This is quite easy, assuming you're writing out some steps to disk. Once your car animation is complete, delete all but 4 points, one from each wheel (you could use some point on the axle, or really any point if the wheels aren't turning). Write out this as a sequence to disk. This will be fast to read back in because it's only 4 points per file. Next, load back in with a file SOP followed by a trail SOP. Set the trail SOP length to $F and ‘connect as geometry’. Presto, you have your 4 tire marks. Now use a ‘measure’ SOP to measure the perimeter of these. Use a prim() function to get access to the primitive attribute, and use that in your rotation calculation.

graham
From the first script that is run you could execute a source command to manually source your other script.

It's a bit trickier than that, as you want to make sure that your environment is consistent, i.e. that all appropriate scripts get read, regardless of whether they are present. Ergo daisy chaining scripts doesn't work really well.

One solution is to write a wrapper to launch Houdini which effectively does:

houdini myFile.hip /path/to/setup.script

And then the setup.script looks at the HOUDINI_PATH, reverses the order, and then looks in each location for a 123/456 or some other named script. If it finds it, it runs it; repeat for all remaining locations in the HOUDINI_PATH.

Immediate openings for Houdini Technical Directors / Generalists who can create realistic terrains, environments and effects for a heavily science driven project for National Geographic. We are looking for one junior/mid level TD and one senior level TD through September 2011 with a possible extension into October. We are a small team working on an all CG stereo fulldome (Planetarium) show about weather on the planets and moons of the Solar System. We are based in Culver City, a few blocks form Sony Studios.

Please send resume to resumes@floqfx.com

sdugaro
Does every expression in every parameter of a network
inside an hda always evaluate? I have custom expr dso
in a file sop that always prints out to the shell when it
can't find a file. I'd like to suppress this.. but nothing
seems to work. i tried using switch sops, thinking that
the node wont cook unless its in the cook path, but it
appears that the file sop expression is evaluated prior
to the network being cooked and makes no difference.

That seems a little odd. Have you tried the performance monitor to see what's in the cook chain? It might be that something unexpected is referencing your file sop. Does an hscript expression function behave the same way?

Albin
Hi.
When I move points with handles it's fine but if I try to enter specific values in spreadsheet the edit node goes in to soft lock state.What's the reason for this and can it be changed ?
Thanks…

That is because you've interrupted the flow-through nature of SOPs by hand-editing specific values. Thus that particular SOP is now holding precious data, which is what locked data is – it's modified in a non-procedural manner. Personally, for the rare cases where I do edit data with the spreadsheet, I usually write it out as a .bgeo and read it back in – that way I don't have to keep the data in the .hip file.

roughsporty

Antoine Durr
FYI, Apple no longer sells that card, at least on store.apple.com

But it is still available even if it's not available on apple store

But you have to wonder why they took it off the market so soon after introducing it. There are still plenty of 2008/2009 Mac Pros out there to warrant having it available unless…

twod
Is Houdini causing this, or is this a general systems stability problem? Have you tried downloading the driver from Nvidia instead of using the Apple-supplied driver?

I ran into one issue with rendering to a framebuffer object with multiple attachments that caused the system to freeze, then go into a “flicker-fit” as vsync went all crazy, requiring a reboot. Is this the sort of thing you're seeing?

We've seen the flicker-fit on just about every machine we have. But what's happening here is that flipbooks cease to function, mplay will have a sequence loaded and start showing black frames. Houdini will start eating 100% of a single core, and once you've killed Houdini and all the mplays, the rest of the UI is unresponsive (as if the Finder was busted). Eventually you're stuck in a corner and have to reboot.

roughsporty
What do you think about the EVGA GEFORCE GTX 285 PCI-E GRAFIKKARTE 1 GB RAM ?

FYI, Apple no longer sells that card, at least on store.apple.com

twod
Lately, ATI cards have been much more stable on OSX than Nvidia cards. The exception may be the Nvidia GEForce 285 and Quadro 4800, which Nvidia

If you mean the GTX-285, I beg to disagree on its stability. We're rebooting 3-4 times a day, as the whole machine is brought to a grinding halt. Even ssh'ing in and running reboot won't successfully shutdown the machine.

On the issue of drivers, does doing a System Update automatically update the drivers to their latest version, or is that something I should do separately?

So we're running Houdini on Mac Pros, OS X 10.6.4, mostly Nehalem 2009 Mac Pros, some 2008's, but no new 2010 Westmere versions yet). Are there any graphics card options for these that don't have huge GL issues? I feel like I'm back on Houdini version 4 or 5, on the bleeding edge!

