High resolution machine vision cameras: how many pixels?

More pixels help only when the inspection needs finer spatial sampling and the optics preserve that detail. Otherwise, a larger image mainly increases transfer load, processing time and storage. Buyers comparing high resolution machine vision cameras should start with the smallest relevant feature, not the largest megapixel figure.

This guide turns field of view and feature size into a pixel requirement, then checks the lens, motion and acquisition rate. You will be able to distinguish a useful resolution upgrade from one that merely produces larger files.

Decide what the extra pixels must improve

Define the inspection decision before selecting a camera. Detecting a dark mark, separating adjacent edges, reading a character and measuring a diameter impose different requirements. A feature visible in one image may still be unreliable when contrast, position or lighting changes.

Write down these inputs:

• The complete inspection field, including placement variation and tracking margins.
• The smallest feature that changes the pass or fail decision.
• The feature contrast and its possible orientations.
• Part speed, height variation and required captures per part.
• The acceptable false-reject and missed-defect rates for validation.

For initial detection sizing, allocating 3–5 pixels across the smallest feature is a useful starting assumption, not a performance guarantee. Low contrast, irregular edges or classification tasks can require more. A single bright pixel is not dependable evidence that a small feature can be inspected consistently.

Dimensional measurement needs a separate uncertainty budget. The closest companion guide, choosing camera resolution for dimensional inspection, explains why pixel spacing alone does not establish measurement accuracy.

Convert field of view into required image dimensions

Calculate horizontal and vertical requirements separately. Total megapixels can conceal an unsuitable aspect ratio or insufficient pixels along the critical direction.

Use consistent units for field of view and feature dimensions. For a circular defect, its diameter can be the feature dimension in both directions. For a narrow scratch, evaluate its width and orientation rather than substituting its much longer length.

The calculation describes sampling at the object plane. It assumes the feature lies in focus, has adequate contrast and is not blurred by motion. It also assumes the chosen field covers the entire inspection envelope.

A lens and sensor together establish that field. Changing sensor dimensions while retaining the same lens and working distance can change coverage, so a camera replacement is not automatically a pixel-only upgrade.

Use the machine vision lens selection guide when the sensor, working distance or available mounting space also changes.

Worked example: a small defect across a wide field

Suppose the inspection initially covers 200 mm horizontally and 150 mm vertically. The smallest relevant defect is 0.20 mm across. Choose an initial target of 4 pixels across that feature.

The arithmetic is:

Now include part-placement variation. If the required field becomes 220 mm × 165 mm, the same sampling target requires:

The table compares hypothetical image grids over that expanded field. These are sampling examples, not product specifications.

The largest grid offers more samples, but the table cannot prove better detection. That depends on the optical image and the inspection algorithm.

Also check the actual dimensions of any candidate. A hypothetical 4,000 × 4,000 image contains 16 MP yet still misses the horizontal requirement. Use area scan cameras to identify candidates, then verify active image dimensions in their documentation rather than selecting by megapixels alone.

Check lens detail and sensor coverage separately

Sensor coverage asks whether the lens forms a usable image across the sensor. Resolving performance asks whether that image preserves the contrast needed for your feature. Passing the first test does not establish the second.

An undersized image circle can cause dark corners or poor edge performance. Even with adequate coverage, fine detail can lose contrast toward the field edge. Review lens sensor format and coverage before treating matching mounts as proof of compatibility.

The LS-FFL-0602 is an 8.5 mm, 1", 10 MP, C-mount machine vision lens. The LS-FFL-0567 is a 16 mm, 1.1", 20 MP, C-mount machine vision lens. Their different focal lengths and coverage specifications make them selection examples, not interchangeable recommendations.

A lens megapixel designation is useful for screening, but it is not a universal guarantee for every sensor, aperture and working distance. Ask for resolving-performance information at the intended conditions, especially when small pixels or field-edge defects drive the purchase.

Before choosing among machine vision lenses, check:

• Sensor dimensions and required image circle, not just the optical-format label.
• Focal length and focus range at the intended working distance.
• Detail contrast at the center and corners.
• Aperture settings that balance light collection, focus tolerance and diffraction.
• Mechanical clearance and camera mount compatibility.

Use the lens finder to narrow the selection. Then use the lens configuration step to organize the matching requirements. Do not assume that closing the aperture indefinitely improves detail; diffraction eventually reduces fine-detail contrast.

Limit motion blur before adding resolution

A moving feature can cross several pixels during one exposure. More pixels do not remove that motion; finer sampling makes the same physical blur cover more pixels.

For motion approximately parallel to the image plane:

Using the expanded-field example, object sampling is 0.050 mm/pixel. At 500 mm/s, allowing 0.5 pixel of motion gives:

At the same speed, a 1 ms exposure produces 0.5 mm of movement, or 10 pixels of blur. That can overwhelm the benefit of additional sampling.

