Floating Point Accuracy
| Date : Fri, 19 Jan 2007 01:44:44 -0600 |
| To : XSI(at)Softimage.COM |
| From : "Kent Bowling" <kentamanos(at)gmail.com> |
| Subject : Floating Point Accuracy |
I have a few questions relating to floating point accuracy within XSI 6.0. I assume it's using "doubles" within XSI (since the SDK uses those types of values), but I noticed a few things.
The Transform (SRT) panel only seems to show 4 digits after the decimal point, despite the fact what is behind the scenes might actually be accurate than that. I've also noticed that when changing those values in that bar, it only accepts 4 digits after the decimal point. Is there any preference settings to change the shown/used accuracy for those text fields? I realize it seems trivial, just wondering.
Another area where stored accuracy comes into play is the COLLADA export (which is now part of Crosswalk, as opposed to dotXSI). I've noticed for instance if I set the position of a point's X value to for instance
4.000001, that exports fine. If I add one more digit of accuracy (4.0000001), the COLLADA file exports "4.000000". I looked through the SDK and didn't find it, but Is there any way to get the source to the COLLADA exporter so that I might make it "more accurate" (and potentially do some other things with it).
I realize it all seems pedantic (and if anyone needs THAT much accuracy they're probably doing something horribly wrong with scale etc.), but I'm just trying to figure out any potential gotcha's I might encounter while using XSI in my tool chain.
-Kent
- Follow-Ups:
- Re: Floating Point Accuracy
- From: "Bernard Lebel" <3dbernard(at)gmail.com>
- Re: Floating Point Accuracy
| DATE: | << | >> | THREAD: | << | >> | INDEX: | Main | Thread |
|---|
- Previous by Date: Re: paste UVs - vertex order shuffled on file I/O?
- Next by Date: Re: paste UVs - vertex order shuffled on file I/O?
- Previous by Thread: Re: paste UVs - vertex order shuffled on file I/O?
- Next by Thread: Re: paste UVs - vertex order shuffled on file I/O?
- Index(es):
| Search the XSI List archives here or use the advanced search form to search across mailing lists. Searching help is available. |