Application Guides

Roll-to-Roll Film Defect Inspection with Line Scan Cameras

Plan roll-to-roll film defect inspection by web width, line scan camera route, line light uniformity, encoder timing, smallest defect and false reject control.

Roll-to-roll film defect inspection with line scan camera line light and encoder synchronized web

Illustrative inspection layout. The sizing assumptions and proposed tests below do not describe measured results from this image.

Direct answer

Roll-to-Roll Film Defect Inspection with Line Scan Cameras

Size roll-to-roll film inspection from usable web width, defect contrast and maximum travel speed. Calculate cross-web pixel spacing and travel per acquired row separately, then check exposure, encoder timing, illumination and sustained data capture. Line scan fits continuous moving film; indexed or stationary small fields may be simpler with area scan. Nominal pixel spacing alone does not prove a defect can be detected.

Sources and editorial review

Primary references for this selection guide.

Content updated: October 23, 2026. Calculations and selection guidance; no new physical test is claimed.

About the engineering team

Basler

GigE Line Scan Use Case Descriptions and Diagrams

Manufacturer documentation explains fixed line rate for predictable motion and encoder triggering when roller-driven material speed varies. Example numbers above are declared assumptions.

Teledyne Vision Solutions

What drives line-scan image quality at production speeds

Manufacturer method article dated 2026-09-18; supports separate sampling axes, light/exposure limits, optical coverage, dark/bright calibration and data-path checks. It does not establish Deyi performance or universal camera feature support.

Where this matters

Start with the inspection condition.

For continuous film, line scan is an imaging system: camera, line light, lens, encoder and defect logic must be designed from web speed and defect size together.

Why projects fail

Confirm the limits that change hardware.

Edge-to-center lighting uniformity controls false defects.

RFQ preparation

Send enough context for a real review.

Encoder synchronization protects defect shape and measurement scale.

What engineering should check

What this page should help teams decide.

Line scan fits continuous film and wide web inspection.
Edge-to-center lighting uniformity controls false defects.
Encoder synchronization protects defect shape and measurement scale.

Related Deyi routes

Product families commonly involved.

Machine Vision Lighting Machine Vision Lenses

Practical note

Continuous web motion favors line scan.

Use line scan when continuous motion provides the second image axis and the lens can cover the usable web width. Use area scan for an indexed or stationary field that fits the required image and timing limits. If web height, flutter or available light prevents a repeatable image, resolve that physical limit before selecting a detector or quoting a camera.

Practical note

Lighting uniformity is a core acceptance item.

Compare reflected and transmitted illumination on the actual film and defect types; scratches, inclusions and coating variation need different contrast. Hold exposure, geometry and processing fixed while comparing light arrangements. Check both edges and the center for glare, clipping and low signal. A longer light or a flat-field correction cannot recover detail that was never imaged.

Practical note

Encoder timing stabilizes image geometry.

For variable roller motion, verify encoder distance per pulse, scaling to each acquired row, slip, direction and acceleration limits. A fixed timer can stretch or compress the reconstructed image when speed changes. Log missing rows and test the actual trigger input and receiving data path; an encoder connection alone does not prove correct geometry or reject timing.

Practical note

False reject control must be defined early.

Freeze labeled clean, defective and nuisance samples before tuning; retain a separate evaluation set. Agree missed-defect and false-reject limits, minimum defect classes, speed and height range, sample counts and downstream output timing before testing. Report observed counts with the number inspected; a camera datasheet or illustrative image is not a detection-rate measurement.

Practical note

Keep calibration and production evidence reproducible.

After the specified warmup, acquire dark and bright references using the selected model's supported calibration procedure. Record lens, aperture, exposure, gain, pixel format, light settings and encoder scale. Capture raw film images at both edges and the center across minimum, nominal and maximum speed and permitted web height. Recheck after cleaning or changing optics or light geometry; retain configuration files with the images.

How to test before buying

Use this guide as a pre-RFQ decision filter, not as a part-number shortcut.

Machine vision selection is usually stable when the project starts from the inspection condition instead of a catalog model. Before requesting a quote, define what must be detected or measured, how the part moves, what surface behavior affects contrast and which factory constraint cannot change.

Use this guide to translate the requirement into testable inputs: sample images, target tolerance, line speed, field of view, working distance, mounting envelope and the current failure mode. That gives the factory enough evidence to map the request to camera, lighting, optics, reader or 3D routes.

Decision checks

Three checks before locking the route.

Cross-web sampling

Usable field width ÷ active pixels across that field = nominal mm/pixel; check edge coverage and lens performance.

Down-web sampling

Maximum speed ÷ chosen travel per row = required rows/s; exposure and model-specific acquisition limits must also fit.

Image payload

Pixels per row × rows/s × transported bytes per pixel; add overhead and verify continuous delivery, processing and storage.

Engineering method and assumptions

Film inspection sizing: two sampling axes and a sustained data path

Illustrative sizing exercise, not a Deyi camera specification or a measured defect-detection result: assume a 200 mm inspection width, a 2,048-pixel line sensor, web speed of 1,000 mm/s and a target of 0.10 mm travel per acquired row. Use the active pixels across the inspected width; leave any edge margin out of this simplified calculation.

