Author Topic: Simplex noise tutorial  (Read 4920 times)

Simplex is a faster and better revision of Perlin.
But if your not conviced look at this.
Linking me to the wikipedia articles is probably the dumbest thing I've ever seen anyone do. In case you've somehow missed my existence on the forum, I only argue about topics I know I can win, meaning I have done my research.

First off, Simplex gives undesirable results in comparison to Perlin. Look at the visual differences in your example. They're far too drastic in change to support a desirable terrain.

Second, they're using Perlin incredibly inefficintly. I'm positive this was intentionally done to exacerbate the speed difference. Just look at their code, it's horrid.

Third, there's more considerations than speed to take into account. Perlin noise is significantly more manipulatible and customizable than Simplex. Simplex noise loses on every frontier besides speed, which you're right, it's faster. That does not make it better.

simplex has a complexity of n^2 for dimension n, and perlin has a complexity 2^n for dimension n. Therefore perlin is faster by ONE.
simplex has no visual artifacts. perlin does. Therefore simplex is better in that sense.

In the tradeoff, it seems fair to be 1/8 faster for no visual artifacts.

I like the look of Simplex better, but i guess it's just a matter of /opinion/.
Xalos asked for Simplex, and simplex he should get.

I like the look of Simplex better, but i guess it's just a matter of /opinion/.
Xalos asked for Simplex, and simplex he should get.
I never said he shouldn't. I said that Perlin is better, and I even offered to help him understand Simplex.

Did some googling, people seem to say that simplex is better. I thought Perlin was faster for 2D?

n^2
2^n

Therefore perlin is faster by ONE.
fair to be 1/8 faster
Um, where and why are you getting these conflicting numbers

Did some googling, people seem to say that simplex is better. I thought Perlin was faster for 2D?
perlin is faster for 1D actually.

Um, where and why are you getting these conflicting numbers
what do you mean conflicting numbers

simplex noise has a complexity of O(n2)
perlin noise has a complexity of O(2n)

32 = 9
23 = 8

and if you mean where I got the n2 and 2n from, just look up perlin/simplex nose complexity on google.
« Last Edit: October 23, 2012, 05:50:44 PM by Ipquarx »

what do you mean conflicting numbers

simplex noise has a complexity of O(n2)
perlin noise has a complexity of O(2n)

32 = 9
23 = 8

and if you mean where I got the n2 and 2n from, just look up perlin/simplex nose complexity on google.
That looks familiar, I thought I remembered the numbers 8 and 9 had to do with those.

what do you mean conflicting numbers

simplex noise has a complexity of O(n2)
perlin noise has a complexity of O(2n)

32 = 9
23 = 8

and if you mean where I got the n2 and 2n from, just look up perlin/simplex nose complexity on google.
"1/8 faster"

"1/8 faster"

He said meant 1/8th slower


9 - 8 = 1 / 8 = 1/8th slower

He said meant 1/8th slower


9 - 8 = 1 / 8 = 1/8th slower
Ah, now it makes sense

He said meant 1/8th slower


9 - 8 = 1 / 8 = 1/8th slower
Even then, big-oh notation does not directly translate to time taken, so it's incorrect to say it's an 8th slower.