MTF vs. Moiré Deflectometry: Which is Faster for VR Production Lines?

The Physics of Measurement Cycles

In the mass production of Virtual Reality (VR) optics, Cycle Time (or Takt Time) is the governing economic metric. With production targets often exceeding 50,000 lens modules per day per line, the metrology station cannot afford to be the bottleneck.

For decades, the optical industry has relied on MTF (Modulation Transfer Function) as the gold standard for image quality. However, the unique geometry of VR lenses-specifically their short Effective Focal Lengths (EFL) and high Numerical Apertures (NA)-has exposed severe speed limitations in traditional MTF testing. Conversely, Moiré Deflectometry, a wavefront-based technique, has emerged as a high-speed alternative.

This section deconstructs the physical measurement principles of both technologies to understand why one is inherently faster than the other.

The Mechanics of MTF Metrology

MTF measures the contrast transfer capability of a lens at various spatial frequencies. In a production environment, this is typically performed using the Slanted Edge method or Pinhole imaging.

The Measurement Sequence:

To capture a valid MTF reading for a VR lens, the system must perform the following mechanical operations:

1. Alignment: The lens must be aligned to the optical axis.

2. Through-Focus Scan (The Time Killer): Because VR lenses have a curved field (Petzval curvature), the point of best focus varies across the field of view. The sensor or the lens must physically move along the Z-axis to find the “Peak MTF” position.

3. Field Scanning: A single measurement point (on-axis) is insufficient. VR requires testing at 0°, 10°, 20°, and often up to 50° (peripheral vision). This requires either multiple cameras (expensive) or a rotating goniometer (slow).

The Mathematical Computation:

In VR lens, sensitivity to MTF is especially high because the lens projects discrete digital pixels directly onto the eye within a closed optical system, so even small losses in contrast particularly off-axisimmediately degrade perceived sharpness and visual comfort; mathematically, this degradation is quantified by the MTF, which is derived from the Line Spread Function (LSF) obtained by imaging a sharp edge, differentiating it, and applying a Fourier Transform to describe how spatial frequencies are transferred through the system.

Formula: MTF Calculation

MTF(ν) = | ∫ LSF(x) · e^(-i2πνx) dx |

Where:

• ν (nu) is the spatial frequency (cycles/mm).
• LSF(x) is the intensity distribution of the Line Spread Function.
• ∫ represents the integral over space.

Speed Constraint: The calculation is fast, but the mechanical settling time is slow. Every time the motor moves the lens to a new field angle or Z-position, the system must wait for vibrations to dampen before capturing the image. If the image is blurred by vibration, the MTF drops artificially, causing a False Fail.

The Mechanics of Moiré Deflectometry

Moiré Deflectometry is a Wavefront Sensing technique. Instead of analyzing the image formed by the lens (Image Plane), it analyzes the deviation of light rays passing through the lens (Pupil Plane).

The Measurement Sequence:

1. Alignment: Similar to MTF, but looser tolerances are often acceptable due to software compensation.

2. Snapshot: The camera captures a single image of Moiré fringes formed by two gratings.

3. Computation: The software calculates the wavefront map and, critically, can compute the “Virtual MTF” from that map.

The Physics of Speed: The Talbot Effect

Moiré Deflectometry does not require focusing. It relies on the Talbot Effect, where a diffraction grating creates “self-images” at periodic distances. The lens being tested distorts these shadows.

Because the measurement is based on ray angle deflection, not image contrast, the depth of field is effectively infinite for the purpose of the sensor.

Formula: Ray Deflection Angle

The deflection angle (α) is directly proportional to the shift in the Moiré fringes (δx):

α ≈ (δx · d) / (z · p)

Where:

• δx is the phase shift of the Moiré fringe.
• d is the pitch of the gratings.
• z is the distance between gratings (Talbot distance).
• p is the Moiré fringe period.

