General Optics Principles

Optics is the study of light's behavior as it propagates through space and interacts with matter. The field divides naturally into two regimes based on the relationship between the wavelength of light and the size of the objects it encounters. When optical components and apertures are much larger than the wavelength — typically by a factor of 100 or more — light behaves as though it travels in straight lines called rays, and the mathematics of geometry suffices to predict its path. This regime is geometric optics, and it governs the design of nearly every imaging system, from simple magnifying glasses to multi-element camera lenses and astronomical telescopes [1, 2].

When features approach the wavelength in scale, diffraction and interference become significant, and the wave nature of light must be accounted for. This regime is physical (wave) optics. Both descriptions are limiting cases of electromagnetic theory, and a working optical engineer must know when each applies and where the boundaries lie [3].

This guide establishes the foundational principles that underpin all other optics topics on this site. The laws of reflection and refraction, the concept of optical path length, Fermat's principle, paraxial approximations, stops and pupils, f-number and numerical aperture, étendue, and ray matrix methods — these form the common language of optical system design. Subsequent guides on Lenses, Mirrors, Filters, Coatings, and other topics build directly on the framework presented here.

1.1The Ray Model of Light

The ray model treats light as a collection of straight lines that change direction only at interfaces between materials or at reflecting surfaces. A ray represents the direction of energy flow — more precisely, it is perpendicular to the wavefront in an isotropic medium [1]. The ray model is valid when the following condition holds: the smallest feature in the optical system (aperture diameter, lens clear aperture, surface defect) is much larger than the wavelength of the light being used. A common quantitative criterion is that geometric optics applies when the Fresnel number of the system is large:

Where a is the characteristic aperture radius (m), λ is the wavelength (m), and L is the propagation distance (m).

When $N_F \gg 1$, diffraction effects are negligible and ray tracing accurately predicts system behavior. When $N_F$ approaches unity or falls below it, diffraction dominates and a wave-optics treatment is required [3].

For a 25.4 mm diameter lens used with 632.8 nm light at a 1 m propagation distance, the Fresnel number is approximately 255 — comfortably in the geometric regime. For a 100 μm pinhole at the same wavelength and distance, the Fresnel number drops to about 0.016, placing it firmly in the diffraction regime.

1.2Geometric vs. Physical Optics

The boundary between geometric and physical optics is not a sharp line but a gradual transition. Three regimes are useful to distinguish:

Geometric optics applies when apertures and features are hundreds of wavelengths or more. Ray tracing, Snell's law, and the imaging equation accurately predict system behavior. This covers the vast majority of optical component selection and system layout tasks.

Physical optics applies when features are comparable to or smaller than the wavelength. Diffraction gratings, thin-film coatings, single-mode fibers, and sub-wavelength structures all require wave analysis. Interference and diffraction are the dominant phenomena.

The overlap zone exists in systems where both descriptions contribute. A well-corrected lens may form an image whose quality is limited not by aberrations (a geometric concern) but by the Airy disk diameter (a diffraction phenomenon). Understanding where a system transitions from aberration-limited to diffraction-limited performance is a core skill in optical engineering [2, 5].

The present guide focuses on the geometric regime, establishing the principles that govern ray behavior. The Light Fundamentals guide on this site covers the wave properties of light — interference, coherence, polarization — while specific topics like Coatings and Diffractive Optics address phenomena that require physical optics treatment.

The single most powerful principle in geometric optics is Fermat's principle, from which all ray behavior — reflection, refraction, and image formation — can be derived. First articulated by Pierre de Fermat in 1662, it provides a unified foundation for the entire field [1, 4].

2.1Optical Path Length

The optical path length (OPL) along a ray is the physical distance weighted by the refractive index of each medium the ray traverses:

Where $n_i$ is the refractive index of the i-th medium and $d_i$ is the physical path length through the i-th medium (m).

For a continuously varying medium:

Where the integral is taken along the ray path from point A to point B, and $n(\mathbf{r})$ is the refractive index at position r [1, 6].

