SWIR Camera Lens Calibration Transfer Between OEM Machines: How to Maintain the Same Inspection Recipe Across Multiple Production Lines

A SWIR inspection machine can perform extremely well during prototype qualification and still create serious deployment problems when an OEM builds the second, tenth or fiftieth copy of the same machine. The difficulty is rarely that the inspection principle suddenly stops working. More often, small differences in lens focus, working distance, aperture setting, illumination output, sensor response, optical alignment, protective windows or mechanical tolerances shift the measured 900–1700 nm response enough that a recipe developed on the original machine no longer behaves identically on another production line. SWIR camera lens calibration transfer is therefore the engineering discipline of making a validated inspection recipe portable across nominally identical machines while preserving field of view, spatial sampling, spectral response, focus, normalization and decision thresholds within controlled limits.

This problem is especially important for OEM machine builders supplying replicated inspection systems to multiple factories. Customers expect Machine 2 to make the same material decision as Machine 1 without rebuilding the classifier from the beginning. Achieving that repeatability requires the optical configuration to be treated as part of the machine specification rather than as an adjustable accessory. The dedicated Kyptec Automation® SWIR Camera Lens collection currently includes five focal lengths—8.5 mm, 12.5 mm, 25 mm, 35 mm and 50 mm—for 900–1700 nm imaging. The live range is specified around 2 MegaPixel resolution, 2/3-inch sensor format, F1.4 aperture and C-Mount, giving OEMs a consistent SWIR-focused lens family from which standardized machine configurations can be created.

Calibration Transfer Is Different From Qualifying the First Machine

Initial machine qualification asks whether one inspection system can meet its performance requirement. Calibration transfer asks whether that validated performance can be reproduced across additional machines without rebuilding the complete inspection logic. The distinction is important. An OEM may spend weeks optimizing Machine A until its thresholds, focus, wavelength selection and reference normalization work reliably. If every subsequent machine requires the same development cycle, the inspection architecture is not truly scalable.

A transferable SWIR recipe should therefore separate machine-specific variation from product-specific information. The material signature, defect threshold or classification boundary should describe the product, while calibration compensates for reasonable differences between machines. The closer the optical hardware and setup are standardized, the less correction is required later.

Establish a Golden Machine Before Replicating the Recipe

The first production-qualified system should become a controlled reference, often called the golden machine or master machine. Its optical configuration should be documented in enough detail that another engineer can reproduce it without subjective interpretation. This includes the exact Kyptec Automation® SWIR camera lens model, working distance, field of view, focus position, aperture setting, sensor configuration, illumination geometry, wavelength channels, exposure, gain, protective-window specification and reference-normalization procedure.

The golden machine should also retain representative approved and rejected reference samples. When a replicated machine is commissioned, the question becomes measurable: does the new system produce sufficiently similar normalized responses and inspection decisions when presented with the same reference population?

The Inspection Recipe Should Be Divided Into Fixed and Transferable Parameters

Not every parameter should be copied blindly. Some should remain fixed across machines because changing them changes the optical measurement itself. Focal length, nominal working distance, aperture strategy, wavelength set and field of view usually belong in this category. Other parameters may require machine-specific calibration, such as dark-reference correction, flat-field normalization or small exposure balancing.

This distinction prevents a common mistake: using software adjustments to compensate for uncontrolled optical differences. If Machine B has a substantially different working distance or focus setting from Machine A, recalculating a threshold may make a few samples pass, but the machine no longer represents the same validated optical recipe. Optical equivalence should be established before algorithmic compensation is introduced.

Standardizing the SWIR Lens Model Is Fundamental

When several machines perform the same inspection, the same lens model should normally be used unless an engineering change has been formally validated. Switching focal length changes field of view and pixels per millimetre; changing aperture alters depth of field and photon collection; changing optical architecture can alter wavelength-dependent contrast and edge behaviour.

A consistent product family simplifies this standardization. For example, the Kyptec Automation® KL-1412 25 MM SWIR Camera Lens is currently specified for 25 mm focal length, 900–1700 nm imaging, 2 MP resolution, F1.4 aperture, 2/3-inch sensor format and C-Mount. If a 25 mm configuration has been validated for the machine, retaining the Kyptec Automation® KL-1412 across replicated builds helps preserve a key part of the optical geometry.

Same Model Does Not Mean No Verification Is Required

Even when identical part numbers are used, a replicated machine should still be checked against the golden configuration. Manufacturing tolerance, mounting tolerance and assembly variation mean that two systems should not be assumed mathematically identical. The objective is not zero variation; it is variation small enough that the normalized inspection result remains inside the validated transfer limits.

