16.4 Contrast and Resolution

The ASNT NDT Level II visual general exam lists Contrast and Resolution as official VT topic 4. Topics 1–3 gave you scope, an eye, and a lamp. Topic 4 is why a discontinuity that is physically present still does not appear. The exam language is short: contrast (luminance and color), resolution versus magnification versus working distance, the idea of a modulation transfer function (MTF) at technician level, and cleanliness / surface condition as contrast killers. A high-magnification image of a dirty toe is the standard trap.

Contrast: luminance and color

A discontinuity is visible only if it differs from its surroundings in a way the eye or camera can use. That difference is contrast.

Luminance contrast

Luminance contrast is a brightness difference. A crack may be a dark line on brighter metal, a bright specular glint from a groove wall, or a shadow in a pit. A simple shop expression is (L_feature − L_background) / L_background (other normalizations exist). What matters operationally is the ratio, not the absolute lux. Two gray tones that differ by 2 percent under 2000 lx are still a 2 percent contrast problem. Turning the lamp up does not automatically create contrast if both the groove and the adjacent mill scale get brighter together.

Color contrast

Color contrast is a difference in hue or saturation at similar brightness: rust red on gray steel, straw heat tint on stainless, a yellow crayon layout mark on primer, a remaining dye stain next to oxide. Color contrast can save a call when luminance contrast is poor, and it can vanish under a low-CRI lamp even when the lux meter passes.

The two are independent:

Do not say "only color contrast matters on steel" or "contrast means magnification." Contrast is why the feature is different. Magnification is how large you draw that difference.

Grazing light raises luminance contrast of texture. Daylight-balanced, high-CRI light raises color contrast of rust and coatings. Cleaning raises both by removing the film that made the feature and the background the same.

Resolution, magnification, and working distance

Resolution is the smallest separation at which two features (or a feature and its background edge) can still be told apart. It is set by the eye's acuity, the optic, the camera pixel pitch, the display, the lighting, and the contrast of the target. A high-contrast black line on white is resolved at a much smaller width than a 5-percent gray groove of the same width.

A useful order-of-magnitude for the eye: high-contrast acuity is on the order of 1 arcminute. At the ASME V direct-viewing maximum of 24 in (610 mm), that angle is roughly 0.007 in (0.18 mm) for a high-contrast target. That is not an ASNT pass/fail number and it is not a promise that a 0.007 in low-contrast undercut will be seen. It is why working distance matters: the same eye at 48 in has half the linear resolution, and you are already outside the common direct-VT distance.

Magnification makes the image larger. A 2× loupe or a borescope zoom can help if it moves a near-threshold feature above the eye's or sensor's resolution and the feature still has contrast. Empty magnification is enlargement past the point where new detail appears — a bigger blur. Zooming a 2 percent gray smear to fill the monitor does not turn it into a crack.

Working distance is the eye-to-surface or lens-to-surface gap.

• Closer (within the procedure's direct-viewing limit) increases angular size, usually improves inverse-square illuminance from a handheld lamp, and can improve resolution — until you block the light or lose the 30° angle.
• Farther shrinks angular size, drops point-source illuminance, and is how direct VT silently becomes an invalid examination past 24 in.
• Remote systems trade working distance inside the part (probe tip to surface) for a monitor viewing distance. Resolution is then the system resolution at the tip, not how close your chair is to the screen.

High magnification collapses depth of field. A steep weld toe has the cap, the fusion line, and the plate in different planes. At high zoom, the undercut can sit just out of focus while a scratch on the cap looks razor sharp. Focus is part of resolution.

MTF at exam level

You do not need to plot a laboratory curve. You need the idea.

A modulation transfer function (MTF) describes how much contrast a system delivers as a function of spatial frequency (how fine the detail is — line pairs per millimetre, cycles per mrad). At low spatial frequency (wide bars), a decent eye, lens, or camera transfers almost all of the target contrast. At high spatial frequency (fine bars), the system blurs bright into dark and the modulation falls toward zero. The frequency where modulation becomes unusable is a practical resolution limit — but only for the contrast of that target.

