Fenolis

If you consider that evenly spaced points on a triangular grid are the "optimal" result of a point-relax algorithm, it shouldn't be hard to see why patterns emerge when you turn up relax iterations. The first image with gaps in the distribution is more "random" than a relaxed scatter (which causes the distance between nearby points to be more uniform) If the default pseudo-random noise used for Scatter does not suit your purpose, you can try using the Density attribute or write your own scatter algorithm using VOPs. Inserting an Attribute Noise before Scatter and then setting the Density on Scatter to use that attribute should give you more control over the distribution and better visualization of the output of the node.

The easiest way is to just select the preset so that all the parameters are set to the preset values, delete the preset, change your values, and save as a new preset with the same name. You can also look into your nodes presets files in Documents\{houdini_version}\presets\Sop but they are saved as .idx and not easy to work with.

You can use the Orientation Along Curve SOP to create alignment vectors for Sweep. I'm not too sure what you mean by deformed, do you have an example of the "correct" shape?

There is a tool under development here that may be of use to you. Other methods you can try are using Cluster attribute and Voronoi, although it will take some experience to control well.

If only VEX had a switch() function that let you go case by case... You can implement your own, example here. The idea is to set up an enumerator (which is kind of like using integers anyway) that compares an input to a kind of key-value in a dictionary. Depending on which value is matched, a different output can be chosen, and a default for if no match is found. I'd still recommend converting/outputting integers as it's probably faster than comparing string values, and the Switch works with ints anyway. One thing to note is that the Switch SOP doesn't know what its inputs are, only that they are ordered by input index. It also has no way of ensuring that inputs are unique, so if you accidentally plugged 2 Carrots into it, it wouldn't be able to tell which one you wanted. Best to just stick with ints, imo.

"Display particles as Discs" is like copying a Primitive Circle/Sphere to Points. The Copy to Point SOP will assign pscale to primitive objects as well and should suit your visualization needs.

The JoyOfVex series will be of much help to you, this chapter in particular for this problem.

Some links that may be of use to you: - SideFX Building Generator - Simon Verstraete's 2-min House Generator And you can always install Labs tools and see how the Building Generator works.

Skip to 49:35 of this video for information on adding images to Help. Jeff Wagner talks about the process in great detail and I would recommend watching the video in its entirety if you have the time.

1. If you think about this carefully, random may not be the way to go here. You can use the modulo operator "%" to step increments - such that the last switch in the chain cycles between say, 10 inputs, 1 per frame. The switch before that uses floor() function, cycles only every 10 frames, so that every combination of the last 2 switches is produced uniquely. switch_input1: $F%10 -> goes 0, 1, 2, ..., 8, 9, 0, ... switch_input2: floor($F/10) - > goes 0, 0, 0, ..., 0, 0, 1, ... One way to simplify this idea - if I have 4 switches 0000, each with <=10 inputs, then I can get every unique output by cycling the digits like counting upwards - 0001, 0002, ... 9999. 2. To control this kind of statistical weighting, you will likely already have a "Has hat" toggle parameter. You can use Wedge TOP to batch out the 100 with hats, and the other 9900 without. 3. I'm not sure how to help with this.

https://www.sidefx.com/forum/topic/80778/

Circle SOP, set to Open Arc. PolyExtrude - use Transform Extruded Front to set height You may have to use Reverse to flip normals since I expect you'll be viewing the inside of the curve screen_arc.hipnc

1. Converting line art to geometry can be done using Trace SOP - import image to trace, resample/remesh to get detail. Works best if shapes are closed. 2. You can use the integer attribute active to define RBD pieces that cannot move. Example 3. After running the simulation, you can Group by Bounding Box and use a Blast SOP at the end to remove pieces outside the box before exporting .abc

You can take a look at the Sweep SOP's documentation by clicking the question mark on the node, or just Googling.

The problem occurs because parts of the mesh are thinner than the voxel size specified during conversion. This results in the creation of voxels that are not "full" and thus allow other voxels like fluids to leak through them. The solution to this kind of leaky problem is to reduce voxel size during conversion (to more accurately represent the bounds of the geometry), matching the voxel size of the fluid sim to the voxel size of the geometry (or have the fluid voxels be smaller, so they don't penetrate the thicker boundaries), or use thickened/solidified proxy geometry.