Rendering in 3D - am i wasting my floating point values?

OK.

I render out in Floating point from maya every day. Generally though my colour values goes from 0 to 1. Apart from specular hotspots and such.

Which is fine, though i do wonder sometimes if I should make my colours range from 0 to 100000000000000000000000000 or something.

Am i wasting the available range in floating point? Will i be more susceptible to rounding errors? Is it the same as only rendering in values 0-35 in 16 bit when you have 0-65535 available.

Although I haven't seen any evidence of this in after effects.. it is one of those things that keeps popping into my head.
Maybe i'm just misunderstanding things though.

Gareth

Floating point is weird. Larger numbers are less precise than small ones, because they have fewer decimal places to work with. So no, you would not gain anything by re-mapping white to some crazy big number (Real Math People please correct me if I am lying).

Experientially I can tell you with confidence that there is, uhm, A TON of fidelity between 0–1 in even a half-float (such as most EXRs) image. A ton. Worry not.

-Stu

_________________
The DV Rebel's Guide

Oh okay. So it doesn't have, say, a set amount of numbers before and after the decimal place.

Actually that makes sense the more i think about it.

So essentially you have 32 or 16 bits to represent the number and large numbers take up more of the allocation so theres less room for decimal places. (which is pretty much what you said)

So really, keeping everything around the zero to one range could be the best way to do it as I get more decimal places that way.

Nice!

Cool yeah i use 16bit half EXRs here just because they take up less space. Though I was thinking of upgrading to 32bit as I recently put together a server with a lot of hard drives in it. And whats the point of having all that empty space on those drives! Smile

Thanks Stu!

I think this forum will be handy.