There have been a lot of new sampling functions added to VEX in the last couple of releases. I am no expert on statistics and would love an overview tutorial going over the sampling functions and what the various distributions might be useful for.

The docs have almost zero examples for these.

I would agree that having tutorials that go more in depth of not just the sampling but of other vex functions is ideal.

Perhaps Master Class tuts on each of the vex function catagories.

Mainly because it would help in exposing and helping ‘novices’ like myself realize what more I can do with vex.

I also, would like this ‘extra’ info/help to be seperate from the docs themselves.

Because the more I use the Vex functions docs, as I frequently do as quick reminders, the more I appreciate their current status of how they are structured.

Namely, they are succinct and lend themselves to being very useful for those already familiar ( not with the functions themselves ) but with the ‘topic’ at hand.

Otherwise, cramming too much information with examples and drawn out explanations in order to satisfy beginners and long time users, simply becomes too bloated. And really, in such a case they no longer really could be described as docs anymore in such a case.

It seems like a good idea to mix a comprehensive ‘how to’ with ‘documentation’ and I use two other software packages that do just that. I find in such cases they actually become difficult to find new info and/or do ‘reminder’ searches.

My experience ( with Houdini ) is that once I understand the topic/application the function is utilizing, the description of the docs becomes crystal clear and I am able to dive in and make use of the function.

I would think in your case, if true ( because I really don't know ) that there is a relationship between statistics and the sampling functions, that if one studies some statistics related to the terms described in the docs of the specific function one is interested in; I am sure those docs will become clear as to how to apply the function.

This happened with me when I wanted to make use of the vex spline functions. I could not figure out from the docs alone. But when I did a net search on the math behind spline functions, it became clear to me what the docs where talking about and I was able to start using them right away. And it( the docs ) also served as a good reference when I forgot about the function and needed to reference their application again or to try a variation of their application.

The Attribute Randomize SOP doc page [sidefx.com] has some images depicting the different distributions and some of the options. (There's a bug with those images in the Houdini build; it should be fixed for tomorrow's build, Houdini 16.0.571.) I've considered adding some of those images to the VEX function docs too, so maybe now's a good time.

Attribute Randomize internally is mostly an Attribute Wrangle SOP calling those VEX functions based on the parameter values, so you can dive inside to see what it's doing. If you turn on the “Visualize as Points” option, it'll give you a visualization of the distribution as point positions in 3D space, which is useful for visualizing some of the 2D and 3D distributions. If you're sampling orientation quaternions (4D), you can add a Marker viewport visualizer set to Axes and give it the quaternion attribute being output, it will show a gnomon depicting the orientation.

Please feel free to ask any questions you might have.

Wow, thank you Nick, I had no idea there were so many examples on that page. It looks like a great start. I think Attribute Randomize is also relatively new and I never really think about using it, but I certainly will now! It's certainly more promising than when I usually roll my own hacky randomizing functions.

At my current job I occasionally deal with statistical distributions of data given to me by scientists, which I then use to make more general-audience appealing images from. I often need to re-synthesize a new distribution from the source data. The scientists almost certainly apply many of the same statistical distributions to their work provided by VEX, so the more I know about them the better.

Cheers,
Jon