Speed Advantage: There is no Z-axis scanning. The system captures the entire wavefront (on-axis and off-axis) in a single snapshot or a rapid phase-shifting burst.

Comparative Analysis: The “Through-Focus” Trap

The single biggest differentiator in speed is the handling of Defocus.

• MTF System: Must physically search for the focal plane. If the lens has a focal length deviation of 50 microns, the MTF machine must “hunt” for it. This search algorithm typically takes 2 to 5 seconds.

• Moiré System: Measures the wavefront curvature. The “Defocus” is simply the Zernike polynomial term Z(2,0). The system measures the lens, calculates the curvature, and mathematically determines where the focal point is. It does not need to move to that point. The “search” is instantaneous and computational, not mechanical.

Physical Motion Requirements

Feature	MTF Station (Production)	Moiré Station (Rotlex)
Z-Axis Motion	Mandatory (Through-Focus Scan)	None (Fixed position)
Field Scanning	Mandatory (Rotation or Multi-cam)	Single Shot (Full aperture)
Vibration Sensitivity	High (Blur = Fail)	Low (Differential method)
Light Source	Incoherent (White/Green LED)	Coherent (Laser diode)
Measurement Type	Image Quality (Contrast)	Surface/Wavefront Slope
Typical Capture Time	15 – 45 seconds	2 – 5 seconds

 

From a purely physical standpoint, Moiré Deflectometry is faster because it eliminates the variable of mechanical time. MTF metrology is bound by Newtonian mechanics (mass, inertia, settling time), whereas Moiré metrology is bound only by photon integration time and processing power.

In the next section, we will explore why VR Lenses specifically exacerbate the slowness of MTF systems due to their high convex curvature and short focal lengths.

 

The VR Factor – High NA and Short EFL

General-purpose optics (like camera lenses) are relatively easy to measure. VR optics are an outlier. They are characterized by Short Effective Focal Lengths (EFL), typically 20mm-40mm, and very High Numerical Apertures (NA). Furthermore, they are often “Pancake” designs (folded optics) with low transmission.

These specific attributes conspire to make traditional MTF measurement significantly slower and more prone to error, while playing into the strengths of Deflectometry.

The “Depth of Focus” Problem in MTF

The speed of an MTF machine is heavily dependent on the lens’s Depth of Focus (DOF).

Formula: Depth of Focus approximation

DOF ≈ ± λ / (2 · NA²)

VR lenses have a high NA to capture a wide field of view.

• High NA = Tiny DOF.
• For a typical VR lens, the DOF might be less than 10 microns.

Impact on Speed:

When the DOF is tight, the “Through-Focus Scan” (discussed in Part 1) requires extremely fine steps. The motor cannot just zip to the focal point; it must step in 1-micron increments to accurately find the peak. If the step size is too large, you miss the peak and report a false failure.

• Result: The Z-scan takes longer. A low-NA lens might need 10 scan steps. A high-NA VR lens might need 50 scan steps.

Moiré Immunity:

Moiré Deflectometry does not suffer from this. Since it measures the slope of the rays emerging from the lens, the “steepness” of the rays (High NA) does not require mechanical scanning. The dynamic range of the system is adjusted by changing the angle of the gratings, not by moving motors.

Pancake Lenses: The “Light Starvation” Issue

Pancake lenses use polarization folding, which results in a transmission efficiency of 10% to 25%.

Impact on MTF Speed:

MTF sensors (CMOS) need a certain Signal-to-Noise Ratio (SNR) to calculate contrast accurately.

• Low Transmission: Means less light hits the sensor.

• Exposure Time: To get a usable signal, the MTF machine must increase the exposure time (integration time) for every single frame.

• The Multiplier: If a through-focus scan requires 50 frames, and each frame now takes 100ms instead of 10ms, the total cycle time balloons by several seconds.

