Imaging through water: diagnose blur at high speeds

A submerged object can look sharp through an empty clear tank and blurred immediately after filling it. That change does not establish that you need an immersion lens. Adding water changes the optical path, but filling can also introduce bubbles, reflections, vibration and moving refractive boundaries.

For high-speed inspection, separate stationary optical quality from motion blur before changing hardware. The tests below help you identify the limiting mechanism, calculate an exposure budget and decide whether refocusing, a better viewing window or revised illumination is the appropriate next step.

1. Separate optical blur from motion blur

Start with a fixed, high-contrast target at the intended object position. Use features near the center and corners, rather than judging one central edge. Keep the camera rigidly mounted and save unprocessed images without sharpening or automatic contrast adjustments.

Record exposure time, aperture, gain, working distance, object depth behind the window and lighting direction. Disable automatic exposure and gain during the comparison. Otherwise, filling the tank may change brightness and cause the camera to lengthen exposure without making that change obvious.

Run these tests in order:

Do not remove and replace the target between tests unless its position can be repeated accurately. A different target distance would confound the comparison.

If the dry reference already contains localized defects, use the controlled comparisons in camera lens scratch detection before blaming the water. A useful diagnosis requires a known-good baseline.

2. Refocus through the filled tank before replacing the lens

A flat water boundary changes the apparent position of an underwater object viewed from air. Near perpendicular incidence, with small viewing angles, apparent depth can be estimated using the refractive-index ratio.

This example describes the water portion of the optical path. It excludes the plastic wall contribution and is not a prescription to move the camera by that amount. Actual focus depends on the complete geometry and lens design.

Refocus with the tank filled, the target stationary and the water settled. Capture a small focus sweep, perhaps 5–7 settings around the apparent optimum. Compare the same center and corner features at every setting.

Interpret the sweep as follows:

• All regions become sharp together: a focus shift was probably the main limitation.
• The center sharpens but corners remain poor: investigate viewing angle, wall shape, alignment and refraction-induced aberrations.
• Different regions require different focus settings: check target tilt, wall curvature and field-dependent optical errors.
• No setting restores contrast: inspect scattering, reflections, wall quality and vibration before buying another lens.

A curved tank wall can act as an additional optical element. A flat, parallel viewing window with the camera axis approximately perpendicular to it is usually easier to evaluate. Wall bowing under the filled condition can also change the result.

Stopping down may increase usable depth of field, but it reduces collected light and eventually increases diffraction blur. The lens finder helps narrow aperture and coverage requirements after the filled-path tests establish what the optics must accomplish.

3. Calculate exposure from motion, not frame rate

If the stationary submerged target is sharp but the moving object is not, calculate the permitted exposure. Frame rate determines how often images arrive; exposure determines how far the object travels while each image integrates light.

Use local object-space sampling at the inspection plane. Refraction and wall geometry can change magnification, so do not assume the dry calibration remains valid. The guide to object-space pixel size explains the sampling relationship.

For a general example, suppose the filled-path image covers 80 mm across 1600 pixels:

At an object speed of 1000 mm/s and an allowed blur of 1 pixel:

These are application examples, not specifications for any listed component. One pixel is a starting criterion, not a universal acceptance limit. Fine edge measurement may need less blur; a coarse presence check may tolerate more.

Confirm the required pixel count against the smallest useful feature, then verify exposure against actual speed. For variable-speed motion, use the maximum speed during the measurement interval. For rotation, use the tangential speed at the feature being inspected.

A short exposure does not remove every rolling-shutter effect because different rows may be exposed at different times. The explanation of exposure time versus shutter speed helps separate integration time from sensor readout behavior.

4. Identify water movement, bubbles and reflected light

Clear, homogeneous water does not become an image-blurring mechanism simply because it is flowing. Look instead for changes associated with that flow: entrained bubbles, temperature gradients, surface waves in the viewing path or mechanical vibration.

With a stationary submerged target, record a sequence before circulation starts, during circulation and after it stops. Keep exposure and focus unchanged. Compare individual frames rather than relying only on a live display.

A dry reference fixed to the outside of the tank can help separate camera vibration from changes within the water path. If both references move together, investigate mechanical movement. If only the submerged target distorts, investigate the intervening optical path.

For reflections, darken the space around the lens and move the light away from the direct reflection geometry. A temporary matte shield outside the tank can reveal whether room lighting is entering the image through the wall.

