View Single Post
Old 19-10-2010, 03:14 AM   #1
Registered User
Crispy4004's Avatar
Join Date: Nov 2006
Location: SF Bay Area, California
Posts: 1,121
Thanks: 4
Thanked 18 Times in 15 Posts
Awards Showcase
Threedy Top Poster - Silver: Contributing to the forums with 1,000+ posts - Issue reason: Target reached! The Golden Grenade IV: For participating in the Dominance War IV and successfully finishing, well done soldier! - Issue reason: For participating in the Dominance War IV and successfully finishing, well done soldier! 
Total Awards: 2
32 vs 16 Bit Displacement Maps

For as long as I can remember working with Zbrush I've been choosing 32 over 16 for displacement maps. At the time they were easier to setup and I just assumed 16-bit would be inaccurate based on some poor results I saw. But after I gave 16 bit a shot with Go to Z yesterday I'm starting to ask myself why have I been using 32 for so long.

Much like how 16-bit files are ideal for Compositing, I now think it is also true of displacement maps. After all if you can get over 30 stops of exposure with 16-bit EXRs there should be plenty of value information for displacement. Unless your base mesh is wildly off compared to the highest subd, I just don't think all those extra bits are necessary.

One thing I've noticed with 32-bit and Mudbox, there is no need to offset the values in Maya, unlike 16-bit with Zbrush that scales the values to maximize the stored definition. This tells me for the sake of making it easier to plug into Maya, a lot of that 32-bit displacement map generated from Mudbox is simply wasted space. It's dead weight taking up precious RAM I need for rendering.

Maybe when I have a future super computer with enough RAM and CPU power to not worry about any of this I'll throw caution to the wind and adopt 32-bit vector displacements, but until then I'll stick to 16. So am I completely off the mark with all this or are these good conclusions?

Last edited by Crispy4004; 19-10-2010 at 03:17 AM..
Crispy4004 is offline   Reply With Quote