MetroX global marker scanning and marker detection issues!

[mr_cg]

It has been mentioned here in the forum and on facebook also by many other MetroX users that marker detection is terribly bad on MetroX. There has been many promisses to improve marker detection but since after months … Nothing! The global marker registration of larger environments is simply not working at all. When Revopoint is improving this?

Some screenshots from recent try to do global marker registration of 3d scanning frame attached. MetroX was calibrated before and i tried several time but never (never!) had any success. Not even that the marker recognizion in bad, also quite often the markers are simply registered wrongly making whole global marker registration useless. It simply cannot be user fault or laptop (i9-13900HX, 32GB RAM, RTX4070, SSD) performance issue because with exactly same equipment and RaptorX it’s working without any issues. I also never had any such issues with wrong marker alignment using the RaptorX.

The level of frustration with MetroX is already on quite high level and Revopoint should take this issues seriously and provide solution plan.

Please escalate this topic to dev team and provide solution plan!

[PUTV]

The markets need to be separated and in unique pattern , if they are too close you will get error and that is w exactly what is happening here .

Normally they need to be at 3 cm /30 mm distance from each other , unless they are on larger geometric figures where they are not visible all at the same time at a specific angle.

If the markets are too closer, they will jump and confuse the sensors.

Just test it out , you will see.

BTW improved version will be release soon , but it still requires the markers to be at the proper distances from each other.

[mr_cg]

Everything is prepared as it should. Normal marker dices have very short distance between die markers and this should definetly not cause any issue. As said, Raptor does not have any issues when scanning marker environment. Only MetroX has and this is purely because of the SW. After some months on the market this issue should be fixed. Also many other users are complaining about the exact same issue.

I want to get propper ticket ID from Revopoint and answer from dev team about solution plan.

[PUTV]

Please zip your project folder including the global market setup and send it to customer@revopoint3d.com for technical evaluation .

You can include the link to this thread if you like it or add the all images to the email.

[mr_cg]

With new RevoScan version 5.6.5 it was - first time ever - possible to register 3d frame. But it was a nightmare to get this done and several tries were needed. The small fov of the MetroX seem to be the reason why marker detection and registration is so weak.

But even with global marker registration the scanning was finally always ending up with alignment error. Really frustrating and time to give up!

[MonsterMaxx]

Probably going to get told your markers are too close together. I only had a few that were less than 3cm and it caused issues. Looks like you have a LOT that are closer than 3cm.

[mr_cg]

All this is not a problem for the Raptor. Not at all!
Only the MetroX is not able to handle such real live marker scenario.

[TheBoatScans]

I’m going to be honest, I don’t know that “it should work with this marker setup and I will not adjust it” is a valid argument in and of itself. This is not about the ability to scan an object, we’re talking about an arbitrary pile of markers that you say should work just because it works with another scanner. It may in fact be a limitation in comparison, but it’s not necessarily one that makes a difference in the actual scanning process.

I will say, if you are actually interested in working with the MetroX, is that you should learn from the global marker creation step. If you have trouble creating the global marker file, you’re going to have trouble during the live scan as the scanner is seeing only a small subset of that and trying to match it.

[mr_cg]

I found now a big implementation issue in RevoScan v5.6.5 which is one of the main reason for very bad marker scanning and marker detecion issues. Also causing to wrong alignments. This is existing since day one.

First i was measing the fov for the marker detection and realized that from MetroX it’s extremely small (only 10-15cm). Just for comparison the fov measured with RaptorX is 50cm. The fov mentioned in this case is the distance in which the markers are recognized by the scanner.

And after some more thinkering i realized that the scanning distance is not only limiting the distance in which point clouds are getting collected, it is also limiting the marker detection. To me this is bad implementation and should be changed. It only makes sense to limit the distance for point cloud detection but not for the marker detetion. I checked implementation in CR-Scan and here the scanning distance has no effect on marker detection, only point cloud capturing. This is how it should be implemented.

