Projective Geometry: Vanishing Points & Perspective Transforms

simulator advanced ~10 min
Loading simulation...
2-point perspective with 60° field of view

A 60° field of view with slight tilt and pan produces a natural two-point perspective. Parallel lines in 3D converge to vanishing points on the canvas, demonstrating the core principle of projective geometry.

Formula

Cross-ratio: (AC·BD)/(BC·AD) — invariant under projection
Perspective projection: x' = f·X/Z, y' = f·Y/Z
Homogeneous coordinates: (x,y) → [x : y : 1]

The Geometry of Sight

When you look down a long straight road, the parallel edges appear to converge toward a point on the horizon. This is not an optical illusion — it is the fundamental operation of projective geometry. Renaissance artists discovered perspective drawing by formalizing this convergence, and mathematicians later built an entire geometry around it where parallel lines always meet — at infinity.

Invariants Under Projection

Projective geometry strips away distances and angles, keeping only what survives projection: collinearity (points on a line remain on a line) and the cross-ratio of four collinear points. This single numerical invariant, (AC·BD)/(BC·AD), remains unchanged no matter how severely you project or distort the figure. It is the cornerstone of projective measurement.

From Art to Computer Vision

Projective geometry is the mathematical foundation of computer vision. Every camera performs a projective transformation from 3D world to 2D image. Homography matrices describe how planar surfaces transform between views. Structure-from-motion algorithms recover 3D geometry from multiple 2D images using projective invariants — the same mathematics that Brunelleschi used to paint convincing depth in 1420.

Exploring Perspective

This simulation lets you manipulate field of view, tilt, and pan to see how a 3D grid projects onto a 2D canvas. Watch parallel grid lines converge to vanishing points. Increase the field of view to create dramatic wide-angle distortion. The cross-ratio readout confirms that this projective invariant remains constant as you transform the view.

FAQ

What is projective geometry?

Projective geometry studies properties that are invariant under projection — stretching, skewing, and perspective transformation. Unlike Euclidean geometry, it has no concept of distance or angle, but it preserves collinearity (points on a line stay on a line) and the cross-ratio of four collinear points.

What is a vanishing point?

A vanishing point is where parallel lines appear to converge when projected onto a 2D plane. In one-point perspective, all depth lines converge to a single point. Two-point and three-point perspectives have multiple vanishing points for different sets of parallel lines.

What is the cross-ratio and why is it important?

The cross-ratio of four collinear points A, B, C, D is (AC·BD)/(BC·AD). It is the only numerical quantity preserved by projective transformations. It enables camera calibration, 3D reconstruction from images, and is fundamental to algebraic geometry.

How does projective geometry relate to computer vision?

Computer vision uses projective geometry extensively. Camera projection is a projective transformation. Homography estimation, stereo vision, augmented reality, and panorama stitching all rely on projective geometry to relate 3D scenes to 2D images.

Sources

Embed

<iframe src="https://homo-deus.com/lab/geometry/projective-geometry/embed" width="100%" height="400" frameborder="0"></iframe>
View source on GitHub