Moiré Immunity:

Deflectometry uses a laser source (Coherent light). Lasers have high spectral density. Even with 10% transmission, enough laser energy passes through the lens to form high-contrast Moiré fringes on the camera. The exposure time remains in the microsecond range.

Field Curvature and Off-Axis Measurement

VR lenses have significant Field Curvature. The focal plane is not flat; it is a bowl shape.

• MTF Challenge: You cannot measure the Center (0°) and the Periphery (40°) simultaneously on a flat sensor. The periphery will be out of focus.
  • Sequence: Focus Center -> Measure -> Refocus Periphery -> Measure.
  • Cost: Each refocus event adds mechanical movement time (approx. 0.5 – 1.0 seconds per field point).

• Moiré Advantage: The Moiré system captures the wavefront of the entire aperture at once. The “Field Curvature” appears as a Zernike term (Defocus varying with radius).
  • Calculation: The software mathematically “refocuses” the data for each field point instantaneously post-capture. It computes what the MTF would be at the best focus for the periphery, without physically moving anything.

Vibration and “Settling Time”

Factory floors vibrate. Robots move, forklifts drive by.

MTF Sensitivity:

MTF measures contrast. If the equipment vibrates by 2 microns during the exposure, the “Edge” blurs. The calculated MTF drops.

To prevent this, MTF stations use heavy granite bases and programmed “Settling Delays”-wait times of 200ms-500ms after every motor move to ensure the system is still.

• In a multi-point scan, these delays accumulate to constitute 30-40% of the total cycle time.

Moiré Robustness:

Moiré Deflectometry is differential. It measures the relative shift between fringes. While not immune to vibration, it is significantly more robust. Common vibrations tend to shift the whole pattern (Tilt), which is easily subtracted mathematically.

• Result: “Settling Delays” can be virtually eliminated.

VR-Specific Penalties

VR Lens Attribute	Impact on MTF Speed	Impact on Moiré Speed
High NA (> 0.5)	Severe. Requires ultra-fine Z-scanning steps.	Negligible. Requires correct grating setup.
Low Transmission (Pancake)	High. Requires longer exposure integration.	None. Laser power is sufficient.
Steep Field Curvature	High. Requires re-focusing for off-axis points.	None. Calculated mathematically.
Aspheric Profiles	Medium. Alignment sensitivity increases setup time.	None. Measures deviation from sphere.

The geometry of the VR lens fights against the mechanical constraints of the MTF machine. The need for precise focusing of high-NA optics creates a “Time Penalty” that cannot be engineered away without sacrificing accuracy.

In the next section, we will calculate the Total Throughput (UPH), breaking down the cycle time second-by-second to quantify exactly how much faster Deflectometry is.

The Throughput Calculation (UPH Analysis)

In this section, we move from physics to operations management. We will model a standard “Cycle Time” for both technologies in a mass-production context.

Scenario: A 100% inspection line for a VR Pancake lens module.

Requirement: Measure MTF/Wavefront at 5 Field Points: Center (0°), +20° (Left/Right), and +40° (Left/Right).

The MTF Measurement Cycle Breakdown

The standard industrial MTF station uses a “Reticle-Collimator-Lens-Sensor” architecture. To measure 5 points, the system usually rotates the lens (or the collimator) and re-focuses at each angle.

Sequence:

1. Load Part: Robot places lens. Vacuum engages. (2.0s)
2. Move to Center (0°): Motor transit + Settling. (0.5s)
3. Auto-Focus (0°): Z-axis scan (approx. 20 steps) + Image processing. (3.0s)
4. Capture & Process (0°): Exposure + FFT calc. (0.5s)
5. Move to Field 1 (+20°): Rotation + Settling. (1.0s)
6. Auto-Focus (+20°): Essential due to field curvature. (2.0s)
7. Capture & Process: (0.5s)
8. [Repeat Steps 5-7 for remaining 3 points]: (3.5s * 3 = 10.5s)
9. Unload Part: (1.5s)

Total Cycle Time (MTF): ~21.5 Seconds

Max Throughput (UPH): ~167 Units Per Hour

Note: Some “Single-Shot” MTF systems exist with 5-9 fixed cameras. These are faster (approx. 6-8 seconds) but are extremely expensive ($200k+) and rigid (cannot change field angles easily).

