Text archives Help

---

But I noticed in the checkin before that it was changed from 1.e-8 to 1.e-4:

r362 | abe | 2005-06-02 18:22:10 -0600 (Thu, 02 Jun 2005) | 1 line

Changes to kdtree code for shadows, also note sphere code wasn't self 
shadowing

===================================================================
--- Parameters.h        (revision 331)
+++ Parameters.h        (revision 362)
@@ -3,6 +3,6 @@
 #define Manta_Interface_Parameters_h

 #define MAXCACHELINESIZE 128
-#define T_EPSILON 1.e-8
+#define T_EPSILON 1.e-4

 #endif

That seems a bit more drastic.

James

Abe Stephens wrote:

Yes I did change that and nobody noticed until now!!!

I don't remember why I made the change exactly (it was several months ago--during intensive Manta development which was happening at SGI ) -- might have had to do with precision in the Double Eagle model which has a lot of long range views--just a guess.

This was about the time of the first Newport News demo.

May have had something to do with shadows since this was the introduction of the headlight.

(btw e-4 to e-3 is a "little" change ;-) )

Abe

James Bigler wrote:

You changed the T_EPSILON value from 1.e-4 to 1.e-3 without even mentioning it in the commit message:

"Big change: Added computeLight method to Light. Added HeadLight class. Parallelized kdtree load, is now around 3 minutes, down from 10. Only triangle parse is parallel, io takes a while still."

What was the reason?

James

---

• [MANTA] T_EPSILON value, James Bigler, 10/20/2005
  • Re: [MANTA] T_EPSILON value, James Bigler, 10/20/2005
  • Re: [MANTA] T_EPSILON value, Abe Stephens, 10/20/2005
    • Re: [MANTA] T_EPSILON value, James Bigler, 10/20/2005

---

Archive powered by MHonArc 2.6.16.