More Types of Noises

I think there should be more options for noise, and more kinds of noise! Blender, an open source 3d program has a tremendous amount of different kinds of noise, ROFL, it has it's own little FF in it (Nodes), though not as advanced. Could you please try to make the kinds of noise they have made?

We've discussed this quite a while ago.

Clicky! for Vlad's opinion on this!

Way back then I realized that Vlad was right: You can build almost any kind of noise in FF by combining what's there: Clicky!

i've been thinking about this recently, also. why not just make one noise component, but make it a versatile noise creator. i could even see having a separate image input just for this noise creator for seeding. imagine having just one or two noise creator type components, with all the current nodes/sliders/adjustments, but where you would have even more control and versatility of the thing. for instance, right now it's a bit tricky to adjust the white to black ratio in a perlin, let's say. but, what if you could adjust it any way you wanted, or what if you could actually draw in your own noise or input a small file to use as the base seed of the noise creator?

noise components are very useful, as dilla points out in those other threads, but i do find myself wanting more versatility in these.

I honestly agree with some things here. See, in my humble opinion I was expecting a ton more than just several noise patterns. Then I eventually learnt that you can make noise from combining noise. So I just assumed that FilterForge kept everything minimum as possible.

I think the next step in making it a much "Versatile" component is to provide a ton of presets or pre-programed snippets, macros, whatever. We already have fairly open discussion on this like "Noise lab" and often, while we discuss a lot, most of the snippets aren't available either way.

Put it this way: we're only provided the idea that we can make anything. We're hinted, clued, and given a vague idea that it's possible with just several components. But we're never given the actual thing.

I was one day showing a friend Filter Forge, and as we got to it's workings, he became confused at what was going on. Especially when he saw me connect node after node on each other to generate a particular pattern. That to him was a big mysterious field. Basically the problem was that he doesn't think "component based" as much. In fact Filter Forge and many other Nody-programs end up becoming "specialties" or something of people's preference in working environments. Now I tend to think node-based and my thought process works in nodes, which is why I can live with it. Others just can't figure it out.

I eventually ended up making a ton of snippets myself. It helps when you have a quick deposit of noise patterns for later use. It especially helps when you have an idea for a filter, but you run into a big wall. Now just apply that thought to Filter Forge. No need to make new components, probably just have "pre-created" patterns.

That's my way of thinking versatility.

Quote
Skybase wrote: I was one day showing a friend Filter Forge, and as we got to it's workings, he became confused at what was going on. Especially when he saw me connect node after node on each other to generate a particular pattern. That to him was a big mysterious field. Basically the problem was that he doesn't think "component based" as much.

Precisely. Most people don't think in terms of setting up processes that generate or alter images -- they think in terms of brush strokes and similar incremental changes. This mindset is the biggest obstacle in the way of mass-market adoption of node-based systems.

My opinion on the noise spam that I stated in 2006 still stands:
http://www.filterforge.com/forum/read...=9&TID=980

The opinion was strong enough back then, and it's even stronger now, considering these two things:

1. FF now has the Bomber component, with its often overlooked Layering:Balanced parameter, that can be used as a custom noise maker.

2. FF4.0 will have Groups which can wrap subtrees into singe noise components, making noise spam unnecessary.

But anyway, let's take a look at Blender 2.6 procedural nodes. I'm looking at this page: http://wiki.blender.org/index.php/Doc...Procedural. I'm not a Blender expert, but so far it looks to me as follows:

Clouds = pure Perlin noise.
Distorted Noise = FF's Noise Distortion does exactly the same, and more.
Magic = not a noise. Can be easily made with FF.
Marble = looks like a diagonal pattern perturbed with Perlin.
Musgrave = this is a good one. More on that later.
Noise = Perlin noise with small grain and high contrast.
Stucci = looks like Perlin.
Voronoi = FF's Worley-based noises.
Wood = can't judge because I haven't seen its "ring-shaped patterns".


Having said that, there are a few things I'd like to add to FF's repertoire of noise functions.

The main thing is the third (Z) axis for Perlin or simply Phase. Perlin is a 3D noise function, and for 2D application the Z coordinate is usually set to 0. However, there's an interesting use for it in a 2D app. If we expose it in the component intrerface, it would allow to smoothly, continuously "animate" the noise, as shown in this video: http://www.youtube.com/watch?v=OAABNjp7s1Y

This Phase input can be made mappable, which would lead to a number of interesting applications.

However, there are potential problems with this, the main one being the fact that our perlin implementation uses domain rotation for octaves to get rid of "cubing" artifacts (suggested to us by Ken Perlin himself), so the Z planes of the octaves are not parallel to each other.

Seamless tiling shouldn't be a problem since we edge-blend octaves in 2D after the domain rotation.

(Worley-based noises could also support Phase, but their code for seamless tiling is so complex and messy that I shudder when I think about revisiting it again.)

Also: before the release of the first beta, we experimented with octave distortion in our Perlin and other noise. Basically, imagine a Noise Distortion with different random seeds applied to every octave before they are blended together. This produced a liquid-like fluid look. It would be fun to revisit those experiments.

As for the implementation of the above, I hesitate about putting it into Perlin, mostly because this may slow it down even when these parameters are set to defaults. Perlin is a very widely used component, so it must remain as fast as possible. Perhaps, we could add a Perlin Plus component that included these things.

Vladimir, a z-component to perlin noise would be great! Actually, you already provided us with this within the "get_perlin_noise" script function, but I noticed that the z-coordinate does not "seamlessly tile". I am curious to see what a "z" animation looks like with the worley noises. Would the "noise distortion" component also have a "phase" input? And could we see the result of the "octave distortion" experiment?

Quote
Would the "noise distortion" component also have a "phase" input? And could we see the result of the "octave distortion" experiment?

That would depends on our decision regarding rendering speed. Noise Distortion, as Betis correctly noted, is just a "shortcut" for Offset and two Perlin noises (one for X, another for Y). If we implement it on the 3D Perlin, it would take longer to render.

At the moment, it seems to me that Z-Perlin, 3D Perlin, Noise Plus, or whatever we want to call it, should be implemented as a separate component, without affecting everything else within FF that relies on Perlin.

Quote
Vladimir Golovin 1. FF now has the Bomber component, with its often overlooked Layering:Balanced parameter, that can be used as a custom noise maker.

Sorry that I am not an FF expert to understand this, how can be used the bomber as a custom noise maker?

Also want to put here some nice examples of Worley Noise found on Google images or more especific also This one here

FilterForge finally starting to sink in?