Shorter exposure requires adequate illumination. Increasing gain can brighten the image but cannot restore photons that were never collected. Test feature contrast and noise at the required exposure rather than evaluating only stationary, brightly illuminated samples.

Shutter type also matters. A global shutter avoids the row-to-row exposure timing skew associated with rolling shutters, but it does not prevent blur during the exposure itself. The sorting conveyor camera selection guide connects these checks to moving-part inspection.

Separate frame rate from exposure time and payload

Exposure time controls motion blur. Frame rate controls how often images can be captured. A short exposure does not, by itself, guarantee a high sustained acquisition rate.

For regularly spaced parts, a starting calculation is:

At 500 mm/s with 100 mm part pitch, throughput is 5 parts/s. If the inspection plan requires 2 captures per part, acquisition demand is 10 frames/s before allowing for timing variation. Check the actual spacing of those captures, because a burst can impose tighter timing than the average rate suggests.

Next estimate image payload. For a hypothetical 4,400 × 3,300 image at 20 frames/s, with one byte per pixel:

This excludes transport overhead and additional memory copies. A two-byte pixel container doubles the image payload. Packed pixel formats and multi-channel output need their own calculations.

Check camera readout, interface capacity, host reception, processing and storage independently. The camera frame-rate bottleneck guide provides the next diagnostic path when delivered performance falls below expectations.

If an upgrade changes the acquisition interface, use camera and frame grabber compatibility checks. That connection check comes after establishing that the extra pixels benefit the inspection.

Compare catalog candidates without assuming a complete match

The LS-CMR-0053 is a 12 MP GigE rolling-shutter camera specified at 9 fps. The LS-CMR-0166 is a 12 MP USB3 rolling-shutter camera specified at 30 fps with 4000 × 3000 resolution.

All 2 share: Sensor size 1/1.7". Values as printed in each product's data sheet; open a part for its drawing and full specifications.

These examples show why equal megapixel figures do not establish equal acquisition suitability. Compare interface, shutter, active dimensions and operating conditions alongside the headline resolution. Confirm the supported rate at the intended image size, pixel format and exposure.

The MV-CH120-11UC is a 12 MP USB3 global-shutter color camera.

A global shutter makes this a different shutter-type candidate, not an automatic system recommendation. Confirm its required operating rate, color-processing path and optical compatibility before selection. Color sampling also differs from monochrome sampling, so validate the actual defect rather than assuming identical fine-detail performance.

The camera finder helps narrow the catalog by interface, resolution and shutter. A useful shortlist still requires documentation checks and representative images; the cards are a starting point, not evidence that a particular inspection will pass.

Know when a higher pixel count is the wrong answer

A resolution upgrade is unlikely to help when the limiting factor is weak illumination, glare, defocus or motion. Fix those causes before purchasing additional pixels. Compare candidate images at the same physical field and inspect the smallest feature, not just the apparent sharpness of a resized preview.

Other approaches can be more effective:

• Reduce the field: move inspection closer to a controlled feature location if the process permits it.
• Use multiple views: separate distant inspection regions when one wide image demands excessive resolution.
• Improve contrast: change illumination geometry when the defect blends into the background.
• Control presentation: reduce placement or height variation before spending pixels on unused margins.
• Use a region of interest: reduce transmitted area when the camera supports it, while retaining the needed inspection coverage.

Cropping an existing image does not create finer object sampling. Likewise, digital enlargement increases displayed dimensions without recovering missing optical detail. A camera-side region of interest may reduce payload or improve frame rate, but the available improvement depends on camera implementation.

Check before ordering and validate before acceptance

Use this checklist to turn a shortlist into a purchase specification:

1. Record the inspection envelope. Include placement variation, field corners and part-height extremes.
2. State the sampling target. Document the smallest feature and the chosen pixels across it, including why that assumption is reasonable.
3. Verify active image dimensions. Check both axes instead of accepting a total megapixel label.
4. Confirm optical compatibility. Review coverage, mount, working distance and resolving performance together.
5. Calculate exposure limits. Use production speed and an explicit motion-blur allowance.
6. Check acquisition timing. Include sustained demand, closely spaced triggers and host processing time.
7. Test representative parts. Include borderline defects, acceptable variation and challenging feature positions.
8. Define acceptance evidence. Record detection outcomes, false rejects, missed defects, image loss and processing latency under agreed conditions.

Keep camera settings and lighting conditions fixed when comparing candidates. Otherwise, an apparent resolution improvement may actually come from a brighter exposure or different processing.

Start your shortlist with the camera configuration step. If lens detail, feature contrast or motion remains uncertain, send the field dimensions, minimum feature size, speed and representative images through application support before choosing a higher pixel count.