On this page
Introduction: Why Optical Performance Is More Than Just Resolution
Two lenses can carry identical resolution numbers on their datasheets and still produce visibly different images. One renders fine texture with crisp, confident contrast; the other turns the same detail into a gray mush that no amount of sharpening fully recovers. The difference is not resolution — it is how each system transfers detail from object to image.
The tool that captures this behavior is the Optical Transfer Function (OTF). Think of an optical system as a spatial filter: coarse features pass through nearly untouched, while progressively finer features are transmitted with less and less contrast until they vanish entirely. The OTF describes exactly how much of each level of detail survives the trip through your optics, which is why it is one of the most important tools for evaluating real-world optical performance.
The Fundamentals: Resolution and Contrast
Two concepts underpin everything that follows.
Resolution is an optical system's ability to distinguish fine detail, conventionally expressed as spatial frequency in line pairs per millimeter (lp/mm) — how many adjacent black-and-white line pairs can be squeezed into a millimeter of image before they blur together.
Contrast (or modulation) describes how well the system separates light from dark:
Contrast = (I_max − I_min) / (I_max + I_min)
where I_max and I_min are the intensities of the bright and dark regions. A perfect black-to-white pattern has a contrast of 1.0; a uniform gray has 0.
The two are inseparable. Diffraction and aberrations progressively erode contrast as spatial frequency rises — fine detail doesn't disappear abruptly, it fades. Any honest statement of resolution must therefore say at what contrast the detail is resolved, and that is precisely what the OTF formalizes.
What Is the Optical Transfer Function?
The OTF describes how an optical system transfers both contrast (amplitude) and phase (spatial position) from object to image, as a function of spatial frequency. For spatially incoherent imaging under a linear, shift-invariant approximation, it is the normalized Fourier transform of the system's Point Spread Function (PSF) — the image the system forms of a perfect point source.
That Fourier relationship is what makes the OTF so powerful: within a region where the system is approximately shift-invariant, the point response predicts the image of an extended object. Systems whose response varies across the field require separate evaluation at different field points.
The Two Components: MTF and PhTF
The OTF is a complex function, and its two parts carry distinct physical meaning.
Modulation Transfer Function (MTF) — the magnitude of the OTF — quantifies contrast transfer:
MTF(u) = Image Contrast(u) / Object Contrast(u)
at each spatial frequency u. An MTF of 1.0 means contrast is transferred perfectly; 0 means the detail at that frequency is completely washed out. MTF is the metric you will encounter most often, because contrast loss is what the eye (and most algorithms) notice first.
Phase Transfer Function (PhTF) — the phase of the OTF — describes spatial shifts of image features. A non-zero PhTF means certain frequencies land in slightly wrong positions, which distorts edges and asymmetric features (the classic signature of coma). In precision applications — metrology, star trackers, machine-vision gauging — phase errors matter as much as contrast, because feature position is the measurement.
How OTF and MTF Are Measured
Three measurement approaches dominate practice:
- Point Spread Function imaging — image a sub-resolution pinhole or star; the Fourier transform of the recorded PSF gives the full OTF directly.
- Line Spread Function (LSF) analysis — image a narrow slit; the transform of the LSF yields the OTF along one direction.
- Slanted-edge method — image a straight edge tilted a few degrees to the pixel grid. The edge is super-sampled across many rows, differentiated into an LSF, and transformed into a continuous MTF curve. This is the workhorse of camera-system testing (standardized in ISO 12233) because it needs only a printed target and delivers sub-pixel-resolution MTF data.
Reading an MTF Chart
An MTF chart plots contrast transfer (0–100%) on the vertical axis against spatial frequency (lp/mm) on the horizontal axis. To read one like an optical engineer:
- The diffraction limit is the theoretical ceiling — no lens at a given f-number and wavelength can exceed it. How closely the measured curves hug this ceiling tells you how well-corrected the design is.
- Multiple field positions (center, mid-field, corner) appear as separate curves. A lens that is superb on axis and collapses at the corners will show it here immediately.
- Sagittal and tangential curves (solid and dashed, by convention) describe performance for detail oriented in perpendicular directions. A separation indicates direction-dependent contrast transfer; astigmatism is one possible cause, but it should be checked alongside other aberrations.
- Frequency of interest matters more than any single number: a lens for a 3.45 µm-pixel sensor needs contrast out to ~145 lp/mm (the sensor's Nyquist frequency), while a projection system for human viewing may only care about 10–30 lp/mm.
Why System-Level MTF Is What Actually Matters
The image your customer sees is filtered by every element in the chain, and, for cascaded linear, shift-invariant stages at consistent spatial-frequency scales, MTF composes multiplicatively:
MTF_system = MTF_lens × MTF_sensor × MTF_electronics × MTF_display …
This has a hard engineering consequence: the system is weaker than its weakest link, and over-speccing one component cannot rescue another. A 0.9-MTF lens on a 0.5-MTF sensor delivers 0.45 — barely better than a mediocre lens would have. Balanced budgeting across lens, sensor sampling, motion blur and processing is how experienced teams hit image-quality targets without paying for performance the chain throws away.
This is also why MyntOptics designs to a system MTF budget from day one — allocating contrast between optics, detector and margin, then verifying each allocation with analysis before hardware exists.
Try It Yourself
You can explore every concept in this article interactively with our free, browser-based Diffraction MTF Calculator — physical-optics MTF from pupil autocorrelation, with defocus, spherical aberration, astigmatism, central obscuration and detector Nyquist overlays. No installation; email signup is required to use the calculator.
Key Takeaways
- Resolution numbers alone are meaningless without contrast; the OTF/MTF framework unifies both.
- MTF measures contrast transfer; PhTF measures positional fidelity — precision systems need both.
- Read MTF charts against the diffraction limit, across field, in both azimuths, at your frequency of interest.
- Use the multiplicative MTF model where its linear, shift-invariant assumptions apply; account separately for sampling, aliasing and nonlinear processing.
Specifying or verifying image quality for a real product? Talk to our engineers — we can discuss the scope of an MTF budget, lens-selection study or design review.
Working on a similar optical question? Discuss your project with us. We can agree the scope, inputs and next steps together.