow you to understand better what was going wrong with that particular step of the recipe.
I remember @mostapha telling me that Radiance octrees can freak out when you have two .rad polygons that are perfectly coplanar and overlapping with each other. Perhaps that’s what was happening here?
Honeybee uses the exact same command but here is an issue when large amount of warnings are printed by Radiance. I pushed some changes that should resolve this issue. The error was the result of a recent change that we made to accommodate an edge case on Mac.
I suspect your rad file has several slightly non-planar faces which results in Radiance printing out several warning messages when you try to create the octree.
Thank you @mostapha! I can confirm this fix resolve the issue I had. Thanks for the quick fix!
Happy New Year!
Bing