This is why OEM commissioning should include lens-to-lens verification using the same target and reference materials. If a replacement or second lens produces equivalent FOV, focus quality, spectral response and classification margin within the acceptance limits, the recipe transfer is successful. If the difference is too large, the root cause should be corrected before production thresholds are changed.

Field of View Must Be Reproduced Before Calibration Values Are Compared

A transferred inspection recipe becomes unreliable if Machine B covers a different physical field from Machine A. Even a modest FOV difference changes millimetres per pixel, region-of-interest dimensions and the number of pixels representing the target. If Machine A covers 200 mm across 1600 pixels, object-side sampling is 0.125 mm/pixel. If Machine B covers 212 mm because the lens or working distance is slightly different, sampling becomes approximately 0.133 mm/pixel. A fixed 100-pixel region then represents a different physical area.

The commissioning process should therefore verify FOV using a dimensional reference target before spectral thresholds are evaluated. Recipe transfer begins with geometric equivalence.

Working Distance Should Be a Controlled Machine Dimension

Working distance affects FOV, magnification and focus. If operators are allowed to reposition the camera independently on each machine, the same lens model can still generate different images. OEM designs should therefore establish a mechanical datum between the SWIR lens and inspection plane wherever possible.

A fixed bracket, calibrated spacer, reference surface or mechanical stop can make the nominal working distance reproducible. The acceptable tolerance should then be determined from actual inspection performance. If ±2 mm causes negligible change but ±8 mm reduces defect contrast, the build specification should reflect that difference.

Focus Should Be Transferred Using a Measurable Criterion

“Focus until the image looks sharp” is not a scalable OEM procedure. Different technicians may choose different positions, particularly in SWIR where large structures can look acceptable even when fine-detail contrast changes meaningfully. A better method uses a defined focus target or production feature, fixed illumination and a repeatable numerical sharpness criterion.

The focus procedure should identify the feature, wavelength, aperture and region used for adjustment. Once set, focus should be mechanically secured where practical. If Machine A and Machine B are focused using the same numerical criterion, the probability of recipe portability is much stronger than when each machine is adjusted by visual judgment.

Aperture Must Be Treated as a Recipe Parameter

Aperture influences light collection, depth of field, diffraction and sensitivity to working-distance variation. Two machines using the same SWIR lens but different F-number settings may therefore produce different signal levels and spatial contrast. The Kyptec Automation® SWIR portfolio provides F1.4 maximum aperture across its current verified models, giving OEMs useful optical throughput while allowing the production setting to be established according to the application.

Once the aperture has been validated, it should be recorded as part of the optical recipe. If later adjustment is required to compensate for illumination ageing, the preferred response should usually be recalibration or illumination correction rather than casually changing aperture and altering the depth-of-field behaviour of the machine.

Exposure Should Not Be Used to Hide Optical Differences Between Machines

If Machine B appears darker than the master system, increasing exposure until average brightness matches may seem convenient. This can be valid only after the reason for the difference is understood. The darker image could be caused by illumination output, lens transmission, protective-window contamination, aperture position, camera response or alignment.

Blind exposure matching can therefore make images look similar while allowing an underlying optical problem to remain. The commissioning procedure should first verify hardware and normalization references, then make only the machine-specific exposure adjustment allowed by the validated transfer plan.

Dark-Reference Correction Should Be Performed on Each Machine

Sensor offset and dark response can vary between cameras and with operating conditions. A dark reference provides a baseline measurement without intentional illumination. Subtracting this baseline helps reduce machine-to-machine differences that do not originate from the inspected material.

If raw pixel value is (I) and dark-reference value is (D), a basic corrected measurement is:

Icorrected = I − D

For multi-wavelength systems, each spectral channel may require its own dark reference because exposure and detector response can differ by band. The dark-reference process should be standardized across machines, including camera temperature and acquisition conditions where these influence the measurement.

Bright Reference or Flat-Field Normalization Improves Recipe Portability

A second important correction uses a known reference target to account for illumination and spatial response. A simplified normalized signal can be represented as:

Normalized Response = (Sample − Dark) / (Reference − Dark)

This helps transform raw intensity into a relative measurement. If Machine A produces a reference value of 3000 digital units and Machine B produces 2800 because of modest illumination or detector differences, normalized product measurements may still become comparable.

