As someone who works with 3D printing, I have always had an interest in 3D scanning. So when Creality offered to send me their CR-Scan Lizard 3D scanner for review, I jumped on the chance. Below you will find my thoughts about the Creality CR-Scan Lizard and a showcase of some of my scans.
Disclaimer: Creality provided me with this scanner free of charge, but this review is as unbiased as possible. Creality did not pay for this review and these are entirely my own thoughts.
I’ve been working with 3D printers for more than a decade and consider myself something of an expert on the subject. But I have remained skeptical of both the utility and the capability of consumer 3D scanners.
My skepticism of the utility came from the fact that 3D scanners produce mesh models, which have limited use to those of us who prioritize parametric CAD modeling. Because I need parts to fit together exactly and need to quickly modify dimensions, it makes more sense to me to use calipers and other tools to gather exact measurements and then use those measurements to build CAD models.
My concerns about capability came from seeing the results posted by others over the years. Those forced me to question the accuracy and precision of 3D scanning.
But 3D scanning technology has come a long way over the years and I wanted to know if it is now useful. The Creality CR-Scan Lizard gave me the opportunity to test a consumer 3D scanner to find out if it is worthwhile for makers like myself.
The CR-Scan Lizard is small consumer 3D scanner about the size of a USB power bank. Creality claims that it can achieve up to 0.05mm accuracy and doesn’t require any markers for scanning. The basic model doesn’t include any accessories and is meant for handheld scanning. Stepping up, the Premium Combo includes a small tripod and rotating platform for table scanning. At the top, the Luxury Combo adds a light box and tripod setup for smartphones, which lets users add color textures to their scans.
Creality sent me the CR-Scan Lizard Luxury Combo. I tested both handheld and table scanning (both geometry and texture), but did not test the color texture feature enabled by smartphone pairing.
The CR-Scan Lizard scans objects using a projector and a pair of cameras. Like any other camera, they capture the light reflected off of the object and therefore scanning performance is a result of the reflectivity of the object. In ideal circumstances, this works well. But conditions aren’t always ideal.
In my testing, I found that the “perfect” object to scan is: about the size of a closed fist, a medium color (like blue), mostly convex, and with an opaque surface that isn’t too shiny.
The further you stray from those conditions, the more difficult it is to get a good scan. For example, Creality claims that the CR-Scan Lizard can handle black objects. It does, but only some of the time. Some black objects are essentially invisible to the CR-Scan lizard, as are transparent and very shiny objects.
Size matters, too. If something is too small (like a ring), the Lizard will struggle to pick up enough content to generate a mesh model. If something is too big, it will require repositioning and multiple scans to capture the full object.
Concave details are also difficult for the Lizard to scan. It works fine with relatively shallow features, but not with deep features or anything too detailed. Finally, it struggles with thin features like lattice structures.
Creality says that the Lizard has “up to 0.05mm accuracy,” but I think the word they were looking for was “precision.” On top of that, “up to” is doing a lot of heavy lifting.
In my testing, the practical accuracy wasn’t anywhere close to 0.05mm. This is easy to test by scanning something like a ping pong ball. It should have a smooth surface with very little variation and an accurate scan would reflect that. But when I performed this test, the surface was rippled and lumpy—far exceeding 0.05mm (which is about half the thickness of a strand of hair). You can see the same effect on all prints.
This limits the utility for mechanical engineering work, because one cannot rely on the scans to accurately represent the model. If you need to scan something so you can design a mating part with a tight tolerance fit, you may be disappointed.
A scanner is only as good as the software it works with, which is why it is a good thing that CR Studio is pretty decent. The more scans you perform, the better your final result. But you need to align those scans to process the final mesh model.
Fortunately, CR Studio has two ways to do this. First, it can attempt to automatically align scans. That works well when the object as a distinct shape. But it will usually fail if the object is symmetrical or uniform (like a cube, cylinder, or sphere).
As a backup, you can manually align scans and this works very well. Just select two scans and click on three matching points. You don’t have to get it exactly right, just in the same neighborhood. The software will then use those points to orient the scans and align them. That was almost always successful in my tests.
This was probably the most frustrating aspect of scanning with the Lizard for me. As it scans, it captures frame after frame. To turn those frames into a point cloud (and later a mesh), the software has to know the orientation of each frame relative to the previous frame. It does that by looking for matching geometry. If it doesn’t see any matching geometry, it loses tracking.
On a table scan, lost tracking will almost always result in a failed scan. Repeating the scan doesn’t necessarily solve the problem, as some objects that don’t have distinct features (or that have a very narrow side) will always lose tracking. One would think that CR Studio would use the table platform (with its special geometry patterns and printed symbols) to maintain tracking regardless of the scanned object geometry, but it doesn’t and I was really annoyed by that. This means that some objects simply can’t be scanned.
In my testing, this usually presented as “skips” when CR Studio would think it had regained tracking but actually hadn’t. That would produce an unusable scan with the features in the wrong places.
The tracking situation with handheld scanning is a little better, because you can move back to the last successful position and regain tracking. It is still frustrating, because it is sensitive and easy to lose tracking. But you do have some recourse.
In general, you want to table scan anything small enough to fit and handheld scan the rest. Table scans are quick and easy to capture, but handheld scans take longer and very careful movement (your arm will get sore).
Both can produce some very nice results and both can fail completely in conditions that aren’t ideal. Table scans are easier, because you can perform multiple scans in a row and simply reorient the part. You also have more control over lighting. Handheld scans require you to very carefully move the scanner around the object at the exact distance needed (a fairly small window), which is more difficult than it sounds.
Normal scans only pick up physical geometry, but you do have the option to perform texture scans. Those use the cameras to also capture images, which CR studio will use to texture the model after processing the mesh.
In my testing, this worked well anytime that the mesh model itself scanned properly. If your scanned object has graphics, this lets you add them to the model.
In some situations, it seems like this can also help with tracking as CR studio seems to take visuals into account. I’m not certain of this, but it did seem to help in some cases.
However, those texture scans are grayscale. To get color textures, you’ll need to also use your phone. I didn’t test this feature and so I can’t confirm whether or not it works well.
I really wanted to like the Creality CR-Scan Lizard. If it performed well, it would be useful for many of my projects. Unfortunately, I found the experience to be disappointing.
Some objects won’t scan properly at all, while many others don’t scan well. Even those that do scan well are only suitable for low-tolerance applications, as the geometry isn’t accurate enough for anything that requires high tolerances.
That said, when the conditions are ideal the Lizard is easy to use and captures some nice looking scans. The CR Studio software does most of the hard work, so users don’t need expertise to take advantage of the Lizard.
But if a friend asked me if they should buy the CR-Scan Lizard, I would probably tell them not to. You can get similar quality scans using an iPhone Pro with free/cheap apps, which makes it hard to justify the cost of the Lizard. If the Lizard were more reliable or produced better scans, it might be worth the cost. Unfortunately, that wasn’t the case with my experience.