Because of this very small fov for the marker detetion the global marker scan experience is very bad and also wrong alignment are happening because only small amount of marker are detected. Larger fov for maker detection will significant increase number of captured markers which will reduce probability of wrong alignments.
After changing the default scanning distance (210-350mm) to maximum distance (200-400) the global marker detection experience is very much improved. Also 3d scan using the 3d scanning frame from top and bottom of the object was - first time - successful.

Revopoint should immediately change the implementation of scanning distance setting and not limit the marker detection, only the point cloud capturing. I believe this bad implementation is the culprit of marker detection and wrong alignment issues mentioned by so many users including myself.

I hope Revopoint is taking this feedback seriously and provide update in RevoScan very soon.
This will help to improve the scanning experience massively.

[GuyH77]

Have you seen this video? It seems the MetroX is limited and shows why so many of us are frustrated with it. Thank you for raising these issues.

[mr_cg]

I know this video. And the issue with global marker registration (24:20) failed exactly because of the reason mentioned here in this thread. The scanning distance is not set to maximum and therefore the marker at the back side of the controller are not recognized (even it’s visible, see screenshot) which is needed for successful and robust alignment. This is implementation issue because scanning distance should not limit the marker detection distance.

[SphaeroX]

Someone on Discord once did some reverse engineering with the Metro X 3D scanner. The main issue turned out to be that the camera has a fixed focus instead of an autofocus, which is crucial for proper calibration.

Other manufacturers handle this differently—their RGB cameras have autofocus, allowing them to detect markers that are farther away more accurately. So, to put it simply, it’s a hardware flaw and not something that can be fixed through software. And even though people often say you just need to position your markers differently and so on: yes, that might help a little but in my opinion, it’s just adapting to a hardware limitation that shouldn’t be there in the first place.

In short: unfortunately you have to live with it (yes, it annoys me too)

[mr_cg]

You definetely do not need to have autofocus.
fov is only about aperture x focal length.
All metrology grade structured light 3d scanner have fixed lenses and they even not allow to change the focus any more once they are calibrated.
The higher the aperture the larger the fov but the light is getting lower which require more sensitive image sensor.

The RGB camera is usually only used for texture where no high precision is needed.

[PUTV]

@SphaeroX

The main issue turned out to be that the camera has a fixed focus instead of an autofocus, which is crucial for proper calibration.

If you had auto focus you will not capture anything . The sensors need to be calibrated to its RGB camera att a fixed focal distance the " Excellent " distance and nothing else as there is only one distance that allow the accuracy perfectly at that distance .

[PUTV]

@GuyH77

It seems the MetroX is limited and shows why so many of us are frustrated with it. Thank you for raising these issues.

This video is the biggest BS I saw in a while , he don’t have idea what he doing with it , so wrong on so many level

But what you expected when competitor. Pat for it .. it can’t be good
For that reason the video is also removed , don’t follow his advices as you will get not far .

[PUTV]

@mr_cg

You definetely do not need to have autofocus.
fov is only about aperture x focal length.
All metrology grade structured light 3d scanner have fixed lenses and they even not allow to change the focus any more once they are calibrated.
The higher the aperture the larger the fov but the light is getting lower which require more sensitive image sensor.

The RGB camera is usually only used for texture where no high precision is needed.

That is very well said , the RGB camera is used for calibration as well to calculate the digital adjusting between the sensors and their positions but only digitally since there are not moving parts in the lenses.

[GuyH77]

All I can say is I see how easily the other scanner identifies markers and how painful I find the MetroX when it comes to markers, its sensitivity to settings, distance, spacing etc. I know you’ll tell me it’s an art and it’s down to me but in reality, if scanning is ever to become more useful to the average person then it has to be easier to use. I know it seems like I’m moaning and lacking skills but I just want to enjoy using a tool. I’m not new to the principles of scanning, CAD etc.

Rather than remove videos, why not educate us and explain why it’s BS. Censorship here wont help much as anyone looking for guidance on the MetroX will check out youtube.

[PUTV]

@GuyH77

Rather than remove videos, why not educate us and explain why it’s BS. Censorship here wont help much as anyone looking for guidance on the MetroX will check out youtube.

Sorry about that , that’s no my rules
Negative propaganda is sadly not allowed in the forum . It was paid propaganda to bashing the product on purpose.

