I saw this demoed, but have never found it in shipping version.

How do does one do a region render like in Maya and XSI?

Stupid question but thanx.

Also where can I find a solid tut on Image3D proccess. Creating volume shaders, integrating them with metas and particles, and render outputs.

Dave

Hey Dave,

In the Object or SOP viewport, hit Tab, type ‘ren’ and you should see ‘Render’ operation. Once you've selected the Operation, click and drag a box in the viewport.

There are some Image3D tutorials available at either of the following links:

http://tut.techimage.co.uk/video_tutorials/by_topic/effects/index.html [tut.techimage.co.uk]

or

http://www.vislab.usyd.edu.au/sidefx/houdini_video/by_topic/effects/index.html [vislab.usyd.edu.au]

There is a Contrail tutorial here:

http://www.techimage.co.uk/products/houdini/houdini_tutorials.html [techimage.co.uk]

Cheers!
steven

Also, if you want render regions in your renders as opposed to viewport, you can play with the Crop settings in the camera object.

I am sorry, and with all repect to Techimage. But their tutorials are badly written. So hard to follow. They use confusing wording. For example they use
'8.26 Firstly to create the new attribute lay down an attribute operation and rename it growth.'

Now does that mean append or not to append. Thats what I mean. Also they don't italize names of operations or names of tiles. I see this throughout their written tutorials. They need to use Houdini standard semantics.

DaveR

DavidRindner
I am sorry, and with all repect to Techimage. But their tutorials are badly written… …Also they don't italize names of operations or names of tiles… … They need to use Houdini standard semantics.

DaveR

Alright, fine…. I'll say it, then:

"You're welcome, David".

Assuming of course, that at least the bit about how to render regions was helpful (if perhaps not semantically sound).

Yes, I was remiss in thanking you for the links. Please let me apoligizre and offer a belated thank you. The REALLY good one was the Pete Bowmar's video. Image3D is actually simple, and to me reminds me closely of RealFLow procedure.
I am sorry, I did find Techimage tuts a bit confusing. I already knew the material they were covering, except Image3D part. The issue is that they tutorials don't make explicit statement, on some steps, as to whether or not the newly created tile was to appended to network or lay free.

The entire Image3D and Houdini volumetrics can be distilled into following;
Define point cloud from SOPs or POPs
Create Metacloud SHOP
Create Image3D ROP using point cloud for placement and SHOP for volume texture. I.e. baking a volume texture.
Apply resulting i3d sequence to Fog object.

Dave

DavidRindner
Image3D is actually simple,

(… from the user's point of view). Yes it is.
They've done an admirable job of packaging all the nasty details into a simple series of steps.

DavidRindner
and to me reminds me closely of RealFLow procedure.

And this is where you're reading way too much into similarities that are actually quite superficial.

DavidRindner
The entire Image3D and Houdini volumetrics can be distilled into following…

I'm assuming you're referring to the comparison you were trying to make between POPs/I3d and RealFlow in your other thread, so I'll respond here:

Yes; they both start with a bunch'o particles/primitives, and yes; they both end by interpreting this set as discreet components of some volume, which they then sample at render time.
Alas, this comparison is far from exhaustive. Specifically, it fails to look at exactly *HOW* the particles got to where they are in the first place! Wait. Before you rush out and say “well… *forces*, of course; how else?… duh!”; let me point out that, even though you'd be right in principle, your statement would be roughly equivalent to saying “to get to the moon, all you need is a big ACME rubber band”.

Without getting into technicalities, I propose a little experiment; a little “challenge” if you like, to get a taste of what I'm talking about.
Ready? Here goes:

1. Create a cube (box) with 8 points and take it into POPs.
2. Give each particle a random velocity as an initial state.
. Impulse Activation: $F==1
. Impulse Birth Rate: 8
. Const. Activation: 0
. Initial velocity: 0
. Variance: 0.1
3. Append a Force POP and add some turbulence.
. Turbulence: 3
. Exponent: 0.1
. Amplitude: 0.8

And here's the challenge…

4. Ensure that, after step 3, the lengths of the segments joining any two particles in the original cube haven't changed. i.e: those lengths must remain the same regardless of the forces acting on the particles. You may *not* modify steps 1 to 3 to achieve this – those forces are allowed to “push” your particles around in whatever way they see fit.

5. Ensure that your solution works equally well for 1000 particles connected in any arbitrary way.

That's it. Simple.

In steps 1-3, the particles are autonomous: a change in one particle has no effect on any other particle – a bunch of flying pebbles. You may choose to think of them as defining a volume if you like but that changes nothing.
In step 4, however, any change in one particle affects *every* *other* *particle*! (roughly equivalent to solving a cloth simulation).
And a full CFD solution would be *significantly* more complex than this little example.

So what's my point?
A robust fluid dynamics simulation is far, far, *far* from trivial, and not in *any* meaningful way comparable to something like Houdini's POP and/or Image3D systems (except in the most basic of general terms).

A likely candidate for a quickie VEX plugin? No.
Likely to show up as a freebie post on some forum? No.
Any solutions available? Yes! Mark's Houdini<->RealFlow geo translator (see elsewhere in this forum).