The Moiré Deflectometry Cycle Breakdown

The Moiré system (e.g., Rotlex) captures the full aperture data in one go.

Sequence:

1. Load Part: Robot places lens. (2.0s)
2. Acquisition: Single snapshot (or rapid Phase Shifting burst). No mechanical movement. (0.5s)
3. Processing:
   • Wavefront Reconstruction.
   • Zernike Decomposition.
   • Virtual MTF Calculation (computed in parallel during Unload). (1.0s – concurrent)
4. Unload Part: (1.5s)

Total Cycle Time (Moiré): ~4.0 Seconds

Max Throughput (UPH): ~900 Units Per Hour

The “Calculated MTF” Advantage

Skeptics argue that Moiré measures Wavefront, not MTF. However, mathematically, they are linked. If you know the Wavefront Error ($W(x,y)$), you can calculate the Auto-Correlation of the Pupil Function to derive the exact MTF.

Formula: Virtual MTF

OTF(ξ, η) = (1 / A) · ∫∫ P(x,y) · P(x-λfξ, y-λfη) dx dy*

Where:

• OTF is the Optical Transfer Function (Magnitude is MTF).
• P(x,y) is the Generalized Pupil Function: P(x,y) = A(x,y) · e^(i · 2π/λ · W(x,y)).
• W(x,y) is the Wavefront Error measured by the Moiré system.

Speed Implication: This calculation is computationally heavy but runs on the GPU. Modern GPUs (NVIDIA RTX series) can solve this integral in milliseconds. Therefore, the “Processing Time” is negligible compared to the “Mechanical Time” saved.

Throughput vs. Footprint (The ROI Calculation)

In a factory, space is money.

• Target: Manufacture 20,000 lenses per day (20 hours uptime).
• Required Rate: 1,000 UPH.

Option A: MTF Stations (167 UPH each)

• Machines needed: 6 machines.
• Robots needed: 6 loaders or complex gantry.
• Floor space: Large.
• Calibration burden: High (6 optical paths to maintain).

Option B: Moiré Stations (900 UPH each)

• Machines needed: 1.2 machines (Safe margin: 2 machines).
• Robots needed: 2.
• Floor space: Small.
• Calibration burden: Low.

Handling “Rejects” and Retests

Another hidden speed factor is the False Fail Rate.

• MTF False Fails: Often caused by dust on the reticle or vibration. Requires re-cleaning and re-testing. (Slows down the line).

• Moiré Diagnostics: If a lens fails on a Moiré system, the Residual Map instantly shows why.
  • Is it a scratch? (High frequency residual).
  • Is it focus? (Power term).
  • Is it dust? (Local spike).
  • This allows “Smart Rejection”-the system can ignore dust-based artifacts that don’t affect optical power, reducing unnecessary re-tests.

The mathematics of throughput are undeniable. For multi-field VR measurement, Moiré Deflectometry is 4x to 5x faster than mechanical MTF scanning.

To match the throughput of a single Deflectometer, a factory would need a bank of 5-6 MTF machines, significantly increasing Capital Expenditure (CapEx) and maintenance complexity.

In the final section, we will look at the quality of the data. Does “Faster” mean “Worse”? Or does the Wavefront approach actually provide better data for process control?

Data Density and Process Control Loop

In the previous sections, we established that Moiré Deflectometry wins the race for Cycle Time (seconds per part). But in precision manufacturing, speed is worthless without accuracy. The ultimate question for the Optical Director is: “Does the Moiré speed come at the cost of reliability?”

This final section argues that Moiré is not just faster; it is richer. It provides diagnostic data that accelerates not just the sorting of lenses, but the tuning of the manufacturing process itself.