Polarization can reduce some dielectric reflections, but it also loses light and may reveal stress patterns in plastic. Evaluate it experimentally rather than assuming it will improve every clear-wall arrangement.

5. Restore useful light without changing the experiment

Shortening exposure often makes the image darker, but increasing gain does not replace missing photons. First improve illumination delivery and suppress unwanted reflections. Keep the target, focus and aperture fixed while changing one lighting variable at a time.

Backlighting can provide strong silhouette contrast when the task concerns outlines, holes or presence. Surface markings usually need front or oblique illumination. In either case, check whether the light must pass through another tank wall or a longer water path.

If you use a strobe, the useful optical pulse must overlap the relevant sensor exposure interval. The electrical command width alone does not establish the amount of usable light. The guide to LED strobe optical rise time explains why pulse startup matters.

For an illustrative pulse lasting 40 µs at 500 pulses/s:

That arithmetic describes timing only. It does not establish a safe operating point or acceptable peak current. Use the light-controller finder to identify candidates, then verify their documented electrical and timing compatibility with the chosen light.

6. Choose evaluation hardware around the diagnosed cause

Choose components only after the stationary and moving tests identify the limitation. More camera pixels cannot restore detail already lost through the window, and a different lens cannot freeze motion during a long exposure.

For machine vision lenses, check sensor coverage, mount, working distance, aperture range and performance through the filled optical path. The LS-FFL-0567 is a 16 mm, C-mount lens with a 1.1" sensor format. It is an evaluation candidate, not evidence that a conventional lens will correct every window geometry.

For area scan cameras, prioritize shutter behavior, sufficient sampling and documented timing capabilities. The MV-CH120-11UC is a global-shutter USB3 camera. Confirm its frame rate, exposure limits and trigger behavior for the required acquisition mode before selecting it through the camera finder.

For light controllers, verify output type, load compatibility and timing rather than selecting by channel count alone. The LS-CTL-0070 has digital control and a strobe feature. Its suitability still depends on the light and the required pulse behavior.

These components are separate evaluation candidates, not a validated system combination. Do not infer immersion suitability from a lens mount, camera interface or controller feature. Any component exposed directly to water needs explicit manufacturer documentation for that use.

7. Verify the filled system under production conditions

Judge success against the inspection task, not against whether the picture appears attractive. Use representative edges, markings or defects at the actual object depths and positions.

Before ordering or approving the installation, complete this checklist:

• Repeat the filled-path focus test. Check the center, corners and full expected object-depth range.
• Measure local scale. Place a suitable reference at the inspection plane and verify that dry calibration has not been reused accidentally.
• Test maximum motion. Confirm the smallest required feature remains usable at the highest relevant speed.
• Check timing consistency. Inspect a sequence for partially illuminated frames, missing pulses or alternating brightness.
• Repeat after warm-up. Compare focus, intensity and measured dimensions after operating conditions stabilize.
• Test realistic water conditions. Include expected circulation, bubbles and changes in clarity rather than only freshly settled water.
• Record an acceptance limit. Use edge width, contrast or measurement repeatability appropriate to the task.

As an example, compare 100 consecutive frames at each test condition and record the worst observed edge width, not just the average. This is a diagnostic sample, not a statistical production guarantee.

Use the procedures in machine vision lighting stability testing when brightness changes remain after the optical path is stable.

If the camera or light also faces washdown, choose protection for the actual cleaning exposure; the distinction between IP65 and IP66 water-jet tests matters there. Those ratings do not certify optical flatness or image quality.

8. Know when refocusing is not enough

An immersion objective is not a generic replacement for a lens looking through a tank from air. Such optics are designed for particular media and geometries, often with constrained working distances and fields. Putting an ordinary lens into water is not an acceptable substitute.

Seek specialist optical evaluation when a stationary target remains unacceptable after refocusing through clear, settled water and a suitable flat window. Large viewing angles, demanding corner resolution, thick windows or wide depth ranges may require a different optical arrangement rather than another focus adjustment.

Likewise, use a shorter exposure when blur follows object speed, and improve water conditions when image distortion follows circulation. These are different remedies for different mechanisms.

For the next step, prepare dry and filled reference frames, the window geometry, object depth, required field of view, smallest feature and maximum speed. Start with the lens configuration step when coverage and focus are the limitation, or send that test package through application support when the complete refracting path needs evaluation.