(Previous title: “How to use the getattrib VOP properly? geometryvopglobal1 for second input?”)

Hi,


I am trying to get an evenly distributed noise (along the Z) on my curved shape by using position from the second input (a grid), but I get lost.

Why doesn't this work what I did? I guessed that the geometryvopglobal1 must be only for input 1 and that I need to also get ptnum of the second input with another getattrib VOP, but that gives me no noise.

How does this work? How would one go about easily getting attributes from other inputs. Can't I have multiple geometryvopglobal's or something?

I really appreciate the help because I hate being confused. I have a HIP file attached.


Thank you.

OdFotan
Why doesn't this work what I did? I guessed that the geometryvopglobal1 must be only for input 1 and that I need to also get ptnum of the second input with another getattrib VOP, but that gives me no noise.

if you plug output of getattrib1 to P input of geometryvopoutput1 you'll see what it's returning
in order for Get Attribute VOP or Import Point Attribute VOPs by ptnum to work correctly the geometries have to have the same topology

however to get even distribution of the noise over XZ plane you can just set Y to 0 on the P before plugging to noise
or to get custom sampling you can either use UVs that you generate on your input geo or store rest attribute and use those to sample the noise

tamte

I'm a bit confused. I thought that it made an uneven noise because the position was more ‘dense’ on the Z axis at the steep area (viewed from top, let's say) so that if you would make @P.y = 0 it would make no difference. (see attachment picture)

I don't think I really understand very well how noise works. What are you feeding in when you feed in pos and how does it handle that? When you disconnect the pos and manually type in a number at ‘3D Position’ you get obviously one point off a noise, but where is that noise? Has it just the normal xyz dimensions and does it exist ‘everywhere’? And when you put in @P into it's pos you just get the actual position of the noise on that actual point where the noise hypothetically also ‘exists’ just as that point does? (Like a 1 to 1 mapping?) And what is it exactly? Is at a value float value that then subsequently gets added to the value of ‘pos’ inside that noise (and it's output)?

Why did this getting thinner mountains on the more steep part happen anyways? (My head spins. )

And why does the topology have to match? What exactly has to be the same? @P or the @ptnum sorting? Isn't there just like a more ‘abstract’ noise that you can apply to everything no matter what?

And so if you use ptnum of the main ‘geometryopglobal’ on that getattrib you automatically get the ptnum of geo input 2?

Seems my initial suspicion was right, merely a flat geo on @P.y doesn't suffice for a truly evenly spaced noise:



Can somebody help me out with this? Is there a method to get it?
Also would like a good explanation of how that pos input on the noise and the noise itself work. Previous comment hopefully clearly displays my confusion.

-

edit: I took a look more carefully to your comment and I realised I hadn't really checked the UV and rest method (I don't know much about the concept of uv and rest yet), but these also have this problem. Even more so as the noise kind of ‘pauses’ where the extrude has started. HIP file included.

-

There are a lot of questions there! I'd first have a look at this SESI video, it will answer a lot of your questions: https://www.sidefx.com/tutorials/working-with-noise/ [www.sidefx.com]

It contains a useful PDF on noise as well.

what do you mean by “evenly spaced noise along Z” ?
Do you have a picture reference for what you trying to achieve?


Something like this?

mrCatfish

Thanks, will check out the video.

Andr
what do you mean by “evenly spaced noise along Z” ?

I don't mean it being squashed or something. Just the noise, in normal square ratio, like ‘transfered’ or ‘put’ onto a geometry properly from the top view, without it being stretched (like it stretches in the Z direction in that GIF I made). A noise that remains what it is, regardless of the position of the points. I guess kinda like Height Field works for volumes?

This is why I tried putting a grid in the second input of the VOP and get a noise on it's position to transfer that to my shape. But that didn't seem to work out.

Hope I'm clear.

If I understand, you see it stretching because you lose resolution density as you increase the polyextrude distance. But it's not really stretching, the noise is the same, but being sampled by fewer points in the area.
Maybe try to link the division parm of the polyextrude to the distance, in order to maintain a consistent resolution?

Andr
If I understand, you see it stretching because you lose resolution density as you increase the polyextrude distance. But it's not really stretching, the noise is the same, but being sampled by fewer points in the area.

You are right, sir.
Thank you.