Is there a deeper explanation of this mode?
I understand that:
My question is - how does it differ from the normal marker mode?
It just seems to have split a single step process to 3 separate steps.
I just don’t understand what the benefit is.
Here is Jane’s explanation in another thread:
imagine this: you use the magic mat or some other surface with the markers. they have such a placement , where optimal tracking of markers is always possible (like 3-5 markers always in FOV of the scanner), because they aren’t obstructed by some object on/over them. As such, you have a greater precision and less stitching (volumetric) errors when scanning some object on such marker surface.
now, scanning an object on such surface starting on part with good marker visibility the scanner will of course always know its position. even going to areas with lesser visible markers ,because some are obstructed by the object itself, it will keep good tracking because of overlayed prescanned marker position.
I am sure , there will be a demonstration video by revopoint and/or pop 3 plus tester anytime got better visualisation. It would be great see a direct comparison of scanning the same object with “normal” and global marker tracking mode. @Revopoint-Jane
cheers
Hi ivan, thank you! Sure, we will see what can we do.
The idea behind marker pre-scanning and “global marker” file creation is to separate the functions and jobs of the scanner into separate passes, so that each of these functions can be done better and with higher precision and more confidence on their own, and by that you free up resources and eliminate confusion.
This is not a new thing, it has been aroud for a good couple of years in other scanner brands software as well.
Technically whats happening:
First you scan the markers alone, which is done by an algorythm, that is different to the one that is responsible for collecting the point cloud. You only scan these marker points and nothing else. This is a small amount of data that is gathered with pretty high confidence, because of the easily distinguishable high contrast spots. The software also knows that they are actual markers, so if theres a small slip between frames, the software can easily align them correctly. If theres a false trigger for marker, those can be filtered out accurately. These points can be matched between frames at a much higher precision compared to the feature tracking mode, where the software is trying to match pointcluds to point clouds between frames. Also the marker point collecting can be done with a larger field of view, so you gain speed here too.
After you collected all the points, the software will fine align their relative positions without any other influence.
At this point you have a skeleton of your scan where you will build the actual point cloud data onto, the muscles if you will…
The second pass is where the actual scanning will happen. you already have the markers detected and stored, the software will no longer look for new ones, the marker framework is fixed now, the software will compare what it sees and what is stored already and will match up the position of your newly scanned frames accordingly. This is a much lighter on your resources than trying to do the two functions simultaneously and also filters alot of possible errors.
In your workflow, you can take advantage of how this function works, you will find clever uses for it online if you look it up.
My personal favourite is this:
Because the marker scanning does not generate much data, you can pre-scan the markers on very large areas at once, that would need to be separated up to smaller sections when you scan the point cloud.
You can basically create a large skeleton for your scan project out of the marker scan, export it out and import this marker file back into each smaller section scan you plan on doing. Your scans will magically align correctly, without any manual alignment. That is because you built each point clouds on the same marker skeleton thus they are aligned right from the beginning.
That is not how it works Ivan , in theory but not in practice.
We skd for global markers to ore scan the space so we can use objects and scan them without worry about tracking … Without the need fo place a new markers on that object … But that do not works you still need markers so double job .
exactly , but it still look for markers so I am lost here .
How this function works, is not exactly finalized in the revoscan software, I was talking general terms.
I know you talked in general term as it is the way it should works
BTW I got the public version 549 , it works now
First pre scanning markers only , I used the turntable markers
Then place your object on top of it and use the Cloud Scan and done , but you still need the exact markers to be there , can’t be removed in the second scan , and that is not what I was hoping for . So if you have your markers on your object they need to be visible for tracking … That what I don’t get it , it should scan it without the second time since it got the cage already . Unless the purpose is only for better stability while tracking the markers and not as we imagine it to be or how it should works .
But honestly I would get the same results scanning just once using the markers in marker mode , so yes it is only for stability .
If you scanning with global markers and use the markers on the turntable (flat) it speed up the work as you don’t have to rescan the global markers each time what js time saving . But if the markers are on an object , it is one ride pony .
Yes, the markers need to be present when creating the pointcloud scan because they are used to position the newly collected pointcloud frames to the
already stored skeleton. If the markers werent there, then the software would have no idea where to place the scanned frames relative to the already stored “scene”. You will notice an increased speed and stability in this mode and it is also very much possible to gain accuracy as well.
Yes, people store the marker files of their turntables and other scanning jigs and fixtures. If you have good jigs and fixtures, most of the cases you dont have to apply any markers on the object itself. You just load in the file that corresponds to which jig/turntable/fixture youre using for the job, throw the object on in and go straight to pointcloud scanning step. Saves time.
Well I made more tests and I am pleased with the results , the RS5 public version is definitely better , including higher fusing and meshing option so all results are better including the cloud scanning .
Ok I think I’m getting it.
The markers act somewhat like measured GPS trackers in aerial photogrammetry.
So the main benefit i’d see here if it worked well would be:
This is actually similar to what was once suggested for Miraco:
Let us do a rough scan with far mode so that we get the entire space / object with its dimensions - then let us rescan in Near mode using the rough scan for orientation.
I would like to add that if you work with previously stored marker files, you need to have your turntables and other jigs to be really really rigid and not bend or deflect under the typical loads you put on them. If they do flex under the weight of your scan subject, your markers will move out of position thus generating inaccuracy in your scan. I dont know how much this affects the revopoint scans, but based on what I know of how they work, I believe it does.
So I just want to say here that the revopoint standard turntable or even maybe the dual axis one might need to have their tops replaced with a rigid material on which the markers are applied. Glass or 3mm and thicker steel sheetmetal is cheap and good for this purpose.
That what I just tried , scanned global markers in regular FOV mode and then scanned actually object in zoom x2 and it worked fine . I remember your suggestions from early …
And yes you can pause and move the object in any position and continue scanning .
Would be possible, but will cross compatibility between scanners get implemented or would they use the same type of files and just work? cannot know.
Thats right!
I noticed it as well … I have old dual axis the markers are on the hard surface so not bending
Cross compatibility between scanners should be possible - at the end of the day what comes out from the first scan is just a cloud of marker points in space - don’t see why there would be a requirement for it to be created with the same scanner as the second scan.
As long as both scacnners are well calibrated and create a 3D scan with accurate dimensions - it should work.
PS. @PUTV - aren’t you supposed to be asleep right now?
Thanks for clarifying, Cath! I would look forward to a video by you on that subject. 
Also wondering which markers can be used for that mode (diameter), and if they can be combined, like bigger and smaller sized markers.
Can you please test and confirm if that presaved “marker grid” can also contain vertical placed markers ?
THX!
annother usecase for the global markers within a a project is you can scan a object in several seperate models and all will line up in the same location and orientation. so you can scan eg a low resultion mesh and aslo a high res one and both will be in third company software at the same location and orientation. it also helps to merge several scanned compounds, for example you can scan a object in a single rotation. fuse the pc then start a new scan with a reoriented scanner and merge both clouds into annother having a perfect line up.https://fb.watch/tIooO3sCso/
Totally agree , however other scanner firmware would need to be updated to support this feature .
We got a heat wave , so I am running early … it is 6:45 am
Ivan , it don’t works this way you described because the markers can’t move or bend so using a magic mat will not work unless you have it completely flat and the markers are exactly in the same position all the time . Sticking them on some black plastic plate would be better .
You can use from 1 to 6 mm inside diameter the outside diameter don’t matters you can use simple white dots on black surface .
However the reflective markers are better when scanning at angle of to use on plane so highly recommend.
