Baked into threejs there are a number of Math utilities that can be used to help with various tasks such as clamping values for one example. Other things that can be done with the various methods include things such as converting a degree value to a radian value, or getting pseudo random values by way of the seeded random method. There are a lot of other great methods that help with the process of creating what is often referred to as an alpha value as well ( a number between 0 and 1 ).
However there is not just thinking in terms of what there is to work with, but also what is missing when it comes to a collection of methods such as this. With that said I think I should also write about one or more additional things that are not in this math utils object, but should maybe be there. I say maybe because I do not expect everything to be there of course. Even though there are some usual suspect type methods to work with here there is still going to be a need for some kind of additional utility library outside of threejs that is a kind of extension of what there is to work with in MathUtils. I still find myself writing my own code for various things in order to fill in the gap sort of speak.
Speaking of filling in the gap when it comes to making my own math utility module, I am always working on top of threejs, so I can just simply wrap many of the Math utils object methods. For one example of this there is the subject of clamping and wrapping values. When it comes to clamping I can just have an abstraction of THREE.MathUtils.clamp and be done with that. However things get a little messy when it comes to having a wrap method. There is an euclidean modulo method, but that alone does not work the way that I would want it to for such a method so there is having something that is maybe a bit more that just an abstraction for that one.
Many libraries will have a clamp method that will clamp a value to the range of a given max and min value, and threejs is no exception with this as there is very much a clamp method in the math utils object. However there is not just clamping values but also wrapping values as well. That is that on top of having a method that will clamp a value between a min and max value in that it will just not let the value go lower or higher there would also be a wrap method that will wrap a value that goes beyond max back to the min value and vice versa. It would seem that there is no wrap method in the math utils object, thus I have to get this method by using or making something outside of the library. There is however an Euclidean Modulo method that is kind of similar to what I would expect, it is just that I also would like to have something that works a little differently when it comes to negative numbers.
I have wrote a blog post on this subject of wrapping Vectors and primitives in threejs in which I was able to find a decent wrap method, and I also made one of my threejs project examples that is a wrap module in which I am building on top of this kind of thing. In the past I also wrote a post on the wrap method of the grame framework called phaser which has a wrap method that works great.
When I first wrote this post I was using r135 of threejs. The last time I came around to do some editing of this post I made sure all the examples where still working okay with r146, and thus I was following the style rules I set for that revision. Always check what version you are using when reproducing things on you end, as code breaking changes are made to threejs often.
1 - Basic example of threejs math utilities using degree to radian method when setting the position of a mesh
Maybe one of the methods that I find myself using the most often would be the degree to radians conversion method that there is to work with in this Object. Although it is not so hard to just do this with a simple expression because it is a such a common task of course there is a method for this in the Math utils method.
As I have mentioned in the clamp example there is not just clamping, but also wrapping number values. In this example I am again doing more or less the same thing as in the clamp example, but now I am using the Euclidean Modulo method to wrap numbers rather than clamping them.
There is using the plain old Math random method and also many other methods that are based off of it. However all of these options are not deterministic in nature, that is that when called they will not give the same numbers each time. In other words some times I might want to have some kind of solution where I have random numbers in a range, but each time I reload the page I get the same set of random numbers. So then they are not really random, but predictable, yet they look kind of random if that makes any sense.
The smooth step function will return a value between 0 and 1 that is a percent kind of value based on a video value to evaluate as the first argument compared to min and max values given as additional arguments. However the value returned will be smoothed or slowed down depending on how close the value is to the min or max value. A good way to go about getting an idea of how this method works might involve having a group of objects that all move by a fixed pixels per second value, then have another group that moves by a variable pixels per second value.
In this example I then have a helper function that will create a group of mesh objects with differing max pixels per second values in the user data objects of each mesh. I then have a simple update method that will move a group of mesh objects by the max value on each frame tick, and another update method that uses the Vector3.distanceTo method and MathUtils.smoothstep to get a variable pixels per second rate.
When this example is up and running group2 will just move at the fixed rate that is the max pixels per second value for the mesh object, while group1 will slow down and speed up based on distance and smooth stepping.
The lerp method can be used to change one value to another value over time. There is also a lerp method of the vector3 class that is also worth checking out that does the same thing as this lerp method only with a vector3 object. So for this example I am makinguse of the vector3 lerp method, but also the math utils lerp method when it comes to dealing with alpha values that are just single number values rather than an vector3 object.
Each time I create a geometry, material, mesh object, or just about anything to that effect a UUID is generated for the object each time for me. There are only so many reasons why I would even use these IDS to begin with, one of which might be to use the get by id method of the object3d class. I can not say that I use that though, as I prefer to use the get by name method. However I have found that there is one use case in which I would maybe want to call this method and that would be when working out any kind of logic that involves generating JSON string data in the object format. This Object format is what I would want to use if I am to use the parse method of the Object Loader to create the JSON text into a workable object.
This example is based on source code that I worked out for my example on the animation mixer as I would like to gain a better sense of how the JSON data is formated for scene objects that will contain lots of geometry, materials, animations and so forth. Anyway the use of the generateUUID method would come into play when it comes to creating the uuid keys of course for the various types of objects.
Also some of the things that come to mind are methods that I can not say that I use all that often such as an nth root method for example. That is an example of the kind of method that should not be in threejs because it is for the most part unneeded bulk. On the rare occasion that I do need an nth root method for whatever reason that is something that I can add by way of another library, or even a single stand alone method.