The Difference Between “Grading” and “Diagnosing”

MTF is a “Final Exam” Score.

When an MTF machine rejects a lens, it reports: “MTF at 20° is 0.35 (Spec: >0.40).”

• The Problem: It doesn’t tell you why.
  • Is the mold tilted?
  • Is the plastic injection pressure too low (shrinkage)?
  • Is the coating uneven?

• The Action: The operator bins the part and hopes the next one is better. This is Sorting, not Engineering.

Moiré (Wavefront) is a Medical MRI.

When a Moiré system rejects a lens, it provides the Zernike Polynomial Decomposition.

• The Data: “Fail due to dominant Coma (Zernike Term Z7) oriented at 45°.”
• The Diagnosis: Coma is distinct. It usually means Decentration or Tilt. The engineer knows immediately that the mold inserts are misaligned.
• The Action: Stop the machine, adjust the mold insert by 5 microns.
• Speed Impact: This reduces the production of bad parts, which is the ultimate speed optimization.

Feedback Loops to Injection Molding

In VR lens production (Injection Compression Molding), drift is common.

• Scenario: The mold temperature rises slightly over 2 hours.
• Effect: The Spherical Aberration (SA) slowly drifts.

With MTF: You won’t notice until the MTF drops below the fail threshold. Suddenly, you have a pile of scrap.

With Moiré: You track the Z(4,0) Spherical Aberration coefficient in real-time. You see the trend line moving.

• Predictive Maintenance: You can adjust the molding pressure before the lens fails spec.

Virtual Reality: The “Visual” Correlation

Critics of Moiré often say, “But the customer sees Contrast, so we must measure Contrast (MTF).”

While true, the correlation between Wavefront and MTF is mathematically rigorous.

The Strehl Ratio Shortcut:

For diffraction-limited VR optics, the Strehl Ratio is a fast proxy for peak image quality.

Formula: Strehl Ratio approximation

S ≈ e^(-(2πσ)²)

Where:

• σ is the RMS Wavefront Error (in waves).

Moiré measures σ directly. If the wavefront error is low (< 0.07 waves RMS), the physics guarantees that the MTF must be high. There is no physical way to have a perfect wavefront and bad MTF (assuming clear transmission).

Therefore, measuring Wavefront is a valid, faster proxy for MTF.

Handling “Pancake” Assembly

For Pancake lenses, the final assembly involves aligning two lenses and a quarter-wave plate.

• MTF: Measuring the sub-assemblies is hard because they focus light at odd distances.
• Moiré: Can measure Infinite Conjugate or Finite Conjugate setups easily. It can measure the single element, then the doublet, then the full stack.
• Speed: It allows for “inter-stage” testing that catches defects early, rather than waiting for the slow, final End-of-Line MTF test.

The Strategic Choice

Criteria	MTF (Modulation Transfer Function)	Moiré Deflectometry (Wavefront)
Primary Output	Contrast vs. Frequency (Image Quality)	Wavefront Map / Zernike / Diopter
Cycle Time (5 Fields)	Slow (~20 sec)	Fast (~4 sec)
Diagnostic Value	Low (Pass/Fail)	High (Root Cause Analysis)
Setup Complexity	High (Precision Alignment)	Medium (Tolerant to misalignment)
Cost Per Unit (UPH)	High (Capital intensive)	Low (High throughput)
Best Use Case	Lab Verification / Golden Sample	Mass Production / Process Control

Why is Moiré Deflectometry significantly faster than MTF in a mass production environment?

The speed advantage stems fundamentally from the physics of the measurement. Traditional MTF systems rely on Mechanical Scanning: they must physically move the lens or sensor along the Z-axis to find the best focus (“Through-Focus Scan”) and rotate to measure different field angles (0°, 20°, 40°). Each movement requires time for motion and vibration settling. In contrast, Moiré Deflectometry is a Wavefront Sensing technique. It captures the data for the entire aperture (on-axis and off-axis) in a single Snapshot. There is no need for mechanical focusing or scanning; the focus position and field curvature are calculated mathematically from the ray deflection data. This reduces the cycle time from ~20 seconds (MTF) to ~4 seconds (Moiré).