Film inspection sizing: two sampling axes and a sustained data path
Input or check	Calculation	What still needs verification
Across-web sampling	200 mm ÷ 2,048 pixels = 0.0977 mm/pixel	Verify the actual sensor active width, optical field, focus and usable edge coverage.
Along-web sampling	1,000 mm/s ÷ 0.10 mm/row = 10,000 rows/s (10 kHz)	Confirm the selected camera and interface can sustain the required line rate at the actual pixel format.
Time between rows	1 ÷ 10,000 rows/s = 0.0001 s (100 µs)	Exposure, readout and trigger timing must fit the model-specific acquisition limits.
If speed rises by 20%	1,200 mm/s ÷ 0.10 mm/row = 12,000 rows/s	A fixed 10 kHz line rate would stretch the along-web pixel spacing to 0.12 mm/row. Review encoder-triggered acquisition for variable speed.
Uncompressed Mono8 image payload	2,048 pixels/row × 10,000 rows/s × 1 byte/pixel = 20.48 MB/s	Transport overhead, buffering, processing and storage are additional; check the actual output format.

Across-web sampling and along-web sampling are separate. A nominal 0.10 mm grid does not prove a 0.10 mm defect can be detected; test contrast, blur, defect orientation and false rejects with representative moving material.

At variable speed, use an encoder and verify the trigger-to-row relationship on the real transport. Encoder pitch, mechanical slip and direction handling must be checked; a stationary sample cannot validate geometry.

Send material width, minimum defect size, speed range, surface finish, available line-light geometry, encoder details and sample good/bad images before selecting a camera or illumination route.

Proposed validation protocol, not a completed physical test: freeze labeled clean, defective and nuisance film; use a separate evaluation set at the agreed speed and height limits. Record missing rows, false accepts, false rejects and trigger-to-reject time with the number inspected and raw images.

The calculation reuses the assumptions of the line-scan primer. It is not a supplier camera rating, an acceptance target or proof of detection accuracy. Manufacturer method references were checked on 2026-10-23; verify supported calibration and trigger functions for the exact chosen model.

Technical basis: GigE Line Scan Use Case Descriptions and Diagrams; What drives line-scan image quality at production speeds.

Line-scan camera candidates Follow the shared sampling calculation Check encoder acquisition Plan line-light geometry

Decision table

Selection factors and practical tradeoffs.

Factor	Practical rule	RFQ impact
Cross-web sampling	Usable field width ÷ active pixels across that field = nominal mm/pixel; check edge coverage and lens performance.	Send full usable width, edge margin, defect orientation and labeled raw images.
Down-web sampling	Maximum speed ÷ chosen travel per row = required rows/s; exposure and model-specific acquisition limits must also fit.	Send speed range, motion profile and encoder distance per pulse.
Image payload	Pixels per row × rows/s × transported bytes per pixel; add overhead and verify continuous delivery, processing and storage.	State the actual output format, buffering, frame grabber/interface and image-retention policy.
Smallest defect	Pixel spacing is a sizing input, not a guaranteed detection limit; prove contrast and outcomes on held-out moving film.	Send defect size, type and sample images.
Speed variation	Use encoder when geometry or defect shape must stay stable.	Send speed range and encoder details.
False reject target	Separate dust, wrinkles and real defects in the acceptance plan.	Clarifies algorithm and operator review scope.

Application proof

Related delivery routes that make this selection decision concrete.

View all cases

Textile and web inspection

Roll-to-Roll Web Defect Inspection

Wide moving material needs even illumination, encoder synchronization and enough line rate to avoid missing small defects.

Common mistakes

Problems that slow down selection.

Selecting by model number before the inspection target is measurable.
Treating lighting as an accessory instead of the main contrast-control tool.
Ignoring fixture stability, part variation and operator maintenance workflow.

Factory handoff

What Deyi Vision reviews after receiving the project details.

The factory route review starts by checking whether the image can be made stable with lighting and fixture control. Then the camera, lens, reader or 3D sensor route is sized against speed, resolution, interface and installation constraints.

If you already have a Keyence, Cognex, Basler, OPT, LMI, Hikrobot or barcode-reader reference, include it as a reference model. Deyi Vision uses it to understand the application class; final selection still depends on real samples and production limits.

Guide to RFQ

Have a real part, sample image or production constraint?

Use the guide to frame the question, then send the details so engineering can recommend a route.

Request engineering RFQ

Guide FAQ

Questions related to roll-to-roll film defect inspection with line scan cameras.

Ask engineering

Why use line scan cameras for roll-to-roll film inspection?

They build an image row by row from controlled motion and can cover a continuous web. Select them only after checking both sampling axes, illumination and timing. For small indexed or stationary fields, an area-scan arrangement may be simpler.

What causes false defects in film inspection?

Uneven lighting, dust, wrinkles, web flutter, reflections, lens shading and speed variation can all create false defect signals.

Does a 0.10 mm sampling grid guarantee detection of a 0.10 mm defect?

No. Sampling does not establish defect contrast, blur, optical resolution or algorithm performance. The declared calculation below sizes an acquisition route; evaluate representative defects and nuisance samples at production speed before agreeing a detection limit.

What should I send for a roll-to-roll inspection RFQ?

Send web width, material type, speed range, smallest defect, sample images, encoder availability, lighting constraints, false reject target and required output.