Here’s some quick (I just hit render, didn’t even bother to set up a nice lighting arrangement) tests of specular rendering.
- Chad
Here’s some quick (I just hit render, didn’t even bother to set up a nice lighting arrangement) tests of specular rendering.
This is a test of the output of our PRT writer. It outputs color, density, and now it does normal. In this case, calculated in a 5x5x5 kernel.
The speculars help to eliminate the Nurf look we were getting before.
Here’s some more.
That guy looks like something out of Doom. Where did you get that data from?
Guy? That’s a lady.
Ewwwwww. Definitely not one I’d like to meet on the street. Or anywhere for that matter.
some realy great stuff!
Your dissolving hand looked great on the nice High res monitors at Siggraph!
How are u getting the spec to show up so strongly, comp trickery? or is this some fancy method Im not aware of?
cheers
The spinner on the specular level can go really high. I was testing it from 400 up to around 10000, I think. I had to compensate with the lighting levels, but it worked just fine.
It is possible to render out a specular pass though. Just override the particle color and set the specular level to be really high.
Those shots look like cross processed film - Gorgeous stuff. Fair play to you chad, you’ve pushed this thing really far.
With the normals, we’re able to make pretty nice looking “solid” objects. So much so that in those full-body images, the surfaces you see are only 3 points thick. The interior points have all been culled out. When we evaluate the normal with a 5x5x5 kernel, if the result is either all 0’s or all 1’s (meaning it’s either entirely inside or outside the “surface”) we cull them. So these renders, while appearing solid, are actually only a couple million points.