DCGlare

[Ali Fatoorechi]

Hi Nathaniel,

I tried your tutorial2 for dcglare program. I shall say the imageless method is great giving that the image method takes hours to finish even for a view direction.

However I am not sure if the program reads the dc1.mtx as DCDirect. 

dcglare -vf uniformgridviews2.ray  -l 0.4 test dc1.mtx dc8.mtx SanFrancisco_1.smx > occupied.ga

On debugging the code "direct_path" is set to dc8.mtx. 

Looks like line 303 ( direct_path = argv[++a]  ) should use a++ rather than ++a ?

[Ali Fatoorechi]

I by mistake had extra parameter "test" on the posted command. Here is the corrected command args which still seems to parse dc8 as direct matrix.

dcglare -vf uniformgridviews2.ray  -l 0.4 dc1.mtx dc8.mtx SanFrancisco_1.smx > occupied.ga

[Nathaniel Jones]

Hi Ali,

Yes, it should be a++. Interestingly, the difference results in a less than 1% change in the average glare autonomy in the tutorial example, which I suppose goes to show how minor the effect of reflection is on glare in typical spaces.

Nathaniel

[Ali Fatoorechi]

Hi Nathaniel,

Sorry I couldn't articulate it properly. 

You said: "  the difference results in a less than 1% change ....which shows how minor the effect of reflection is on glare " .

So my question is which one do you think is more accurate, to use multiple bounces or not ?

Thank you,

Ali

[Nathaniel Jones]

Hi Ali,

The algorithm that dcglare uses is described in this paper. It requires you to calculate both the direct and total daylight coefficients, where the direct is calculated with ab=1 and the total is usually calculated with ab=8. So you have to use both a one-bounce and a multiple bounce calculation. The tutorial describes how to do this.

Nathaniel