Patch the vectors, matrices and quaternions behind 3D graphics yourself.
2 Sessions at 3 hours, live in English language
Recordings will become available the day after the live lesson.
Maths for artists who use computers. Particularly anyone who finds themselves transforming things in game-engine / GPU-powered worlds.
Over two sessions the goal is to learn the relevant mathematical concepts not just conceptually, but to start developing a kind of intuitive understanding that enables playful exploration. Then you too can frolic among the matrices and quaternions like a pony in a field of clover.
It works both ways. On one side, the applied and concrete: vectors, dot and cross products, matrices and transforms, orientation and quaternions — put to work building your own cameras, wrestling gimbal lock, and driving teapots (aggressively). On the other, the fun of visualizing abstract theory until it stops being abstract: signed distance fields, scalar and vector fields, gradient and curl.
It builds on the maths patches Toby K published in VL.ExtendedTutorials and goes further — we patch it all ourselves in Skia and Stride, with exercises to grow the intuition, and dip a toe into VL.Fuse. You help steer, too: propose a topic or two at the end of session one and we cover them in session two.
By the end, the matrices aren’t something to survive. They are somewhere to play.
What you will learn
- Vectors and Unit Vector
- Dot Product
- Matrices
- Stride entity system and transforms
- Cross Product
- The many ways to describe orientation
- Up vectors
- Gimbal lock
- Quaternions
- Screen space vs World space vs Foobar space
- Camera transforms
- How to make your own cameras
- SDFs
- Scalar Fields
- Vector fields
- Gradient and Curl of fields
Requirements
- Windows computer with latest vvvv gamma installed
- 3 button mouse recommended
- Second monitor recommended
- VL.Fuse pack matching your vvvv version (compatibility chart: https://thegraybook.vvvv.org/changelog/nuget-compatibility-chart.html)
- A suitable GPU for running VL.Fuse (run some help patches to test)
Course Outline
Session 1 – 10.11.2026
We will cover the lower level concepts and patch it all ourselves in skia & stride.
Does some specific mathematical abstraction confound you? There is the opportunity at the end of the first session to propose a couple of topics to be included in the second session.
Session 2 – 17.11.2026
We put the knowledge from session 1 together to make fun stuff like cameras and aggressively driven teapots. Plus we try to demystify VL.Fuse a little bit. Finally we will have time for a topic or two chosen by students at the end of session 1.
Glossary
The key terms from the course, with the vvvv nodes that go with them. Session 2 terms are added after the session on 17 November.
Session 1: 14 terms
| Term | Explanation | vvvv node(s) |
|---|---|---|
| Vector | A direction plus a length, usually drawn from the origin. Subtracting two points gives you the vector between them, and the order of the subtraction sets the direction. | -, + |
| Length (magnitude) | How long a vector is. | Length |
| Normalize / unit vector | Scales a vector to length 1 while keeping its direction. The result, a unit vector, always lies on a circle (2D) or sphere (3D) of radius 1. A zero vector has no direction. | Normalize |
| Normal | The direction pointing straight out of a surface or a line. Usually stored as a unit vector, but not the same thing as normalizing. | mesh normals in Stride |
| Dot product | Multiplies two vectors into a single number. For unit vectors: 1 = same direction, 0 = 90°, −1 = opposite. Doesn’t tell left from right. | Dot |
| Determinant (2D) | Like the dot product, but tells you whether one vector is clockwise or counter-clockwise of another (positive or negative). Doesn’t tell front from back. | no built-in node; built in exercise X2 |
| Cross product | Takes two 3D vectors and returns a third one at 90° to both. Swapping the inputs flips the result. It is longest when the inputs are 90° apart and zero when they are parallel. | Cross |
| Transformation matrix (4×4) | Stores position, rotation and scale of an object. vvvv uses row-major matrices: translation sits in the bottom row, the three basis vectors in the top-left 3×3. | TransformSRT, Decompose, ParallelEpiped |
| Basis vectors | The X, Y and Z axes of a space (right, up, forward). Their length gives scale; if they are not at 90° to each other, you get skew. | Right, Up, Forward (Forward in Stride returns −Z, use - (Negate)) |
| Identity matrix | A matrix that changes nothing: 1s on the diagonal, 0 everywhere else. | IsIdentity |
| World space / local space | World space is the outermost space of the scene. Local space is relative to a parent, for example a box placed relative to a group. | Transform, entity children in Stride |
| Cycles | vvvv’s unit for angles: 1 cycle is a full turn, 0.25 is 90°. Trig functions (Sin, Cos, Atan2 …) work in radians and need converting. | CyclesToDegree, DegreeToCycles, FromPolar, ToPolar |
| Euler angles / gimbal lock | Rotation as three angles (pitch, yaw, roll). At exactly 90° on one axis, two axes line up and you lose one: gimbal lock. Fine for static rotations, problematic for animation. | Rotate, TransformSRT |
| Quaternion / Slerp | Rotation stored as four values (X, Y, Z, W), with no gimbal lock. Slerp blends between two quaternions with a 0–1 value for a smooth rotation animation. | Slerp, Rotate (Quaternion), Decompose |