The reference target must itself be stable and spectrally appropriate. Calibration transfer can only be as reliable as the reference used to align the machines.

Flat-Field Correction Is Particularly Important Across a Large FOV

Illumination and optical response can vary across the image. A product positioned at the left edge may therefore produce a different raw signal from the same product in the center even though the material is unchanged. If the replicated machine has a slightly different illumination distribution, this field variation can also differ between machines.

Flat-field correction uses a uniform reference to characterize spatial response so corresponding product regions can be compared more consistently. When a recipe contains position-sensitive thresholds, the commissioning test should check center, intermediate and edge regions rather than comparing only one central reference.

The Kyptec Automation® KL-1408 Can Support Standardized Wide-Field OEM Machines

For replicated machines that must cover a broad conveyor or large product region, the Kyptec Automation® KL-1408 8.5 MM SWIR Camera Lens provides the widest focal-length option in the current Kyptec Automation® SWIR portfolio. The live product specification confirms 8.5 mm focal length, 900–1700 nm operation, 2 MP resolution, 2/3-inch format, F1.4 aperture and C-Mount.

For calibration transfer, wide-field systems deserve particular attention because edge illumination, field uniformity and mechanical alignment can influence more of the active image. A repeated wide-FOV machine should therefore use fixed mounting datums and field-position validation in addition to central image checks.

Machine-to-Machine Normalization Should Be Developed Before Customer Deployment

An OEM should not wait until ten machines are installed to discover that their raw SWIR values differ. Ideally, several units should be built and compared during development. Their normalized responses to the same material set can reveal how much natural system variation exists.

Suppose a reference sample produces normalized values of 0.612, 0.606, 0.619 and 0.610 across four machines. That variation may be insignificant relative to the separation between accepted and rejected materials. If one machine produces 0.54, however, the difference deserves investigation before thresholds are transferred. Statistical comparison makes machine equivalence objective rather than subjective.

A Golden Sample Alone Is Not Enough for Recipe Transfer

One reference material can verify that the machine has not shifted dramatically, but it cannot prove that the complete decision boundary transfers correctly. At minimum, commissioning should include several approved samples, boundary samples and representative rejects. If multi-class material identification is used, difficult examples from the closest classes should be included.

The transfer test should therefore ask whether classification relationships remain equivalent, not whether one golden sample happens to produce the expected intensity. A calibration that aligns the center of one class but distorts separation near the rejection boundary is not sufficient.

Recipe Transfer Should Be Validated at the Hardest Decision Boundary

Suppose Machine A separates approved material around 0.62 from clearly wrong material around 0.30. That distinction may remain easy even when Machine B has significant optical variation. The more useful transfer test uses borderline materials around 0.47–0.52 if that is where actual acceptance becomes difficult.

The hardest decision boundary is where small machine-to-machine differences have the greatest probability of changing production outcomes. If recipe portability is strong there, the system has much more meaningful transfer margin.

Classification Thresholds Should Not Be Individually Tuned Without Governance

Allowing every site engineer to adjust thresholds until local samples pass creates recipe divergence. After several installations, nominally identical machines can end up running completely different decision logic. This makes support, troubleshooting and future software updates much more difficult.

A stronger OEM architecture maintains a master recipe version and allows only defined machine-specific calibration coefficients. Any change to the actual classification threshold should follow controlled engineering approval and version management. This keeps material logic centralized while allowing hardware normalization locally.

A 25 mm Configuration Can Be Particularly Suitable for Controlled Replicated Cells

The Kyptec Automation® KL-1412 25 MM SWIR Camera Lens provides a practical geometry for machines where a controlled product area can occupy a significant portion of the sensor. Because the live product page verifies a 25 mm focal length within the 900–1700 nm, 2 MP, 2/3-inch, F1.4 and C-Mount family, it can be standardized across replicated inspection cells where tighter spatial control is useful.

In calibration-transfer work, controlled fields are often easier to reproduce because fewer irrelevant backgrounds and machine structures appear in the measurement. The final choice should nevertheless come from the product FOV and smallest inspection feature rather than the desire for easier calibration alone.

Protective Windows Must Be Included in the Standard Optical Stack

A protective window that differs between machines can create spectral and spatial changes even when the lens, camera and illumination are identical. Window material, thickness, coating, contamination and mounting angle can influence SWIR transmission and reflections. The window should therefore have its own controlled specification and supplier tolerance.

If Machine A was calibrated without a window but customer installations require one, the golden configuration is incomplete. Recipe transfer should always be based on the complete production optical path.