The OPL has a direct physical meaning: it is the distance that light would travel in vacuum in the same time it takes to traverse the actual path through material media. Equivalently, multiplying the OPL by $2\pi/\lambda_0$ gives the total accumulated phase along the ray, where $\lambda_0$ is the vacuum wavelength. This phase relationship is the bridge between geometric and wave optics — equal OPLs mean equal phases, and equal phases mean constructive interference at an image point [3].

2.2Fermat's Principle

Fermat's principle states: the path taken by a light ray between two points is stationary with respect to small variations of the path. In most practical cases this means the OPL is a minimum, though it can also be a maximum or saddle point [1, 4].

This variational principle is the geometric-optics counterpart of the Huygens–Fresnel principle in wave optics. It encodes the fact that constructive interference of wavelets selects the ray paths that have stationary phase — paths where neighboring paths have nearly identical OPLs and therefore interfere constructively [3].

From Fermat's principle, three foundational results follow directly:

1. Rectilinear propagation. In a homogeneous medium (uniform n), the shortest path between two points is a straight line. Rays therefore travel in straight lines in any uniform medium.

2. Law of reflection. Requiring the OPL from a source to a reflection point to an observation point to be stationary yields the result that the angle of incidence equals the angle of reflection.

3. Snell's law of refraction. Requiring the OPL across an interface between two media to be stationary yields the relationship between the angles and refractive indices.

These derivations are carried out in Sections 3 and 4. The power of Fermat's principle is that it unifies all three laws under a single statement and generalizes naturally to graded-index media, curved surfaces, and multi-element systems [1, 6].

For imaging systems, Fermat's principle implies a critical condition: a perfect image is formed when all rays from an object point to its corresponding image point have equal optical path lengths. This is the equal-OPL condition for perfect imaging, and it explains why aberrations arise — they represent the failure of different rays to maintain equal OPLs [5].

When light encounters an interface between two media, a portion of the energy is reflected back into the first medium. The geometry of this reflection is governed by a simple and exact law that applies to all reflecting surfaces — flat, curved, smooth, or rough — at the local level [1, 2].

3.1Law of Reflection

The law of reflection states that the angle of incidence $\theta_i$ and the angle of reflection $\theta_r$, both measured from the surface normal at the point of incidence, are equal:

Additionally, the incident ray, the reflected ray, and the surface normal all lie in the same plane, called the plane of incidence [1].

This law follows from Fermat's principle by requiring the OPL from a source to a mirror surface to an observation point to be stationary. The law holds for every type of reflecting surface — the only requirement is that the surface normal is well-defined at the point of incidence [4].

3.2Specular and Diffuse Reflection

Whether a surface produces a clear reflected image (specular reflection) or scatters light in many directions (diffuse reflection) depends on the surface roughness relative to the wavelength. The Rayleigh roughness criterion provides a quantitative boundary [1, 2]:

Where σ is the RMS surface roughness (m), λ is the wavelength (m), and $\theta_i$ is the angle of incidence.

When the surface roughness satisfies this inequality, the surface behaves as specularly reflecting for that wavelength and angle. A polished optical mirror with σ < 2 nm is specular across the entire visible spectrum (400–700 nm). A surface with σ = 100 nm may appear specular in the infrared (10.6 μm) but diffuse in the ultraviolet (250 nm) [1].

This wavelength dependence has practical consequences: surface quality specifications for infrared optics are less demanding than for visible or UV optics, which is one reason infrared optical systems can use materials and manufacturing processes that would be unsuitable for visible-light applications.

3.3Plane Mirror Imaging

A plane mirror produces a virtual image that is located the same distance behind the mirror surface as the object is in front of it. The image is upright (same orientation top-to-bottom), the same size as the object (magnification = 1), but laterally reversed — left and right are swapped relative to the viewing direction [1, 2].

The lateral (parity) reversal produced by a single plane mirror is important in optical system design. Two mirrors at 90° (a corner arrangement) produce a double reflection that reverses the reversal in one axis, while a retroreflector (three mutually perpendicular mirrors) returns light parallel to the incoming direction regardless of incidence angle — a property exploited in laser ranging and traffic safety reflectors [1].

