Start with the physical frequency table
NBS 1952 identifies a pattern family, not one universal frequency range. Scaled glass versions and historical photographic charts can use different label conventions. Confirm whether a label refers to the physical target plane or to an image made at a specified reduction.
Keep frequency, pitch and line width separate
For a physical frequency f in lp/mm, the period is 1/f mm. With equal lines and gaps, each is 1/(2f) mm wide. A line pair includes one dark line and one bright gap. Width alone is insufficient if the duty cycle is not 50%.
| Frequency (lp/mm) | Period (µm) | Line width (µm) |
|---|---|---|
| 2.8 | 357.14 | 178.57 |
| 12 | 83.33 | 41.67 |
| 24 | 41.67 | 20.83 |
| 48 | 20.83 | 10.42 |
| 80 | 12.50 | 6.25 |
Convert between target and image planes
For magnification magnitude |M|, image frequency is f_image = f_target / |M|. Measure magnification from a known target length and its image length. Always state which plane the reported lp/mm refers to.
Understand the historical 26f arrangement
The historical photographic-lens method used approximately 25:1 reduction. In a thin-lens model, an object distance of 26f from the front principal plane corresponds to an image distance of 26f/25 and a magnification magnitude of 1/25. It is a test geometry, not a compulsory working distance for every modern NBS 1952 glass target.
Separate real detail from false resolution
A group is resolved only when its three lines and gaps are distinguishable under the chosen criterion. Defocus can produce contrast reversal, and a sampled image can show aliasing. Both can create apparent lines beyond the reliable boundary. Count the bars and follow the frequency progression rather than accepting an isolated finer group that appears to re-emerge.
Visual resolution, CTF and MTF are different outputs
A visual reading provides a limiting frequency under a stated criterion. A bar-profile measurement can estimate square-wave contrast transfer when input contrast, illumination and detector linearity are accounted for. Finite three-bar groups also have boundary effects, so the scan region and analysis method matter. Neither operation automatically produces a sinusoidal MTF curve.
Square waves contain higher harmonics. Only under appropriate linear-system assumptions and when higher-harmonic contributions are negligible does the familiar CTF ≈ (4/π) × MTF approximation apply. It is not a universal conversion for processed camera images. Use a defined slanted-edge or other validated method when an MTF/SFR result is required.
Sampling and field position
For pixel pitch p in millimetres, the one-dimensional sensor Nyquist frequency is 1/(2p) lp/mm. Near that boundary, phase and orientation can strongly change the appearance of bars. Test the centre, edges and corners separately; record line orientation explicitly so directional results remain comparable.
Control the target and optical path
Flatness and mounting tilt affect whether the full pattern lies in one focus plane. Substrate thickness is particularly relevant when the imaging rays pass through the glass; viewing the pattern from the film side is a different arrangement. Film optical density is wavelength dependent and does not by itself specify the final image contrast.
Uniform temperature changes scale the glass and its pattern together. Temperature gradients and differential expansion between the glass and its holder can instead introduce bending or mounting stress. Allow the assembly to stabilise before comparing small changes in focus or field performance.
Optical density, reflection and pattern polarity
Optical density describes transmitted attenuation: OD = −log₁₀(T), where T is the transmitted fraction under the stated measurement conditions. OD 3 corresponds to T = 0.001 at the specified wavelength. The image contrast also depends on clear-region transmission, stray light, reflections, exposure and the detector response.
Surface reflectance is a separate property. A film that blocks transmission may still reflect substantial incident light. In reflection imaging, the illumination angle, collection angle and coating stack determine the useful contrast; an OD value alone cannot select the coating.
Both opaque bars on clear glass and clear bars in an opaque field can be used in transmission. Their illuminated areas and sensitivity to stray light differ. Specify the clear and opaque regions explicitly, because positive and negative naming conventions can vary. Lowering exposure does not turn a high-contrast target into a calibrated low-contrast target.
Investigate an inconsistent reading
Change one condition at a time and retain the original captures. Use the checks below to narrow the cause before comparing components or adjusting an acceptance limit.
| Observed result | What to check | How to interpret it |
|---|---|---|
| One side of the image is softer | Compare focus through a small axial sweep, check target tilt and repeat with the target rotated in its plane. | Separates a possible setup effect from a repeatable field-dependent result; rotation alone does not identify the cause. |
| Horizontal and vertical readings differ | Keep field position and frequency matched; compare stationary captures with shorter exposures and stable illumination. | Directional motion, sampling and optical aberrations can all contribute. |
| Contrast changes with lighting | Inspect the bright and dark levels before processing; check clipping, reflections, stray light and illumination uniformity. | An exposure change can alter a reading without any change to the target geometry. |
| Results drift after warm-up | Repeat focus and magnification checks after thermal stabilisation; inspect fixture stress and uneven heating. | Thermal expansion and mounting distortion affect different parts of the measurement. |
| Only one group reads incorrectly | Inspect that region for particles, scratches or film defects, then compare an equivalent group where available. | A local defect should not be interpreted as the resolution limit of the entire imaging system. |
Record the reading and its uncertainty
If one group resolves and the next finer group does not, record both frequencies and the observation criterion. This brackets the transition between two discrete test points. Reporting additional decimal places does not create information between those groups.
Measured dimensions can improve the frequency estimate, but their uncertainty remains part of the result. A calibration certificate does not remove target uncertainty. Consider dimensional uncertainty, magnification uncertainty, repeatability and the finite frequency steps when stating the scope of the measurement.
- Target: identifier, artwork scale, frequency table, polarity and available dimensional measurements.
- Setup: magnification, field position, bar orientation, aperture, wavelength and viewing side.
- Capture: exposure, gain, processing settings and the original image files.
- Result: last resolved and first unresolved groups for each orientation, repeat readings and the frequency reference plane.