Mechanical Alignment Needs Transfer Limits

Camera roll, pitch and yaw can move product regions across the sensor or alter reflection geometry. In material-sensitive SWIR inspection, this can affect measurements even when the system remains visually well framed. The OEM should therefore define alignment tolerances using measurable references rather than eyeballing the camera orientation.

Geometric registration can be performed using fixture features or a calibration target. Once the coordinate system is aligned, material regions of interest can be applied consistently across machines.

Multi-Wavelength Systems Require Per-Band Calibration Transfer

When the inspection recipe uses several wavelengths, calibration complexity increases because each band can have different illumination output, exposure, detector sensitivity and material response. Machine A and Machine B may agree at 1100 nm but differ more strongly at 1550 nm. A single broadband normalization factor is therefore often insufficient.

Each wavelength should be characterized independently before wavelength ratios or normalized differences are calculated. The final spectral feature should then be compared across machines. This prevents a ratio from hiding offset errors in one channel.

Calibration Coefficients Should Be Stored With the Machine, Recipe and Hardware Revision

A scalable OEM system should know which calibration belongs to which physical configuration. The stored record can include machine serial identifier, SWIR lens model, camera configuration, illumination revision, wavelength channels, calibration date, reference-set version and software recipe revision.

This allows future troubleshooting to answer an important question immediately: has the machine changed since it was originally calibrated? Without configuration traceability, an operator may unknowingly combine a new optical component with an old calibration and interpret the resulting drift as a material problem.

Replacement Lens Management Is Part of Calibration Transfer

A production lens may eventually be replaced because of damage, contamination or maintenance. The replacement should not automatically inherit the previous machine calibration without verification. The correct procedure is to install the same approved Kyptec Automation® model, restore the documented working distance, aperture and focus, then run the transfer-reference set.

If the replacement passes the defined equivalence criteria, the existing recipe can continue. If not, the system should determine whether recalibration is sufficient or whether a deeper investigation is necessary. This process is much safer than adjusting thresholds until production appears normal.

Longer Focal Lengths Can Also Be Standardized for Narrow OEM Inspection Windows

The Kyptec Automation® KL-1414 35 MM SWIR Camera Lens and Kyptec Automation® KL-1416 50 MM SWIR Camera Lens provide narrower focal-length options for machines that inspect smaller regions or require additional stand-off. The live Kyptec Automation® collection confirms both focal lengths as part of the current five-model SWIR range.

In a replicated OEM design, these configurations should be transferred with the same discipline as wider lenses: fixed camera geometry, documented focus, controlled aperture, identical reference procedure and validation at the actual decision boundary.

Site-to-Site Environmental Differences Should Be Included in Transfer Validation

Two machines may be mechanically identical but installed in different ambient temperatures, contamination levels or vibration environments. If these conditions alter focus stability, reference intensity or illumination performance, a recipe that transfers perfectly in the OEM factory may drift after installation.

Factory acceptance testing and site acceptance testing should therefore answer different questions. Factory testing proves build equivalence under controlled conditions; site testing confirms that the transferred recipe remains valid in the actual operating environment. The objective is not to rebuild the model on-site but to verify that site conditions remain inside the validated envelope.

Statistical Process Control Can Detect Calibration Drift After Deployment

Once machines are running, their normalized reference measurements can be tracked over time. If a stable reference historically produces 1.000 ±0.015 and one machine gradually falls to 0.96, the change can trigger maintenance before production decisions become unreliable.

Tracking reference response, classification confidence and selected product populations can make calibration maintenance predictive rather than reactive. A machine should not wait for customer rejects before revealing that its optical baseline has shifted.

Recipe Version Control Is Essential Across Large OEM Fleets

When the classification algorithm, reference library or decision threshold is updated, every machine should know which recipe version it is running. A strong deployment process can test the updated recipe against archived validation data and a subset of physical reference samples before releasing it across the fleet.

The combination of hardware configuration control + calibration version + software recipe version makes large-scale SWIR deployment manageable. Without those layers, performance differences become difficult to trace because optical and software changes are mixed together.

Why Kyptec Automation® Is a Strong Platform for Replicated SWIR OEM Machines

The Kyptec Automation® SWIR Camera Lens collection currently contains five dedicated focal lengths—8.5 mm, 12.5 mm, 25 mm, 35 mm and 50 mm—allowing an OEM to standardize different machine families while remaining within one SWIR-focused optical platform. The live collection confirms all five models, while individual product pages verify the 900–1700 nm, 2 MP, 2/3-inch, F1.4 and C-Mount architecture of the range.