For a complete treatment of laser beam properties, resonator design, and operating regimes, see the Laser Fundamentals guide.

When light crosses the boundary between two transparent media with different refractive indices, the ray changes direction. This bending of light — refraction — is the physical basis for lenses, prisms, and all refractive optical components [1, 2].

4.1Refractive Index and Snell's Law

The refractive index n of a medium is the ratio of the speed of light in vacuum to the speed of light in that medium:

Where c = 2.998 × 10⁸ m/s (speed of light in vacuum) and v is the phase velocity of light in the medium. Since light always travels slower in a material than in vacuum, the refractive index is always greater than 1 for transparent media (with rare exceptions in metamaterials at specific wavelengths) [1].

Snell's law relates the angles of incidence and refraction at a planar interface:

Where $n_1$, $n_2$ are the refractive indices of the incident and refracting media, and $\theta_1$, $\theta_2$ are the angles of incidence and refraction measured from the normal.

When light passes from a lower-index medium to a higher-index medium ($n_2 > n_1$), it bends toward the normal ($\theta_2 < \theta_1$). When passing from higher to lower index, it bends away from the normal. This asymmetry has profound consequences, including the existence of total internal reflection.

4.2Total Internal Reflection

When light travels from a higher-index medium to a lower-index medium ($n_1 > n_2$) and the angle of incidence exceeds a critical value, no refracted ray exists — all light is reflected back into the first medium. This is total internal reflection (TIR) [1, 2].

For angles of incidence $\theta_1 > \theta_c$, the reflectance is 100% regardless of polarization — a property exploited in prisms (Porro, pentaprism, dove), optical fibers, and waveguides [1].

4.3Dispersion

The refractive index of all transparent materials varies with wavelength — a property called dispersion. In normal dispersion (which covers the visible spectrum for most optical glasses), the index decreases with increasing wavelength: blue light bends more than red [1, 2].

Dispersion is quantified by the Abbe number (also called the V-number or constringence):

Where $n_d$ is the refractive index at the helium d-line (587.6 nm), $n_F$ at the hydrogen F-line (486.1 nm), and $n_C$ at the hydrogen C-line (656.3 nm). A high Abbe number indicates low dispersion (crown glasses, $V_d > 50$), while a low Abbe number indicates high dispersion (flint glasses, $V_d < 50$). Optical designers exploit this difference to create achromatic doublets [2, 5].

The chromatic focal shift of a thin lens made from a single glass type is approximately:

For a 100 mm focal length N-BK7 lens, the focal length difference between the F and C lines is approximately −100/64.17 = −1.56 mm — large enough to produce visible color fringing in imaging applications [2, 5].

Exact ray tracing through optical surfaces requires solving transcendental equations at each interface. The paraxial approximation simplifies these calculations dramatically by restricting analysis to rays that make small angles with the optical axis, enabling the use of linear algebra in place of trigonometry [1, 2, 6].

5.1The Paraxial Regime

The paraxial approximation replaces the trigonometric functions with their first-order Taylor expansions:

where θ is in radians. These approximations are accurate to within 1% for angles up to about 10° (0.17 rad) and within 0.1% for angles up to about 3° [1, 6].

Under the paraxial approximation, Snell's law simplifies from a transcendental equation to a linear one:

This linearity is what makes paraxial optics so powerful: it means that all paraxial imaging relationships — focal length, magnification, image position — can be expressed as linear functions of ray height and angle. Matrix methods (Section 9) exploit this linearity directly [6, 8].

The limits of the paraxial approximation are explored in detail by the Small-Angle Explorer tool on this site, which computes the error between paraxial and exact trigonometric results for any specified angle. As a practical guideline: first-order (paraxial) design establishes the layout and basic performance of an optical system, while real ray tracing and aberration analysis then refine the design for actual (non-paraxial) rays [5].

5.2Sign Conventions

Consistent sign conventions are essential for multi-surface and multi-element optical calculations. The most widely used convention in modern optics is the Cartesian sign convention [1, 2, 6]:

1. Light travels from left to right. The positive direction of the optical axis is left-to-right.

2. Distances are measured from the reference surface (vertex, principal plane, or origin). Distances in the direction of light propagation (to the right) are positive; against propagation (to the left) are negative.

