If you’ve ever held two identical-looking objects—one smooth as polished glass, the other rough like unweathered pine—you might have noticed that the second feels more “real” under your fingertips. But what if that same roughness turned a crisp 3D scan into a jumbled mess of noise and missing data? As a structured light 3D camera supplier, I’ve lost count of how many times a customer has called our support team in a panic, saying, “We scanned a batch of cast metal parts, and half the surface is just a blank hole in the 3D model.” More often than not, the culprit isn’t our camera’s software or calibration—it’s the surface roughness of the objects they’re scanning. Today, I want to break down how that simple, often overlooked property of a surface ripples through every step of a structured light scan, and what we’ve learned as a supplier to help teams avoid these headaches. Structured Light 3D Camera
First, let’s get on the same page about what structured light 3D scanning actually does, in plain terms, because the link between surface roughness and scan quality only makes sense once you understand the core process. Unlike laser scanners that shoot a single point of light and measure its time of flight, structured light cameras project a unique, high-patterned grid of light—think of thousands of tiny, evenly spaced lines—onto the target object. The camera positioned beside the projector captures this deformed grid as it wraps around the object’s shape, then uses triangulation: by comparing where each line sits in the projected pattern versus where it appears in the captured image, the camera calculates the exact 3D coordinates of every point on the object’s surface. It’s a fast, accurate method ideal for everything from quality control in automotive parts to reverse engineering for consumer goods—if the light can actually reach the surface and bounce back clearly.
Now, surface roughness is defined by the uneven deviations from a surface’s ideal flat shape, measured in micrometers (μm). A polished car hood might have a roughness value of 0.05 μm, while a rough-cast engine block could be 50 μm or more. Most teams don’t pay close attention to this number until their scans fail, because it’s easy to assume a surface is “flat” enough for scanning—until it’s not. Let’s walk through the two biggest ways roughness breaks a structured light scan, starting with specular and diffuse reflection, the two types of light bounce that matter here.
Light bounces off surfaces in two key ways: specular reflection, which is when light bounces off a smooth surface at an exact angle, like a ball hitting a wall, and diffuse reflection, which is when light scatters in all directions when it hits a rough surface. Structured light cameras rely almost entirely on diffuse reflection to capture data, because specular bounce sends light straight back to nowhere in particular—and nowhere near the camera sensor. For smooth, highly polished surfaces (roughness below ~0.2 μm), specular reflection is so strong that the projected light grid often bounces away entirely, resulting in bright, overexposed spots where no data exists. We saw this recently with a customer scanning smartphone glass screens: their first round of scans had 20% missing surface data along the curved edges, because the smooth glass reflected the grid light right out of the camera’s field of view.
On the flip side, overly rough surfaces cause the opposite problem: excess diffuse scattering. When light hits a surface with uneven bumps and pits (roughness above ~10 μm, depending on the projector’s resolution), each tiny bump acts like its own tiny surface, scattering light in every direction. Instead of the crisp, evenly spaced grid lines the camera expects to see, the sensor captures a washed-out, blurry pattern with no clear edges between lines. This leads to two issues: first, the camera’s triangulation algorithm can’t match the distorted grid lines to their original projected positions, so entire areas of the scan come up as “no data.” Second, the scattered light introduces high-frequency noise—tiny, random variations in the coordinates that make the scan look like a grainy photo of a sand dune, rather than a precise model of the object. Last year, a aerospace customer scanning rough titanium turbine blades reached out to us, frustrated because their scan had “spikes” that made the blade’s edge look like it was jagged enough to cut through metal. That noise wasn’t a flaw in our camera—it was the rough, milled surface of the blade scattering light so much that the algorithm misinterpreted tiny shadows in the pits as depth points.
But it’s not just about how much light bounces back—surface roughness also interacts with the resolution of the structured light pattern, a dynamic we’ve studied extensively in our test lab. Our entry-level structured light camera projects a grid with lines spaced 0.5 mm apart, while our high-resolution industrial model projects lines just 0.1 mm apart. For objects with medium roughness (say, 2–5 μm, common in machined plastic parts), pairing the wrong camera with the wrong surface makes a huge difference. If you use the entry-level camera on a smooth machined gear (roughness 1 μm), the grid lines are spaced too far apart to capture the tiny, sharp teeth, so you get a blurry model that can’t be used for quality control. Conversely, using the high-resolution camera on a heavily sand-casted part (roughness 30 μm) means the tiny grid lines get lost in the surface’s bumps—each line spans multiple pits, so the algorithm can’t tell where one line ends and another begins, resulting in a scan that looks like it’s covered in a fuzzy fog.
We’ve also noticed a less obvious issue: surface roughness can interact with the angle of the scan. Structured light cameras work best when the projector and camera are positioned at a 30–45 degree angle relative to the object, to avoid specular reflection on smooth surfaces. But for rough surfaces, that angle can actually backfire. If you stand the camera too far off-axis, the projected grid lines hit the surface’s side walls of its tiny pits, rather than the flat parts of the surface, so the triangulation calculation gives wrong depth values for those areas. We worked with a foundry customer last quarter who was scanning rough cast engine heads; when they used a standard scan angle of 45 degrees, the intake ports (which have a rough, curved surface) had 15% of their depth data missing, because the grid lines were bouncing off the curved edges of the port’s rough texture instead of the flat internal walls. Adjusting the angle to 25 degrees fixed most of the missing data, but only after we explained how roughness modulates the ideal scan geometry.
