Perlin noise and fBm help
by Rich Wilson · in Technical Issues · 02/17/2003 (2:29 pm) · 5 replies
Does anyone out there know of any good, plain english tutorials on how to write a perlin noise or fractal brownian motion function. I'm trying to write a multifractal terrain generator, and I can't wrap my head around some of the articles I've seen on the issue. A lot of the writings I've seen on it just give you the code and that's it, which isn't entirely helpful in itself. If anyone here knows of any links, or can sum up the functions in a post, I would be very appreciative.
About the author
#2
For fBM, I'd recommend getting a book about procedural models (if I remember, Ive got a book with a preface by Perlin called something like Texturing and Modelling - A procedural approach).
Try looking for diamond square algorithm if youre interested in doing landscapes.
That book is definitely the bee's knee's if youre going to get into any procedural stuff at any deeper level than just using a function though (it'll explain the principles behind it, and possible uses).
Phil.
02/17/2003 (5:56 pm)
Take a look for Hugo Elias's website (bouncy fun and stretchy), its got a good explaination of Perlin noise.For fBM, I'd recommend getting a book about procedural models (if I remember, Ive got a book with a preface by Perlin called something like Texturing and Modelling - A procedural approach).
Try looking for diamond square algorithm if youre interested in doing landscapes.
That book is definitely the bee's knee's if youre going to get into any procedural stuff at any deeper level than just using a function though (it'll explain the principles behind it, and possible uses).
Phil.
#3
As for the diamond-square algorithm, I'm assuming this is similar to the midpoint displacement, or plasma algorithm. I've actually implemented this in a few different ways, in 2d and 3d using marching cubes, but it just doesn't have the properties I'm looking for for nice looking terrain. What I'd love to be able to spawn is what they use for the multifractal pattern examples in the Procedural book. Defined areas of mountains, hills and valleys, instead of homogenously bumpy terrain.
I'm also going to try a multi-iterative approach on the diamond square, with rougher iterations being scaled down and added to more broad sweeping iterations, and see if that will strike my fancy.
02/17/2003 (6:30 pm)
I checked out the Procedural Texturing and Modelling book, and it's mainly what I'm talking about with the hard to follow comment. I know it's a very mathematical problem, and I'm willing to delve into that as deep as need be, but I feel like I need a more down to earth explanation. This is a case where it'd be nice to have a teacher on such a topic handy. I'm going to give the book a couple more look-throughs though and see if I can gleam something.As for the diamond-square algorithm, I'm assuming this is similar to the midpoint displacement, or plasma algorithm. I've actually implemented this in a few different ways, in 2d and 3d using marching cubes, but it just doesn't have the properties I'm looking for for nice looking terrain. What I'd love to be able to spawn is what they use for the multifractal pattern examples in the Procedural book. Defined areas of mountains, hills and valleys, instead of homogenously bumpy terrain.
I'm also going to try a multi-iterative approach on the diamond square, with rougher iterations being scaled down and added to more broad sweeping iterations, and see if that will strike my fancy.
#4
http://freespace.virgin.net/hugo.elias/models/m_main.htm
Very good site. Marching Cubes is a damn cool concept, I don't see it being all that useable though unfortunatly =\
02/17/2003 (6:36 pm)
Just in case you don't have the link, this is what Phil was talking about:http://freespace.virgin.net/hugo.elias/models/m_main.htm
Very good site. Marching Cubes is a damn cool concept, I don't see it being all that useable though unfortunatly =\
#5
I'm reading through the bouncy fun stretchy site right now and it's a very nice approach to the subject. I can feel the information sliding into my brain a lot more easily. No more square peg-round hole dilemma.
As for the tea, I'm not letting my computer get anywhere near the stuff. :)
02/17/2003 (7:02 pm)
Drifting slightly off topic, marching cubes are nifty for prerendered volumetric data, like the visible female or seeing what plasma in 3d looks like, but it is pretty useless for real-time applications of anything high resolution. I think that's where point billboards or other such technical gizmos could come in handy.I'm reading through the bouncy fun stretchy site right now and it's a very nice approach to the subject. I can feel the information sliding into my brain a lot more easily. No more square peg-round hole dilemma.
As for the tea, I'm not letting my computer get anywhere near the stuff. :)
Torque 3D Owner Pat Wilson
(Ignore if you haven't read Hitchhikers Guide)