Consequences the general exam cares about:

1. A remote-VT camera can pass a high-contrast resolution chart (black-and-white line pairs, a period-dot target, an engraved "E") and still miss a low-contrast undercut of similar width. The chart was high modulation at that frequency; the undercut is low modulation. MTF multiplies the contrast you started with. Ten percent of almost nothing is nothing.
2. Magnification shifts a given feature to a lower spatial frequency on the retina or sensor, which can ride up the MTF curve — if there was contrast to transfer. Magnification does not invent modulation.
3. Defocus, motion, dirty optics, compression artifacts, and automatic exposure that crushes highlights all lower MTF. A smeared, auto-gained image of a shiny grind is an MTF failure, not proof the toe is good.
4. The eye has an MTF too. Fatigue, glare, and poor accommodation are human MTF losses.

If a stem says the probe resolved a 0.025 in chart and the inspector still missed a wide, shallow, mill-scale-filled undercut, the answer is low contrast / MTF / cleanliness, not "therefore magnification was too low" and not "the undercut must be subsurface."

Why a high-mag image can miss a low-contrast undercut

Walk this scenario; it is the topic-4 set piece.

A shallow toe undercut is a gentle groove a few thousandths deep and many millimetres long. On mill scale it has almost no luminance step and no color step. The inspector switches a video probe to maximum zoom.

• The field of view collapses to a few millimetres, so the long groove is no longer seen as a line — it looks like the rest of the dark scale.
• Depth of field collapses, so the valley is soft while a nearby scratch is sharp.
• Automatic gain brightens the dark scale and the groove together; the 2 percent difference is still 2 percent.
• The fiber or LED at the tip is now so close that it either fills the groove with light (kills shadow contrast) or produces a specular bloom (glare).
• The surface is still dirty.

Result: a large, impressive, low-contrast image. The undercut is not subsurface. The system did not fail a chart. Contrast was never created. The fix is clean, change the lighting angle, back off the empty zoom, and inspect the toe as a line at a magnification that preserves context. Then gauge depth if the procedure requires a number.

The same logic applies to a 10× loupe on a greasy fillet and to a drone camera on a bridge that auto-exposes a shadowed web to a uniform gray.

Cleanliness and surface condition

Surface condition is how contrast is manufactured or destroyed. It is in scope for VT, not only for PT.

Contrast killers:

• Oil, grease, and couplant films equalize reflectance and fill openings.
• Water and condensation add specular glare and hide tint.
• Slag, spatter, and grinding dust fill undercut and cracks and add false texture.
• Paint, tape residue, and fireproofing hide everything underneath. Inspection through coating is allowed only when the procedure says so.
• Mill scale and tight oxide drop luminance contrast of fine openings.
• Grinding smear can burnish a crack shut so that even PT will struggle; VT of a smeared toe is not a crack examination.
• Heavy corrosion product can hide the opening that caused it or mimic a linear indication.

Contrast helpers the procedure may require: slag removal before weld visual acceptance, solvent cleaning, controlled grinding that does not smear, local coating removal, drying, and a finish comparable to the standard used to set acceptance (you cannot compare a torch-cut edge to a machined comparator and call it fair).

A dirty surface is not made acceptable by raising zoom. Empty magnification of dirt is still dirt. If the stem offers "the camera could focus, so cleanliness does not matter," reject that option.

Putting the four ideas on one stem

A remote inspection of a vessel weld is performed at high zoom after a resolution-chart pass. The cap is oily. Lighting is a coaxial LED at near-normal incidence. The report says no undercut. A later direct visual, after degreasing, with a flashlight at a grazing angle, gauges 1/32 in undercut for 40 mm. Nothing about vision acuity or the 1000 lx ticket was wrong on paper. Color and luminance contrast were never presented to a system whose MTF, at that zoom and lighting, had nothing to transfer.

Topic 4 language is short: name the contrast type, do not confuse it with magnification, invoke MTF as "high-contrast chart ≠ low-contrast groove," and send the inspector back to cleanliness and lighting angle before raising zoom again.