POV-Ray : Newsgroups : povray.general : real world size : Re: real world size Server Time
7 Aug 2024 03:20:15 EDT (-0400)
  Re: real world size  
From: Jan Walzer
Date: 2 Dec 2001 15:31:27
Message: <3c0a8f9f$1@news.povray.org>
As Povray uses floatingpoint algorithms, it shouldn't your problem, to choose
the right scale ...

The problems will arise, if you have different sized objects, that hafe sizes
in different orders ...

That means:
It is perfectly OK, to compute everything with 10^5 units...
Say you have one planet, size 5000, distance 10000000, and another planet,
size 5500, distance= planet1+500 ...

should make no big problems ...

/BUT:/
If you choose to create structures on planet2, that have the size of 0.1
units, you will get accuracyproblems as in floatingpoint aritmethics they
will probably cancel out with the distance of which is a number, 10^8 times
higher. This alone maybe won't make the problem, but often there are
calculations, that involve solving of high-order equations(x^2 , x^4 or even
x^6 ...). Then the power wil raise to the double (10^16), or even more, and
then even a double-float-number is no longer able to hold these values ...

... of course, it can stre numbers like, maybe, 10^54 or 10^55... but in
these regions there is no longer any difference between 10^54+1 or 10^54+2
... they will be the same...


So your problem shouldn't be how many POV-Units is one kilometer, but
instead: What is the largest number you will use, and what is the highest
accuracy, you need...

hence, it should be perfectly possible, to scale a complete scene by a
factor, maybe 10^20, to make it bigger, or by 10^-20 to make it smaller ...

As long as the complete scene gets scaled (including cameras ligths and
everything) there shouln't any big difference ...


I hope this hole story makes sense to you...
Maybe someone else can explain better ...


--
Jan Walzer <jan### [at] lzernet>


Post a reply to this message

Copyright 2003-2023 Persistence of Vision Raytracer Pty. Ltd.