ikoon

Hi, I am not skilled in vops, but here is the vex code, and file is attached. The principle is: For each found point (pciterate, while loop), you pcimport the point.number, and set this point.number to the "max_ptnum" attrib, only if currently iterated point.number is bigger than what is already stored in this attrib (all previously found point.numbers). At the end, you should have max_ptnum of all the found points. In the Set Attribute VOP, it is called Merge Mode: Maxiumum string Pchannel = "P"; vector P = @P; float radius = 10.0; // search within this radius int input = 1; // search in this input int maxpoints = npoints(1); int handle = pcopen(input, Pchannel, P, radius, maxpoints); while( pciterate(handle) == 1 ) { int found_ptnum; pcimport(handle, "point.number", found_ptnum); // next line stores the found_ptnum, only if it is bigger then previously found setpointattrib(0, "max_ptnum", @ptnum, found_ptnum, "max"); } pcclose(handle); Nearest_MaxPtNum_v001.001_VEX.hiplc

I am quite a Python noob, so I may be using wrong terminology. My situation: - I define my scripts as "independent methods" in .py files, and I import those files - I run those methods from the Tab menu or from the Shelf (with hotkeys), or from HDA buttons - more and more often I import "smaller" methods into "bigger" methods - my "way of organisation" is simple: I keep methods of one kind in one .py file with a descriptive filename - some methods add functionality to hou.Node, hou.PathBasedPaneTab, or so Please, should I start to bind methods to hou classes? Should I define my own classes? Is this an important milestone in the "evolution" of FX TD, to move from Python scripting to Python OOP?

Hi Daniel, I had a quick peek at your file yesterday, you resampled the boxes to a very high detail. With a Resample SOP node. Then you packed them and sent to DOP. I don't think that it is necessary and it slows down the DOP a lot. You may check the tutorial, what is the purpose of their Resample (and then Fuse), and how it should be used.

Hi Michal, you are scaling the object, but at the same time, you are translating it (because of the pivot of the scale). This is the main reason for the unusual behavior. Set the translate in testgeometry_pighead1 to 0,0,0 and set the translate in transform1 to 0,5,0. Then you could also increase the substeps of the dopnet and/or the bullet solver.

Hi Karan, it is not detail() but details() to read a string: https://www.sidefx.com/docs/houdini/expressions/details.html With details() you only need 2 arguments: @name_orig=`details(-1,"value")`

This is great Masterclass: https://vimeo.com/14612897 You don't need to see the whole video, first hour (or less) should help you do amazing things.

I am not sure what exactly do you need, but that python can be run in any String parm. Even in a new custom string parm in any node. Btw in Switch COP, the parm is called "index", not "input". So the third line will be this: active = switch.parm("index").eval()

Do you have a sample scene?

Hi Mono, something like this, in Python? If you want the result inside a String parm, you have to Add a keyframe to that parm and then change the language to Python. The file is attached. switch = hou.node("../switch") inputs = switch.inputs() active = switch.parm("input").eval() return inputs[active] switch.hiplc

Hi Tagosaku, I am just shooting in the dark: https://www.sidefx.com/docs/houdini/hom/hou/maxThreads.html Or environment variable HOUDINI_MAXTHREADS: https://www.sidefx.com/docs/houdini/ref/env.html

Hi Osmeldu, unfortunately I cant help you with the script, but have you considered using the Wedge ROP? Wedge Method, By Take. Is it sufficient for your actual needs? Otherwise, if you really need it to be scripted, then maybe you have to use hdefereval.executeDeferred()? I have only blurred knowledge about the executeDeferred(), and I would also be thankful if anyone could shed some light on that method.

Hi Mark, unfortunately I don't know how to distribute one OpenCL simulation to multiple GPUs in one computer. I think, that Deadline cannot help to distribute one OpenCL sim to multiple GPUs, as long as Houdini can utilize only one GPU. Simulations rely on previous sim frame. But maybe you could split the sim into more clusters, or slices, and try to distribute those? EDIT: the sim usually relies heavily on the CPU, so I think that the benefit of second GPU being used is not that significant. But of course, I am also curious about the future development.

Hi, I am using Houdini 18. I had this conversation with SideFX in April 2018. I don`t know, in which version the Lock method was implemented. But Houdini 16.5 was released earlier, November 2017. Sent: Wednesday, April 25, 2018 6:05 PM Subject: Online Bug: python or hscript to toggle the lock [SESI #60637] I've added you to an existing RFE requesting this functionality. RFE #85473 You may try the following, in the Python Shell. The autocomplete will show you available methods in your version of Houdini, but lockCameraToView is probably not there:

Hi Michael, you need to add the argument, True or False into the lockCameraToView() hou.ui.paneTabOfType(hou.paneTabType.SceneViewer).curViewport().lockCameraToView(True)

Hi @Lutojar I have Intuos4 and Houdini 18.0.353 With wacom driver 6.3.38-2 I get these @stroke_pressures: - with Use Windows Ink ON ... the pressure is recorded correctly (~0.01 - 1.0) - with Use Windows Ink OFF ... the pressure is always full 1.0 I don`t even have to restart Houdini to get these two kind of behaviors ... I just toggle the "Use Windows Ink" in Wacom Tablet Properties.