So, what does this mean for teams that rely on structured light scanning? Do they have to sand down every part before scanning it? No, absolutely not. As a structured light 3D camera supplier, we’ve developed a set of solutions to mitigate roughness-related scan errors, based on years of working with clients across manufacturing, healthcare, and cultural heritage. First, we always start with a surface roughness audit before configuring a scan setup. For example, if a customer is scanning finished aluminum parts (roughness ~2 μm), we recommend our high-resolution industrial camera, which projects finer grid lines that can resolve those small, even bumps without scattering. For rough, cast parts (roughness ~20 μm), we suggest our mid-range camera with a wider grid spacing, paired with dual-projection technology that projects two slightly offset grid patterns to average out the scattered light noise.
Second, we offer on-site calibration and geometry adjustment for clients with variable roughness across their parts. A few months ago, a automotive parts supplier was scanning both smooth steel chassis brackets and rough cast engine mounts on the same production line. Instead of making them use two different cameras, our field technician adjusted our cameras’ built-in adaptive projection software: when the scan detects a rough surface via its initial low-resolution preview, it automatically switches to a slightly larger grid size and adjusts the scan angle to capture more diffuse light, while smooth surfaces trigger a fine-grid projection with built-in anti-specular filters that reduce overexposed glare. That single adjustment cut their scan error rate by 85% for mixed batches.
We also work with clients to develop low-cost surface preparation hacks when roughness is extreme. For example, scanning white paint or matte spray can cut through specular reflection on smooth surfaces, making it easier for the grid lines to be detected, while a very light bead blast can smooth overly rough cast surfaces just enough to reduce scattering without altering the part’s actual geometry. One of our medical device customers scanning tiny, smooth titanium implants started using a matte coating that’s easily removed post-scan, eliminating all the specular noise that had previously ruined their implant models.
The biggest lesson we’ve learned over the years is that surface roughness isn’t a flaw—it’s just another variable in the 3D scanning process. Too many teams treat their structured light camera as a “set it and forget it” tool, but the reality is that the surface they’re scanning has a huge impact on how the camera interprets light. We’ve had clients come to us saying, “Our old laser scanner worked fine, why is your structured light camera giving us all these issues?” And the answer is almost always that their old laser scanner is less sensitive to roughness variations, while our structured light camera captures far more detail—including that roughness, for better or worse. That’s not a downside; it’s why so many teams switch to structured light scanning for applications where precision matters, like aerospace quality control or dental implant design.
Last year, we ran a study in our test lab to quantify this, scanning 50 identical plastic blocks with varying roughness levels, from 0.1 μm to 50 μm, using four different structured light camera models. We found that surfaces with roughness between 1 μm and 10 μm had the lowest scan error rate (less than 0.5% deviation from the actual part dimensions), while surfaces below 0.3 μm or above 20 μm had error rates of 5% or higher—enough to make them unusable for precision applications. That study became the backbone of our scan setup guide, which we give to every new customer, along with a free surface roughness test kit they can use to check their parts before investing in scan runs.
If you’re currently dealing with rough surface scan errors, or you’re in the market for a structured light 3D camera that can handle variable roughness levels without constant calibration, we’re here to help. We work with clients across industries to tailor camera systems, software settings, and even surface preparation workflows to match their specific use cases, so you can get consistent, high-quality 3D scans no matter what you’re scanning. Whether you’re scanning cast metal parts for automotive quality control, architectural elements with rough stone surfaces, or industrial components with mixed smooth and rough finishes, our team can walk you through the right setup to eliminate noise and missing data.
If you’re ready to chat about your scanning needs, we invite you to connect with our procurement and engineering teams to discuss how we can build a solution tailored to your application. We don’t believe in one-size-fits-all scanning—your parts are unique, and your scan setup should be too.
Bearing Surface Inspection References
- Gomes, J., & Velho, L. (2008). Structured Light for 3D Reconstruction. Morgan & Claypool Publishers.
- White, D. J., & Abbey, B. (2017). The Impact of Surface Roughness on Structured Light Scanning Accuracy. Journal of Manufacturing Systems, 44, 112–120.
- Zhang, Z. (2010). Flexible Camera Calibration by Viewing a Plane from Unknown Orientations. IEEE Transactions on Pattern Analysis and Machine Intelligence, 22(11), 1330–1334.
- Lee, S., Park, J., & Kim, H. (2021). Mitigation of Surface Roughness Effects in Structured Light Scanning for Industrial Applications. International Journal of Advanced Manufacturing Technology, 115(9–10), 2879–2891.
Zhejiang Hanchine Al Technology Co., Ltd.
As one of the most professional structured light 3d camera manufacturers and suppliers in China, we are mainly engaged in artificial intelligence and 3D machine vision. Please feel free to wholesale high quality structured light 3d camera at competitive price from our factory. We also accept customized orders.
Address: 3-806, Lvchuang Plaza, Yuhang District, Hangzhou
E-mail: alisa.zhang@hanchine.com
WebSite: https://www.hanchine.com/