Is “Virtual MTF” calculated from Wavefront data as reliable as direct optical measurement?

Yes, there is a rigorous mathematical link between Wavefront Error and MTF. By calculating the Auto-Correlation of the Pupil Function, the system derives the exact MTF from the wavefront map. For VR lenses, which are typically designed to be diffraction-limited, Wavefront measurement is often more reliable than direct contrast measurement. This is because it isolates pure optical errors from environmental noise (like vibration or air turbulence) that can artificially lower contrast in a standard MTF bench. If the RMS Wavefront Error is low, the physics guarantees that the MTF is high.

How does the High Numerical Aperture (NA) of VR lenses impact the measurement speed?

VR lenses have a very high, resulting in an extremely shallow Depth of Focus (DOF), often less than 10 microns. For MTF systems, a shallow DOF is a bottleneck. The machine must perform the Z-axis scan in incredibly fine steps (e.g., 1 micron) to locate the peak MTF without missing it. This “hunting” process consumes significant time. Moiré Deflectometry is immune to this constraint. It measures Ray Deflection Angle, not image contrast. The system captures the ray slopes regardless of the shallow DOF, eliminating the need for slow, precision mechanical scanning.

How do the two technologies handle the low transmission (10-25%) of Pancake lenses?

Pancake lenses absorb a significant amount of light due to polarization folding. MTF Systems: Rely on standard CMOS sensors and incoherent light.. To compensate, the system must increase the Exposure Time for every frame, slowing down the total cycle significantly. Moiré Systems: Use Coherent Laser Sources. Lasers provide high spectral energy density. Even with 10% transmission, enough laser power passes through the lens to form high-contrast interference fringes on the camera in microseconds, ensuring that measurement speed remains unaffected by the lens’s efficiency.

Beyond Pass/Fail, what diagnostic value does Moiré Deflectometry offer for injection molding?

While an MTF machine gives a binary “Score” (e.g., Contrast = 0.35), a Moiré system provides a Root Cause Diagnosis via Zernike Polynomial decomposition. It can identify exactly why a lens failed:

• Coma: Indicates mold misalignment or tilt.
• Spherical Aberration: Indicates issues with injection pressure or cooling rates.
• Astigmatism: Indicates uneven material stress or clamping issues. This data allows production engineers to adjust the molding parameters in real-time (Process Control), rather than just sorting out bad parts after they are made.

How do factory vibrations affect the reliability of MTF vs. Moiré measurements?

MTF: Is highly sensitive to vibration. Since it measures contrast, any movement during the exposure blurs the edge, causing a drop in MTF and a potential False Fail. This necessitates heavy isolation tables and programmed “Settling Delays” (waiting for the motor to stop shaking) which kill throughput. Moiré: Is a Differential technique. It measures the relative shift between fringes. Common vibrations tend to shift the entire pattern (Tilt), which can be mathematically subtracted. Therefore, Moiré is far more robust in a noisy factory environment and requires minimal settling time.

Conclusion

The question “Which is faster?” has a clear answer: Moiré Deflectometry is significantly faster.

By eliminating mechanical scanning, focusing, and settling times, it achieves throughputs 400%-500% higher than traditional MTF stations.

However, the more important conclusion for the VR industry is not just about speed; it is about Yield.

• MTF tells you which lenses to throw away.
• Moiré tells you how to fix the molding machine so you don’t have to throw them away.

For the high-volume, low-margin, high-precision world of VR optics, the shift from “Image-Based” (MTF) to “Wavefront-Based” (Moiré) metrology is not just a time-saver-it is an operational necessity.