@GuyH77

why not educate us and explain why it’s BS.

Why BS ? Because if someone fusing an detailed object at 0.2 mm and the other company’s on higher quality and telling that is the final quality , it misleading , you may not see in that video what I can see because I know exactly the limits of that scanner and I can see if someone doing it on purpose or not .

Each scanner no matter who make it has own rules to be followed , just because other do things differently don’t means MetroX has to follow the same rules .

The team tested marker scanning last week and I did my private test again and there were no issues whatsoever when using recommended system. It worked just the way it should .

Remember MetroX frame fov is small so the markers distance is also smaller and should be at 3 cm when used on flatter surfaces , when used on geometrical surfaces they can be closer since only one angle is visible at the same time .

MetroX laser mode will not fire unless it detect at least 3 markers at the same time in the space .

Laser scanning is time consuming since it capture only 800K points per frame .
That why it is also most expensive scanning service .
Just because you bought the most expensive camera it don’t will make of you the photographer of the year , everything needs time and practice to learn your tool.

Watching videos and reading about will not make you great in scanning , practice will .

You know even after having 30 years of practice and experiences ,it still take me at least 2-3 months of practice with a new scanner to master it optimal the way it works for me.

I recommend you start using Full Field mode and get familiar with it first , it will provide faster better results and then you can switch to laser and continue your workflow process .

[SphaeroX]

I would like to summarize the discussion about marker detection with the Metro X from a technical perspective, as the issues remain unresolved even after several updates. The system continues to struggle with poor marker recognition, which, in my opinion, is primarily due to two technical shortcomings: first, the fixed focus of the RGB camera, and second, the tight coupling of marker detection to the scanning distance window.

The Metro X uses its RGB camera for marker detection, which is set to a fixed focal distance. While this may have some advantages for calibration, it severely limits the depth of field. As soon as the markers are positioned outside this narrow focus range, which can happen quickly with larger objects or more complex setups, they become blurry and can no longer be reliably recognized by the software. The marker algorithms use standard techniques such as edge detection, thresholding and perspective correction. All of these processing steps require sufficiently sharp marker images. If the sharpness is lacking, the error rate increases, and markers are either misidentified or not detected at all.

The structured bluelight projector operates independently of this and is used solely for depth measurement, but it does not contribute to marker detection. The two data channels, RGB for marker detection and monochrome for depth, are synchronized afterwards. However, the problem remains: if a marker is blurry in the RGB image, even the best depth data will not help, as the spatial assignment of markers will be missing or incorrect.

Another issue is that the software strictly ties marker detection to the same distance window used for depth acquisition. This results in markers outside a range of about 10 to 15 cm often not being detected at all, even though they might still be visible in the image if the camera used a variable focus. This limitation makes it particularly difficult to use the scanner on larger objects or more complex marker setups in practice.

Instead of addressing these constructive criticisms, critical posts in the forum are unfortunately often deleted or dismissed as “negative propaganda.” This approach hinders further development much more than it protects the product’s reputation. It would be far more productive to take such feedback seriously and use user input to improve firmware and detection algorithms.

With all due respect, it is noticeable that the only party consistently discrediting other products is @PUTV itself for example, in the case of the OpenScan, where it was claimed that the Metro X would deliver much better results and should provide its own scan comparisons, which never materialized. I actually like Revopoint as a company, but if all criticism is simply dismissed and only new products are released instead of addressing user feedback, this approach will ultimately not lead to long-term success.

From a technical standpoint, it is necessary to decouple marker detection from the rigid scanning distance setting and to design the camera optics so that markers can be reliably detected over greater distances. Alternatively, implementing a variable focus or even autofocus would be a practical solution (in the future). Only in this way can the Metro X become truly competitive in marker registration and offer users the flexibility that is essential for many professional applications…

[TheBoatScans]

@SphaeroX

The Metro X uses its RGB camera for marker detection

This is demonstrably untrue. Even in autoturntable mode it does not use the RGB camera for marker detection. Covering up the RGB camera doesn’t affect laser scanning.

I do agree that the scanning distance parameter affects marker detection is certainly an issue to be addressed.