3. Heights above the optical axis are positive; below are negative.

4. Angles are measured from the optical axis or surface normal to the ray. Positive angles correspond to counterclockwise rotation from the reference direction.

5. Radii of curvature are positive if the center of curvature is to the right of the surface, negative if to the left.

Under this convention, a converging lens has a positive focal length and a diverging lens has a negative focal length. A real image (formed on the transmission side) has a positive image distance, while a virtual image (formed on the same side as the object) has a negative image distance [1, 2].

The Cartesian convention is not the only convention in use — older texts use the “real-is-positive” convention, and mirror equations sometimes reverse the sign of the image distance. Mixing conventions within a calculation is a common source of error, particularly in multi-element systems. The key is to choose one convention and apply it consistently through every surface in the system [6].

The formation of images — the mapping of points in object space to corresponding points in image space — is the central task of geometric optics. Understanding conjugate relationships, image types, and magnification provides the foundation for all optical system design [1, 2].

6.1Conjugate Points and Image Types

Two points are said to be conjugate if every ray leaving one point that enters the optical system passes through (or appears to pass through) the other point. The object point and its corresponding image point form a conjugate pair [1, 6].

Real images are formed where rays physically converge. A screen placed at a real image location displays the image. Real images are formed by converging optical systems when the object is beyond the focal point.

Virtual images are formed where rays appear to diverge from, but do not actually pass through. A virtual image cannot be captured on a screen, but it can be observed by an eye or another optical system that intercepts the diverging rays. The image in a plane mirror and the magnified image seen through a simple magnifier are both virtual [1, 2].

6.2Magnification

Three types of magnification characterize optical systems:

Lateral (transverse) magnification is the ratio of image height to object height:

The negative sign in the distance ratio (under Cartesian convention) means that when both object and image distances are positive (real object, real image), the image is inverted ($m < 0$). A magnification of $m = -2$ means the image is inverted and twice the height of the object [1, 2].

Longitudinal (axial) magnification describes how depth in object space maps to depth in image space:

(In air, where $n = n' = 1$.) The negative sign means that depth is always reversed, and the quadratic dependence on lateral magnification means longitudinal magnification is always greater than or equal to lateral magnification, causing 3D images to appear stretched along the optical axis [1, 6].

Angular magnification is relevant for visual instruments (telescopes, microscopes):

For a simple magnifier with focal length f, the angular magnification when the image is at infinity (relaxed viewing) is:

where 250 mm is the conventional near point distance of the human eye [1, 2].

6.3The Imaging Equation

For a thin lens or spherical mirror in the paraxial regime, the relationship between object distance s, image distance s', and focal length f is given by the Gaussian imaging equation:

Note: Under the Cartesian sign convention used throughout this guide, the object distance s for a real object to the left of a lens is negative, so the equation takes this form with a minus sign. Some texts write $1/s' + 1/s = 1/f$ with a convention that makes s positive for real objects — both forms give identical results when their respective sign conventions are applied consistently [1, 2].

The equivalent Newtonian form measures distances from the focal points rather than from the lens:

Where x is the object distance from the front focal point F, x' is the image distance from the rear focal point F', and f is the focal length. The Newtonian form is particularly convenient for calculating magnification: $m = -f/x = -x'/f$ [1].

Every practical optical system is limited in two fundamental ways: how much light it can collect (determined by the aperture stop) and how large an object it can image (determined by the field stop). Understanding these limitations and their conjugate representations — the entrance and exit pupils — is essential for system design, alignment, and performance prediction [1, 2, 5].

7.1Aperture Stop and Pupils

The aperture stop is the physical element in an optical system that limits the cone of rays from an axial object point that can pass through the system. It may be a dedicated iris, the rim of a lens, or any other physical aperture [1, 5].

The entrance pupil (EP) is the image of the aperture stop as seen from object space. The exit pupil (XP) is the image of the aperture stop as seen from image space. The entrance and exit pupils are conjugate to each other and to the aperture stop [1, 2, 6].

