Faro Creaform's DTEX pipeline update for laser-line scanners has reduced time to usable point-cloud data on difficult-to-scan parts by up to 60 percent, the company told Vision Systems Design. The improvement comes from image-processing routines that adapt laser intensity, camera gain, and exposure to a wide range of surface finishes without manual parameter tuning.

What surface finishes break laser-line scanning?

Laser triangulation scanners build a point cloud by imaging a laser stripe projected onto the part. A camera viewing that stripe from a fixed angle resolves the projected line against the part surface; any finish that disturbs the line's intensity or shape corrupts the resulting cloud.

Three finish classes, in particular, defeat the default scan recipe. Dark surfaces absorb laser energy and return a low-intensity signal that the camera's automatic gain cannot always rescue. Reflective and shiny surfaces create ghost returns, so the image-processing chain can mistake a secondary reflection for the projected line and produce noisy data. High-contrast surfaces — alternating dark and bright regions such as machined features adjacent to polished bores — push the camera's exposure time in conflicting directions within a single frame.

How does DTEX handle each finish class?

Marco Torsello, product manager at Faro Creaform, told Vision Systems Design that the dark-surface problem looks deceptively simple on paper. "The way you solve this is by increasing laser intensity and camera gain, but it sounds simpler than it actually is," he said.

DTEX combines several techniques to lift the burden off the operator:

• Adaptive laser intensity and per-pixel gain control that respond frame by frame to local reflectivity.
• Wavelength selection matched to the substrate, so the scanner can use a wavelength the part reflects rather than absorbs.
• Multi-image redundancy and intensity-pattern analysis that separate the projected laser stripe from specular artifacts.
• HDR imaging paired with auto-exposure, which dynamically retunes the camera as the line traverses bright and dark zones in the same scan.

Orcello, also cited by Vision Systems Design, said the algorithm stack pairs intensity-pattern analysis with laser-wavelength selection to discriminate the genuine laser line from parasitic reflections. Torsello added that the HDR and auto-exposure combination handles the high-contrast case dynamically, in real time, rather than asking the user to plan an exposure map for the part.

How was the 60 percent figure measured?

Faro Creaform tested the updated pipeline on a part it describes as "definitely challenging to scan before the release of the new set of algorithms." Torsello told Vision Systems Design that cycle time on that part dropped by up to 60 percent from the previous workflow. He stressed that the benchmark applies to difficult finishes; for parts that already scanned well, the gain shows up as setup automation rather than raw scan speed.

The company has not published the test part's geometry, the laser class used, or the acceptance criteria for "usable" data. Those omissions leave the headline number tied to a single, vendor-selected benchmark rather than a published, repeatable procedure against which competing systems could be compared.

What does this change for inspection workflows?

For users running inspection, dimensional metrology, or reverse engineering on cast, machined, or coated parts, the practical question is whether automatic tuning replaces the operator skill the previous generation of scanners required. Faro Creaform's pipeline reduces manual parameter adjustment at the cost of trusting proprietary algorithms to interpret ambiguous returns — a tradeoff any quality-management system still needs to validate on its own reference parts.

Until a published, repeatable benchmark lands, any shop evaluating the update should reproduce the 60 percent claim on its own difficult parts before committing the workflow to scan-time estimates.