Text archives Help

---

Ok, well I ported over the coherent grid traversal algorithm and it seems to work, but it's not as fast as DynRT -- most likely because there aren't square packets. I'd rather not spend time learning how ImageTraversers work, so I'll put off the CGT porting and continue working in DynRT for now. If square packets ever get implemented, let me know and we can find out how DynRT compares to Manta (currently DynRT is roughly an order of magnitude faster).

Thiago

James Bigler wrote:

No ImageTraverser class yet implemented generates square fragments. Try adding it. ;)

James

Thiago Ize wrote:

I think step 2. isn't creating square packets. In fact, the fragment has "Fragment::LineShape" according to TiledImageTraverser::renderImage(). So does manta not have all the code to do square packets yet?

Thiago

Abe Stephens wrote:

Try this for primary rays and see if it works:

   1. Set the fragment size and ray packet size to be equal integer
      squares. By default they are both 32.
   2. Make sure TiledImageTraverser can assign pixel coordinates to
      square Fragments correctly.
   3. On the command line specify tile size to be whatever
      fragment/ray packet size you chose.
      bin/manta -imagetraverser "tiled( -tilesize SxS )")
   4. Use the default single sample pixel sampler.

Currently there isn't a modular component for secondary ray sampling.


Abe


Thiago Ize wrote:

Anyone know how to enable square packets?
Thanks,
Thiago

---

• [MANTA] square packets, Thiago Ize, 07/24/2006
  • Re: [MANTA] square packets, Abe Stephens, 07/24/2006
    • Re: [MANTA] square packets, Thiago Ize, 07/25/2006
      • Re: [MANTA] square packets, James Bigler, 07/25/2006
        • Re: [MANTA] square packets, Thiago Ize, 07/25/2006

---

Archive powered by MHonArc 2.6.16.