Two special rays define the geometry of any optical system: the marginal ray originates from the axial object point and passes through the edge of the aperture stop, defining the maximum cone angle and the system's light-gathering capability. The chief ray (also called the principal ray) originates from the edge of the object field and passes through the center of the aperture stop, defining the field of view and image height [5, 6].

7.2Field Stop and Field of View

The field stop is the aperture that limits the extent of the object that can be imaged. In a camera, the detector serves as the field stop. In a visual instrument, a physical ring called a field diaphragm defines the circular boundary of the visible field [1, 5].

Where h is the half-height of the field stop (or half-diagonal of the detector) and f is the effective focal length. For small angles, this simplifies to FOV ≈ 2h/f in radians. A 50 mm focal length lens with a 36 × 24 mm sensor (half-diagonal = 21.6 mm) has a diagonal FOV of 2 arctan(21.6/50) = 46.8° [5].

7.3Vignetting

Vignetting is the progressive reduction in illumination at the edges of an image caused by off-axis ray bundles being partially blocked by apertures other than the aperture stop. Natural vignetting (the cosine-fourth law) occurs even in a perfect system [1, 5]:

At a 30° field angle, the illuminance drops to cos⁴(30°) = 0.5625, or about 56% of the on-axis value. Mechanical vignetting compounds this further, which is why many wide-angle lens designs deliberately control vignetting as a design variable [5].

The light-gathering capability of an optical system — how much optical power it delivers to the image plane per unit area — is characterized by three related quantities: f-number, numerical aperture, and étendue. These quantities are fundamental to exposure calculations, resolution limits, and the coupling of light between components [1, 2, 5].

8.1F-Number

Where f is the effective focal length and $D_{\text{EP}}$ is the diameter of the entrance pupil. A lens with a 50 mm focal length and a 25 mm entrance pupil has an f-number of f/2. A “fast” lens (f/1.4, f/2) collects more light than a “slow” lens (f/8, f/16) [1, 5].

When a lens is used at finite conjugates, the working f-number accounts for the change in effective light-gathering:

Where m is the lateral magnification (negative for inverted images under Cartesian convention). At 1:1 magnification ($m = -1$), the working f-number is twice the infinity f-number, meaning the effective exposure is reduced by a factor of 4. This is the “bellows factor” familiar to macro photographers [5].

8.2Numerical Aperture

Where n is the refractive index of the medium and $\theta_{\max}$ is the half-angle of the maximum cone of rays. For systems in air, the relationship between NA and f-number at infinite conjugate is:

This approximation is accurate for f/# > 2. The exact relationship is:

NA is the preferred specification for microscope objectives and fiber optics because it remains meaningful even when the focal length is not well-defined or when the medium is not air. A microscope objective with NA = 1.25 requires oil immersion ($n \approx 1.515$) because NA > 1 is impossible in air [1, 2].

For optical fibers, the NA characterizes the acceptance cone:

8.3Étendue and the Optical Invariant

Étendue (also called throughput or the AΩ product) is a conserved quantity that characterizes the light-gathering capability of an optical system:

Étendue is conserved as light propagates through a lossless optical system — it cannot be decreased by any combination of lenses, mirrors, or other passive optical components. This conservation law is a direct consequence of the second law of thermodynamics and is also implied by the Lagrange invariant in Gaussian optics [1, 3, 6].

The practical consequence is fundamental: when coupling light from a source into an optical system (for example, coupling an LED into a fiber), the system can accept only light within its étendue. If the source étendue exceeds the system étendue, the excess light is lost regardless of the optical design. Matching étendue between source and system is a critical early step in any illumination or fiber-coupling design [3, 5].

The closely related Lagrange invariant (or optical invariant) expresses the same conservation law in the paraxial ray-trace framework:

Where y, u are the marginal ray height and angle, and y', u' are the chief ray height and angle. The value of H is the same at every surface in the system [6].

The linearity of paraxial optics permits ray propagation to be expressed as matrix multiplication. Each optical element is represented by a 2×2 matrix that transforms the ray state (height and angle). Cascading elements is as simple as multiplying matrices in sequence. This formalism is the computational backbone of first-order optical design [6, 8].