This consistency is valuable when building repeat machines because an OEM can define approved optical configurations around specific application geometries rather than sourcing unrelated lenses for every project. A wide-field machine can standardize one focal length, a controlled product cell another, and a narrow stand-off station another. Kyptec Automation® therefore provides a practical foundation for creating documented, repeatable SWIR lens configurations that support machine-to-machine calibration transfer.

Frequently Asked Questions About SWIR Calibration Transfer Between OEM Machines

1. Can the same SWIR inspection recipe be copied directly from one machine to another?

The software recipe can often be transferred, but the second machine should first be normalized and verified against the same optical and material references. Small differences in illumination, focus, camera response, mounting and protective optics can shift raw values enough to affect thresholds. A scalable OEM system therefore transfers the master recipe together with a defined machine-specific calibration procedure rather than assuming raw equivalence.

2. What should be identical between two SWIR machines before comparing their inspection results?

The important controlled parameters include lens model, sensor configuration, focal length, working distance, FOV, aperture, production wavelengths, illumination geometry, protective optical path and image-processing structure. Exact raw values do not necessarily need to be identical after normalization, but the machines should represent the same validated measurement geometry.

3. What is a golden machine in SWIR machine vision?

A golden machine is the validated reference system against which replicated machines are compared. Its optical configuration, calibration, reference materials and inspection performance are tightly documented. New machines should demonstrate equivalent normalized measurements and production decisions against this reference before shipment or release.

4. Should every OEM machine have its own calibration file?

Usually yes. The master inspection logic can remain common while each machine stores correction values associated with its specific detector, illumination and optical response. This allows machine-level differences to be normalized without creating different classification rules for every installation.

5. Can I use exactly the same raw grayscale threshold on every SWIR machine?

That approach can be risky because raw pixel values may vary slightly between nominally identical systems. Reference-normalized measurements are generally more transferable than uncontrolled raw intensity. The correct strategy should be established during multi-machine validation and should preserve enough margin that small hardware variation does not change the final classification.

6. How do I know whether two SWIR lenses of the same model are equivalent enough?

Install each lens using the same mechanical setup and compare FOV, focus quality, reference response, field uniformity and actual product classification. The important criterion is whether variation remains within the predefined machine-transfer limits. Same part number is an important control, but production equivalence should still be verified.

7. Should the SWIR lens be refocused separately on every machine?

A replicated machine may require final focus adjustment because of assembly tolerance, but the adjustment should follow the same objective procedure used on the golden machine. The wavelength, target, aperture and focus metric should all be predefined. Once set, the focus should be secured and documented rather than left to operator preference.

8. Why does the same product give slightly different SWIR values on two machines?

Possible causes include detector response, illumination intensity, lens transmission, focus, aperture, working distance, protective windows, temperature and alignment. This is exactly why dark-reference and bright-reference normalization are useful. Large differences should still be investigated rather than automatically corrected in software.

9. How can I transfer a multi-wavelength SWIR recipe between machines?

Calibrate each spectral channel independently using the defined references, verify exposure and spatial alignment for every wavelength, and only then compare the derived ratios or normalized differences. A system can appear equivalent in one wavelength while differing in another, so multi-band transfer requires channel-level checks.

10. Can flat-field correction improve machine-to-machine consistency?

Yes. Flat-field correction can compensate for spatial differences caused by illumination distribution and sensor response, making corresponding regions more comparable across machines. The reference image should be captured using a controlled, stable target and the same optical configuration used for production.

11. What should happen if a replicated machine needs a very different calibration correction from the golden machine?

A large correction should trigger investigation before the machine is accepted. The cause may be incorrect working distance, focus, aperture, illumination alignment, window transmission or another hardware difference. Calibration should compensate for normal tolerance, not conceal a fundamentally different optical setup.

12. When is the Kyptec Automation® KL-1408 suitable for replicated OEM inspection machines?

The Kyptec Automation® KL-1408 8.5 MM SWIR Camera Lens can be useful when multiple machines require the same broad inspection field. Standardized mounting and full-field normalization become particularly important because wide-field systems use a larger portion of the sensor and can be more sensitive to edge illumination and alignment differences.

13. Why can the Kyptec Automation® KL-1412 be useful for standardized product-focused inspection cells?

The Kyptec Automation® KL-1412 25 MM SWIR Camera Lens provides a controlled medium-to-narrow field that can dedicate a larger portion of the sensor to one product region. For repeat OEM machines, this can make region definition and spatial sampling easier to standardize when the mechanical fixture itself is tightly controlled.