Does Houdini really run better with the Quadro FX 4800 than the GT-120 that shipped with it? I know that drivers (at least on Linux) are supposed to be better for the Quadro FX series). Remaining choices are few and far between: the nVidia GeForce 8800GT that shipped with the 2008 Mac Pros, the nVidia GTX-285 (of which we have a couple), and the ATI HD 4870.

Does anyone have a setup that's working reliably?

Either with L-Systems, or as a particle system with controlled splitting based on the external data. The latter is probably more in line with how the algorithm works. For example, the split POP can use the external data to decide if the given particle should split. On the split, you can determine where the splitees should start, how many there should be, etc.

phtj
Is there a way to export to obj, and also to export colour?

Nope, afaik .obj does not support color. About the only thing that can be point varying are texture vertices and normals. The whole notion that you could have arbitrary per-point or per-vertex data is not something that the designers envisioned.

http://www.martinreddy.net/gfx/3d/OBJ.spec [martinreddy.net]

Jenny
Lens curvature can have a value of positive or negative, and simply distorts the film plane in a kind of parabolic shape (leaving the edge unchanged).

It's really only there for backward compatibility though, and it would be better to stick a grid in front of the camera with a surface shader to perform the distortion.

I was wondering what the heck it was doing! That visual explains it nicely, thanks.

melander
I am very new to this can you please clarify SOP?

Any of the panes? where do i turn on detail view?

There's a menu option at the top called ‘help’ and the first entry is ‘start here’. Highly recommended.

oat
suppose you have one houdini file A containing the geometry transformation operation nodes.

how to import another houdini file B which contains the geometry to be manipulated only into file A rather than opening two houdini sessions and copying the geometry nodes from file B and paste it in file A?

Or just write out the geometry from file A into a .bgeo file and read it into file B with a file SOP. If you're talking more about a process, then an HDA might be appropriate, as that essentially encapsulates the process and makes it portable.

You could either set up a take that sets your switches and then in the ROPs use that take; or, in the ROP's prerender script enter:
set -g RENDERPASS = 2
and in the switch SOP have $RENDERPASS.

danilo2
Could you please explain one step of the ‘meantime solution’:
“…and make the atjob command source that cmd script.”

atjob -p 1 source /my/path/myscript.cmd

danilo2
Hi!

I'm trying to watch every second for a change in node with a ajob HScript function.

I'm trying to redirect the output of this script to Textport or linux shell (I'm wondering how to do it both ways )

so I'm trying to do something like:

ajob -p 1 opinfo(“/obj/lsystem_object1/lsystem9/”) // but I get no feedback in textport nor in shell

I was trying to connect it with functions such as “unix echo …” or simple “echo…”, but not succesfull

I assume you mean ‘atjob’, not ‘ajob’. I tried some simple things, and I think that command is pretty busted – its parsing is completely borked – put the whole thing into double-quotes, and it ceases functioning!

The problem is that you need backticks, e.g.
atjob -p 1 echo `opinfp(“/obj/somewhere”)`
but the backticks will evaluate now instead of on each itereation. I think this is worth sending in to SESI.

In the meantime, I'd put all your script into a .cmd file and make the atjob command source that cmd script. Note that regular ‘echo’ statements will not work, as the atjob processing is detached from the textport. But doing ‘unix echo’ from within the .cmd file should work.

LFX
Yes, I'm talking about:

vertex(“.”, 1, 0, “Point Num”, 0)

This should give you an infinite recursion error. To evaluate the vertex() function, vertex() has to make sure that the requested SOP (“.”) is cooked. But to cook that sop (which is us!) the SOP has to evaluate its parameters, one of which contains the vertex() function, hence an infinite loop. $PT will evaluate properly even in SOPs that iterate on vertices.

hoknamahn
There is no vertex normals in Houdini. Only point ones.

So as you have discovered, Houdini can indeed use vertex normals. While they aren't advertised too heavily nor is Houdini optimally set up to use them, they will indeed have the intended effect on the shading.

You can, for example, make unique points, calculate normals, promote them to vertex attributes, and then reconsolidate your points. The vertex normals will now create hard edging even though your polys are sharing points. GL will not display this correctly, as it only deals with point normals. However, Mantra will indeed render it correctly.

You can visualize vertex normals by creating a custom attribute vector in the display options ('d' on the geo viewer window, create-attribute-vector, call it vtxN, attribute ‘N’, class vertex).

The process of saving to .obj (which uses the gwavefront standalone) converts point normals and calls them vertex normals inside the .obj. Vertex normals are ignored. It is probably time to refile that RFE for gwavefront to handle vertex normals.