Might have to see the definition to have a guess. Wanna post/send me a file?
In the definition it's the 'G80%-S' component that i had set to auto shades. Here's the model; the GH definition is attached. note that i'm using Human's geometry pipeline
while i'm sending this - a few other issues have come up:
Indeed, I get the same error. The bug has something to do with the way DIVA recreates windows from the DLT file. I won't bore you with the details, except to say I think I fixed it... try running the latest auto update (revision 184.108.40.206).
The new version should also do a slightly better job at trimming planar Grids -- I noticed some unsightly meshing happening on that middle floor. I'd still recommend you clean up your floor layer, however. You've got the sides and underbelly of the upper-floor brep in there, which is creating unneeded grids and adding to the run time.
To your other questions:
1) Invalid thresholds. You reset the DAv upper threshold on the Grid Viewer to 50 lux, which is less than the lower threshold of 300 lux. Hence the warning.
2) Accelerad. Sadly I'm pretty ignorant on this front.
3) Still no way to adjust north? Nope.
Regarding grids - i actually spent much of yesterday afternoon doing exactly that. but had some thoughts to suggest on this front - obviously creating grids from existing geometry is preferable, particularly for wonky floor plates after starting from an architect's revit model. But i found that i was getting light leak in certain locations due to the edges of the grid containing sensors. Given this, i suggest 2 alternatives:
regarding the other items:
1. thanks! yes i figured this was the case...
2. no worries. im finding that the 64bit is helping a ton anyways
3. really?! this would be a huge improvement - c-planes are super annoying and working not in the XYZ is a total pain in the ass. This is a definite feature of Honeybee that i appreciate...
seriously great work on DIVA4 though. it's INCREDIBLE. I have another feature idea that i'll share in a different thread.
Thanks for the feedback, Dan.
Regarding grids, I'm not sure I follow. There are no sensors at grid edges -- only at the center of each face.
As for North, I always assumed rotating the model (or using CPlanes) was not such a big deal. But I suppose this could be added as a parameter. I guess as a vector for the sim components?
grids: right of course the sensor is in the middle! but what i've found is when i apply the grid to trimmed surfaces, there are zones where i get a little sliver of grid that gets a sensor point. so maybe there's a way to cull sensors from grid cells that are smaller than a certain threshold?
for north - yes exactly. if it's easy on your end, it'd make a big difference.
> there are zones where i get a little sliver of grid that gets a sensor point
Yes, that can happen -- maybe a threshold for culling tiny faces, as you suggest. The nice thing is, the grids already account for the face area associated with each sensor, so these "sliver" sensors don't have much effect on the spatial metrics (sDA / ASE).
> for north - yes exactly. if it's easy on your end
For some point-in-time sims it's as simple as changing the azimuth angle, for others a bit more involved, but probably doable. Let me look into it...
FYI, added North axis rotation to the latest revision -- available as a hidden input on all the simulation components. It's just a number, in clockwise degrees, you want to rotate the North vector.
The North rotation tool is much appreciated Jon! Thanks for advocating Dan.