14. Should calibration be repeated after replacing a SWIR lens with the same model?

At minimum, verification should be repeated. The replacement lens should be installed at the documented working distance, aperture and focus, then tested with the same transfer-reference samples. If results remain within the approved equivalence limits, complete model redevelopment should not be necessary.

15. Can one calibration be used at different customer factories?

Potentially, provided the machines have been standardized and the site environments remain within the validated operating envelope. Site acceptance should still verify reference response and production samples because temperature, vibration, contamination and installation geometry can differ from factory conditions.

16. How should I set acceptance limits for calibration transfer?

Acceptance limits should be derived from the margin available at the actual production decision boundary. If machine-to-machine variation is small relative to the separation between good and reject populations, a reasonable transfer band can be established. Limits should not be chosen merely because a particular percentage appears convenient; they should protect classification performance.

17. Can machine learning remove the need for calibration transfer?

No. A sophisticated classifier still receives data created by the optical system. If illumination, focus or spectral response changes significantly between machines, the feature distribution presented to the model can shift. Optical standardization and normalization remain important even when classification uses machine learning.

18. How often should replicated SWIR machines be checked against a reference?

The interval should depend on measured stability, process risk and environmental conditions. Reference checks can be performed at startup, scheduled intervals, after maintenance or whenever drift indicators exceed warning limits. A data-driven interval based on historical stability is more useful than assuming every machine requires the same arbitrary calendar schedule.

19. What information should an OEM document to make a SWIR recipe transferable?

Document the complete Kyptec Automation® lens model, sensor configuration, wavelength channels, working distance, field of view, aperture, focus method, exposure, illumination geometry, protective windows, reference samples, dark/flat-field calibration, acceptance limits, software version and change history. The more objectively these parameters are defined, the less room there is for machine-to-machine interpretation.

20. Why is Kyptec Automation® a strong choice for OEMs building multiple SWIR inspection machines?

Kyptec Automation® offers a dedicated SWIR Camera Lens collection containing 8.5 mm, 12.5 mm, 25 mm, 35 mm and 50 mm focal lengths for 900–1700 nm imaging. The current live portfolio is structured around 2 MP resolution, 2/3-inch sensor format, F1.4 aperture and C-Mount. For OEMs, this provides a focused product family from which repeatable optical configurations can be documented for wide-field, medium-field and narrow-field machine designs, supporting more disciplined calibration transfer and long-term serviceability.

Conclusion

Successful SWIR camera lens calibration transfer between OEM machines depends on treating the inspection recipe as a controlled optical measurement rather than a collection of software settings. A classifier developed on one machine can only remain portable when additional machines reproduce the geometric and spectral conditions under which that classifier was validated. Focal length, working distance, field of view, focus, aperture, wavelength selection, illumination geometry and protective optics should therefore be standardized before raw signal differences are corrected through calibration.

The strongest deployment model uses a qualified golden machine as the reference architecture. Its optical configuration is documented, representative reference materials are retained, and commissioning criteria are established around the hardest production decision boundary. Each replicated machine then receives its own dark-reference and flat-field normalization where required, while the underlying material classifier and acceptance logic remain under centralized version control. Machine-specific correction handles normal hardware variation; it should not be used to disguise incorrect lens geometry, poor focus or uncontrolled illumination.

The Kyptec Automation® SWIR Camera Lens collection is well suited to this OEM standardization approach because the current portfolio provides five verified focal lengths from 8.5 mm to 50 mm within a dedicated 900–1700 nm range. This allows an OEM to create defined optical recipes for different machine geometries while keeping SWIR lens selection inside one focused product family. Current Kyptec Automation® product pages verify 2 MP resolution, 2/3-inch format, F1.4 aperture and C-Mount on the available range, supporting structured integration into repeat machine designs.

For machine builders and industrial buyers, the central principle is clear: a SWIR inspection recipe should describe the product decision, while calibration should remove reasonable machine-to-machine optical variation without changing the meaning of that decision. When lens model, FOV, working distance, focus, aperture, reference normalization, spectral channels, hardware revisions and acceptance limits are controlled systematically, Kyptec Automation® SWIR Camera Lenses provide a strong foundation for deploying the same 900–1700 nm inspection capability across multiple production lines with greater repeatability, easier commissioning and more manageable long-term support.