Hi,
I noticed polywire sop does not ( always) return a constant number of points, when applied to some animated curves ( i.e some lsystem curves with wire simulation )

Is there a way to achieve that ?

Hi matthias, I have planed to share this asset soon, but I lack some time work on it these days. To be honest, I wanted to try and sell it for a couple of bucks on Orbolt asset store, because as I am building it, it's becoming quite a “sophisticated” asset. I spent a little to much time and sweat on that one, to just give it away.

:roll:

That being said. the smooth transitions between trunk and branches is not really my doing. The polywire sop takes care of that automatically when given a set of polygons.

thanks for the feedback ^^

I think nice to have the option not to export a ton of geometry at render time ( if you have millions of grass blades ) when it can be generated directly by mantra. The main advantage is there I guess.

Actually, now I use both in my setup, because of course I wanted my grass to be dynamic. Simulating hundreds of thousands of grass blades was not an options. So I simulate only some guides (Sop version ) . and the mantra interpolates a ton of new blades in between them.

It works pretty fine, but I must say, I though it would be a lot easier to set up.
I learned a lot though

Hi, a good thing to know ( if you don't already ) is that houdini will prompt all the methods of an object if this object is stored in a variable.
if you type :

hou.node('/obj/geo1'). …. # houdini will prompt nothing

#but if you put the object in a variable, and then call a method on this variable

n = hou.node('/obj/geo1')
n. … # this will prompt every available method



Hope it helps

I finally made it work.
I'm still not sure about what I was doing wrong. I tried to work around the problem I had before, so I don't know which one of the changes did the trick. But I suspect it was the guide geometry that was (one of) the problem(s). By that I mean I didn't specify any reference for any guide geometry on the mantra fur_procedural shader
I also suspect that it needs to have absolute paths to nodes: I used “opfullpath” expression everywhere I had a relative reference to a node in the asset.( it seems obvious now, but weirdly , there was no message to warn me about absolute references in the asset, as I have usually)

I attached an image of the current result so far.
I was a little surprised to see that this image took the same amount of RAM ( around 15 giga) , using Sop fur procedural and using Mantra fur procedural.
Besides the fact that Houdini doesn't have to generate fur curves when exporting to mantra ( if I understand it well ), is there some other advantage ? memory wise ?

Hi, I am really scratching my head on this.

I am trying to build a simple grass asset, using houdini fur tools.
I used a furSop to build some curves, and a couple of cvex shaders for the skin attributes and guides attributes.
I can see the effect of the cvex code on the curves in the viewport, and in mantra.
The strands are shaded, I took care of the orient attributes by randomizing it.
Everything works correctly.
I then decide to use a fur procedural, to be able to up the number of grass blades to a bigger without making my pc burn.

I feed the necessary information to the fur procedural node.
It works fine to, I struggled a little with the orient attributes, which I set in sops ( I had to migrate it to cvex guide shader for mantra to pick it)

All is nice, but the bad part is coming …
As soon as I open the type property window to promote some parameters. The fur procedural “stopped working” ( I saw it in the progressive render view, which was active at the time )


I retried the exact same set up from scratch, saw it work again, and then stopped working …
And now I can never get it to work again.
I really don't get it.

Does someone have a clue ?
Ps: I join the asset containing the set up I am talking about
Sorry for the long explaination

Hi, I noticed some pretty long saving time when saving a hip file on my NAS. Defenetly longer than file size would require.
I am not at all a specialist regarding network things.
I do not experience this with any other program, ( after effects , photoshop or others … )

Is there something special to configure for houdini ?

Hi,

here is another result obtained with my tree asset.

https://vimeo.com/139057373 [vimeo.com]

critics are welcome :?

Hi, I am trying to make a progress bar ( in a python panel ) update during the cooking of a copy Sop ….

Is it even possible ?

I have been exploring the hou module, and I can't find a way to access a UI element in a python panel.
I also tried the other way around : making the progress bar “listen” to the copy Sop. But I didn't succeed either. but knowing how power Qt is, there might be more hoep in that approach.

A ticket have been added
“Side Effects Support Ticket: #29530”

At least they'll look into that.
In the mean time I thought of a workaround … simply parent the camera node to a null, and animate the null. I tried it in a hip file with the bug, and the camera just resets to 0,0,0 as usual , but it doesn't matter anymore 8)
I hope it does the trick

hmm nope I made a mistake, It just happened when handle in “aligned to world” mode …

grrr it's getting VERY annoying. I can't think of any workaround that nasty little bug.

I think I finally narrowed something down !
It seems to happended only when the handle of the camera is in mode “aligned to object” .
I use the shortcut very often : “M” key


Can anyone confirm that ?

Hi, just of the top of my head.
I think introducing a temperature attribute could work. trying to mimic caramel becoming more viscous as it cools down, particularly on thin parts.
Maybe treat the “high curvature” zones separately.
Just thinking out loud

Hi,
can you post a hip file ?

I didn't activate self collisions in my sim. Sim times are already high enough

Looking at another sim I ran, I can tell that, for the other wire affectors, wire width is taken into account. I can see some wheeping branches clearly “coiling” around a bigger branch. It's a little hard to see in these conditions though. I should test this in simpler conditions, but it's Sunday, and I am a little lazy

Is this “.” alone ?! or part of a bigger expression
usually in houdini, a “.” can be the relative path to the current node ( the one you're working on).

It would help a lot to read the expression itself though.

Hmmm, but both of them (global geometric and local geometric) are not taking the wire width into account, if I'm right, only SDF is doing it (with really high sym times), do you have a workaround?

No, unfortunately I don't.
But are you sure SDF method is the only one taking the wire width into account ? It seems strange to me.

I found so far that the best results are obtained using “global geometric” method (on the wire solver node). With a low “max collision resolve passes” (also on the wire solver node) ,the param is set to 8 passes in the preview below. Collisions are decent, they play their role. Using “local geometric” method with a far bigger number of samples resulted in crappy collision and deformations.
So for now, global Geometric is the way to go ^^

a preview :
https://youtu.be/FyI91OUXeaY [youtu.be]

Hi everyone, I am currently working on an asset that creates a lsystem tree , and animates it with dops. I am pretty happy with the result so far.
Here is a small example:
http://youtu.be/mECqmnfiOPk [youtu.be]


My question is about handling collisions for such a simulation.
In the example above, there are no collisions at all, whitch is not a problem in that case I think, but in certain situations (e.g stronger wind, denser foliage or weeping branches ) the need for decent collisions is much more obvious.

What is a good strategy for optimizing wire/wire collisions ?
My current system is : a base lsystem that goes for the few first generations, and a bunch of sub-lsytems attached to every end points of that base geometry. With no collisions, obviously, sim times are ok. But as soon as I activate collisions, and up the generations on the tree asset, sim time goes up pretty high.

I can't seem to able to reproduce this behavior right now … I will try to post whenever it comes back.