I use the Revopoint Metro Y Pro, and I am absolutely fed up with a simple problem that ruins the entire workflow: after scanning the parts, I want to merge the scans into a single model. Like any reasonable person, I try to do this automatically based on geometric features—since my models are car interior components, manually identifying matching points is difficult.But absolutely every time, it causes an incredible number of problems—it reaches the point of absurdity where it can’t even stitch together two scans of the same area just because the focus on details differs. How does this even work? What is the point of this function if it can’t handle the task? Why is it that when I use competitor software to scan the inner and outer parts of, say, a car taillight, it automatically stitches them together correctly, establishing both the thickness and the edges?And with Revo MetroX, I have to take six scans, each overlapping the previous one—which isn’t always convenient or even possible. Even when I do manage that, the software doesn’t just stitch them together; instead, it starts flipping them, joining the outer wall to the inner one, or misaligning them.
I can understand your frustration.
Did you try to manually assemble them ? Can you share the file ?
auto alignment needs a good amount of overlap, i think 20-30% between the two scans to work properly. Just do manual alignment for this part. your points dont have to be spot on, just close.
An alternative is to use Cloud Compare. But you will be selecting points manually in that software to align.
It looks like you scanned one side, then flipped it and scanned the other. You typically want 1 more scan to bridge the two scans and tie them together.
I scanned the front face with high quality, then scanned the side along with the front—the idea being to subsequently add the back and the other side to the assembly. I tried using reference points, but the dimensions didn’t match up during reverse engineering because it’s difficult to find three points on this object that are exactly identical.
I tried assembling it six times, but nothing worked. I deleted the assembled file, leaving only the STL with the issue of overlapping scans—this is visible as voids inside the model when I slice it to draw a cross-section, and naturally, the accuracy drops drastically.
If you are reverse engineering anyways, just give yourself some points you can use to align. a small dot of clay or even a paint mark should be enough to show up in your scan and help you align.
It is a little difficult to follow your work flow. Lets start with a picture of the object you are scanning. Then what each scan looks like in its fused state.
Remember, if it was easy, everyone would do it!
I realize it’s not easy, but I do have some scanning experience—albeit limited. I used to own a competitor Raptor, and I can say its stitching function worked perfectly; it was as if the software knew exactly what the object was, and just 2–3 scans were enough. Here, however, I’m dealing with essentially identical surfaces with about 80% overlap, yet it fails to stitch them together. The part I’m scanning is a section of the car interior near the windshield.
HI
Like @Nickluvin said, do yourself a favour and place few easy to identify points or similar on the object itself.
At least 3 of them should be visible at the same time when manually aligning two point clouds. You can delete these areas afterwards and perform hole closing if you want a “clean” mesh. If you reverse engineer the part these points don’t matter.
juste add some markers on the part. scan 3 times. great success !
are you trying to do outside and then inside of the part? Aligning for thin parts can be tricky as so much of it comes down to how you clean them up. You almost always have to do manual mode then because otherwise it just tries to fit a big section of surface to surface and usually gets it backwards.
It looks like there is a lot of geometry at the end you could use to help align. I threw some red dots on areas I’d use.
There used to be an option to align via markers in one version. Not sure why they ever got rid of that one.
Hi @Serhii1
There is no way any software will merge that scans without manual marker point being placed.
In this case you can use mini stickers that can give you some reference points , if the object don’t have any features you can even add some little tape and use it later as reference , I printed myself very mini cubes and stick it to the surface for reference points .
It would actually be great to add the ability to adjust scans relative to one another during manual alignment simply by moving them with the arrow keys; in my view, that’s more practical than picking points and hoping everything lines up—plus, it eliminates the need for “life hacks” involving clay or other DIY methods.
Again, check out CloudCompare. You can bring your scans in and rough align them and then finely align them. Its a free tool. Its not as intuitive as some softwares, but it is powerful.
Hi @Serhii1
Picking points will ensure the proper alignment , it is the same with all point cloud editing software .
Manual alignment will give you just rough alignment
As Nick already stated , try Cloud Compare, it has a great tool and allow you manually align your point clouds.