9.1The Ray Transfer Matrix

A paraxial ray is described by a column vector containing the ray height y and ray angle u (in radians):

An optical element transforms the input ray through multiplication by a 2×2 ray transfer matrix (ABCD matrix):

For systems in uniform media, the determinant of the ray transfer matrix equals 1:

9.2System Matrices

To find the ray transfer matrix for a compound system, multiply the individual element matrices in reverse order of encounter (right-to-left):

where M₁ is the first element the ray encounters and M_N is the last [6, 8].

9.3Cardinal Points from the System Matrix

The cardinal points of a complete optical system can all be extracted from the system matrix [6, 8]:

When the input and output media are different ($n_1 \neq n_2$), the nodal points do not coincide with the principal planes. For a system in uniform media (air throughout), the nodal points coincide with the principal planes [6].

These relationships make the matrix method a powerful design tool: assemble the system matrix by multiplying individual element matrices, then immediately extract all first-order imaging properties without tracing individual rays.

Geometric optics provides the framework for designing optical systems, but every physical system eventually encounters phenomena that rays alone cannot explain. Recognizing the boundary between geometric and wave-optical behavior is a critical skill in optical engineering [1, 2, 3].

10.1The Diffraction Limit

Even a perfect optical system — one with no aberrations whatsoever — cannot focus light to an infinitely small point. The wave nature of light causes the image of a point source to spread into an Airy pattern [1, 3].

The corresponding linear radius at the focal plane is:

For a lens operating at f/4 with 550 nm light, the Airy disk radius is 1.22 × 0.55 μm × 4 = 2.68 μm. This is the fundamental resolution limit — no amount of improvement in lens quality can produce a smaller spot for a given wavelength and f-number [1, 3].

A system is said to be diffraction-limited when its geometric aberrations are small enough that the image quality is dominated by the Airy pattern. The Maréchal criterion gives a quantitative threshold: a system is effectively diffraction-limited when the RMS wavefront error is less than λ/14 [5].

10.2Choosing the Right Model

Use geometric optics (ray tracing) when all apertures are much larger than the wavelength, the goal is first-order layout (focal lengths, image positions, magnification, f-number), aberration analysis is needed, or illumination calculations are required.

Use physical optics (wave analysis) when feature sizes approach the wavelength, coherence and interference effects are important, diffraction-limited performance must be verified, or propagation through single-mode fibers or waveguides is involved.

Use both (hybrid) when evaluating whether a system is aberration-limited or diffraction-limited, designing systems with both refractive and diffractive elements, coupling laser beams into fibers, or performing tolerance analysis on precision optical systems.

In practice, most optical system design begins with geometric (paraxial) layout, proceeds through real ray tracing and aberration analysis, and concludes with a diffraction-based evaluation of image quality (PSF, MTF, Strehl ratio). The geometric foundation established in this guide is the starting point for that entire workflow [5, 7].

10.3Connections to Other Topics

The principles established in this guide appear throughout the Abridged Optics content:

Lenses build directly on Sections 4–6 (Snell's law, imaging equation, magnification) and Section 9 (matrix methods for multi-element systems). The Lenses guide extends these foundations to thick lenses, compound systems, and Sellmeier dispersion models.

Mirrors apply the law of reflection (Section 3) and the imaging equation (Section 6) with appropriate sign convention modifications for reflecting surfaces.

Filters and Coatings operate in the physical optics regime (thin-film interference) but their integration into optical systems is analyzed using the geometric tools from Sections 7–8 (stops, pupils, f-number, étendue).

The principles of kinematic constraint are applied directly in optic mount design — see Optic Mounts §3: Kinematic Principles.

Fiber Optics relies on total internal reflection (Section 4.2) for ray-optic analysis of multimode fibers and transitions to wave optics for single-mode fiber analysis.

Light Fundamentals covers the wave properties of light that complement the geometric treatment in this guide — interference, coherence, and polarization.

Polarization & Polarizers extends the Fresnel equations introduced here into a full treatment of polarization states, Jones and Mueller calculus, waveplates, and polarizer selection.