3D Linear Transformations in Action: Basis Vectors, 3x3 Matrices, and Rotations (3Blue1Brown)

In this quick footnote, the host extends linear transformation ideas from two to three dimensions. The video explains how a 3×3 matrix describes a space transformation by tracking where the three standard basis vectors I hat, J hat, and K hat land. By focusing on these basis vectors, we visualize the action of a transformation and compute the image of any vector XYZ as a linear combination of the matrix columns. The talk also covers how multiplying matrices corresponds to applying two transformations in sequence, a concept that underpins rotations in 3D and practical tasks in computer graphics and robotics. The discussion ends with a teaser for determinants in the next video.

• Basis vectors determine a 3×3 matrix
• Matrix columns encode where basis vectors land
• Matrix multiplication represents composition of transformations
• 3D rotations can be built from simple transformations

Overview of 3D Linear Transformations

This video serves as a bridge from 2D to 3D linear transformations. It emphasizes that, just as in two dimensions, a space transformation is completely described by how it acts on a basis. In three dimensions there are three standard basis vectors I hat, J hat, and K hat, corresponding to the x, y, and z axes. By tracking where these three vectors land under a transformation, we can fully describe the transformation with a 3×3 matrix. The video uses the idea of keeping the origin fixed and moving the entire grid of points to visualize the effect of the transformation on space itself, with every point acting as a vector.

The 3×3 Matrix Comes from the Basis

Each of the three basis vectors lands at a coordinate triplet after transformation. Recording these coordinates as the columns of a 3×3 matrix yields a complete description of the transformation. In other words, the matrix columns contain the new coordinates of I hat, J hat, and K hat after the transformation, and any input vector with coordinates [X, Y, Z] lands at the linear combination X times the first column plus Y times the second column plus Z times the third column.

Rotation as a Concrete Example

A simple yet powerful example is a rotation of space by 90 degrees around the Y axis. In this case I hat maps to the z axis, J hat remains fixed, and K hat maps to the x axis. The resulting 3×3 matrix has columns that reflect these new coordinates, and applying this matrix to any vector XR, YR, ZR gives its rotated image. This illustrates the parallel between the 2D intuition and the 3D generalization: you still scale and add along basis vectors to get your transformed vector, and the process works before and after the transformation.

Matrix Multiplication and Transformation Composition

Multiplying two 3×3 matrices corresponds to applying the transformation encoded by the right-hand matrix first, then the transformation encoded by the left-hand matrix. This composition principle underpins how complex 3D motions, such as sequences of rotations, can be built from simpler, easier-to-think-about rotations. The video also notes that, in practice, this multiplication follows a familiar pattern and is central to fields like computer graphics and robotics where 3D rotations are described as compositions of simpler moves.

Takeaways and Next Steps

Beyond the specific rotation example, the talk emphasizes that 3D linear transformations are approachable once you focus on basis vectors and the columns of the transformation matrix. The narrative sets the stage for determinants, which will be the topic in the next video. Overall, the takeaway is that linear transformations in higher dimensions retain the core structure seen in two dimensions, with the 3×3 matrix providing a compact, complete description of how space is mapped.