Open mostaphaRoudsari opened 7 years ago
@mostaphaRoudsari A quick note. We need to separate out viewmtx and dc for imaging separately (as and when we have it). It would be a good idea to have a naming convention in place for that. Let's get rid of aa completely and let rcontrib handle it. For direct calcs, it would suffice to keep ab to zero and then get rid of all the ambient parameters. If we aren't already doing this, we need a check to ensure that -lw is always lower than 1/ad value. If not a warning or exception should be raised.
For analemma calcs, in case someone wants to save time on a calc with 10,000s of points, we can have an option to reduce -dc to a value less than 1.
@mostaphaRoudsari , These need to be scaled to
{ 'ambientDivisions':1000, 'ambientBounces': 3,
'limitWeight': 1/1000*0.01},
{ 'ambientDivisions': 5000 , 'ambientBounces': 5,
'limitWeight': 1/5000*0.01},
{ 'ambientDivisions': 20000, 'ambientBounces': 7,
'limitWeight': 1/20000*0.01}
Based on the validated models that I received from LBNL and Bartenbach, I don't think the idea of having high-res, low-res etc. makes sense in the context of VMTX. It might be more appropriate to name these as simple geometry, detailed geometry, highly detailed geoemtry or something simple. The idea being, that for more variability in the view matrix a higher number of rays need to be sampled.
@mostaphaRoudsari There might be some unintentional overlap between the default values for DC Method and the multiphase methods.
I added a file which will contain all the default values https://github.com/ladybug-tools/honeybee/blob/master/honeybee/radiance/recipe/parameters.py and used the values that we currently use in recipes to get it started.
We need to review and finalize these numbers.