hello , i have the 2 scanners.
i have in project to use the range to scan an engine bloc fast , and using mini to scan details of the bloc.
it’s ok … but during mergin ( 9 MINI scans add to the range scan ( one after one ) ) then export the
point cloud… the point cloud density fall at each merging … so at the end the firts scan with range is better than my range + MINI adds !!! i don’t understand … there is a parameter somewhere???
First you need to clean each Mini point cloud after fusing it , remove the overlap and lose points , cut out anything that don’t belong to the object , after that you merge all scans made with MINI , then again clean the merged point cloud from Overlapping.
Next you clean the Range point cloud after fusing , and cut out the parts at 50% off that you already scanned and merged with MINI … then you merge them both together and after clean overlap again
Then you are ready for meshing at level 6
If you do not cut out the low accuracy points from the area that you want to be replaced with MINI scan it will not works and ignored by the algorithms.
This is the way for the best results … to keep the accuracy intact.
hello, thank’s.
i’m not sure to understand the method…
my questions below in FAT.
the RANGE have a fast scanning and don’t loose track , but poor accuracy and MIN lose track easily but have high accuracy … a super DIY to merge the better of the two world can be hudge !!.
First you need to clean each Mini point cloud after fusing it , remove the overlap and lose points , cut out anything that don’t belong to the object , after that you merge all scans made with MINI , then again clean the merged point cloud from Overlapping.
Next you clean the Range point cloud after fusing , and cut out the parts at 50% off ( what do you mean by "cut out the part at 50% off "? range and MINI must be cut out by 50% ?) that you already scanned and merged with MINI … then you merge them both together and after clean overlap again
Then you are ready for meshing at level 6
If you do not cut out the low accuracy points from the area that you want to be replaced with MINI scan it will not works and ignored by the algorithms. ( ok , but the algorithm can align correctly the différent mes in that case ? this is for that we only cut out 50% of the area to merge with MINI scan ? )
This is the way for the best results … to keep the accuracy intact.
Mini was designed to scan rings like jewelry and small objects , not for scanning big objects , it’s FOV is too small and requires dense amounts of features to keep tracking correctly, if you have troubles with tracking it is not MINI but the object you scanning with . How smaller the Field of View how more tracking points it needs .
You need to cut out the areas half way that will be replaced with MINI scans to keep the higher accuracy in that areas , you can’t merge 2 scans with the different pitch point and expect it to be the higher accuracy , for that reason you had issues .
Please eliminate the parts from the Range scan leaving some areas for Overlapping .
So the MINI scan can replace the missing areas and you end with the result you are looking for .
The idea of merging scans with 2 types of accuracy( pitch point ) will never works if you don’t follow this workflow .
No matter what you merge , you need always remove the parts that will overlap too much the second object to prevent overlapping , overlap will create artifacts and surface issues that are hard to edit after .
The algorithms align scans that was scanned with the same scanner and accuracy , you will always get seam between 2 different density , that is natural process , how dense the point cloud , how more details it will have , so the difference in both scans will be significantly different.
Algorithms overlap the scans , it don’t remove any parts , it just align them together and the rest of the work need to be done by you . If you have one big scan with lower accuracy and put on top fragments of a scan with higher accuracy , then you get just overlapping that will be eliminated while cleaning as it do not belongs to the low accuracy scan , for that reason you need to cut out the parts on Range scan partially to make space for the higher accuracy scan .
Even if you merge regular scans from one scanner you still need to reduce , clean and cut out parts before merging to avoid overlapping , or use Overlap detection to clean it … sadly with 2 scans with different accuracy you can’t do that automatic . You need to prepare the scans before .
If I have moment today I will show you it illustrated for visualization.
This is caused because Revoscan doesn’t do global bundle adjustment, it only aligns the fused pointclouds instead of the frames consisting each. I hope they rewrite the sofwtare so the fused pointcloud remembers all of its frames.
After the point cloud is fused , there actually no cell frames anymore just fused points at specific pitch point distance , so it can’t use frames or remember since majority of the data is removed in the process and no more needed .
The alignment is based on features of the scanned object , it don’t cares about pitch point distances , it is primitive alignment without full registration , it don’t register with precision , it just overlap the objects based on its features without removing overlaps in the process , for that reason it still need to be cleaned after merging .
The Revo Scan 5 alignment can be compared to the preview alignment in CLoud Compare .
With the same results as you would do it manually in other software so in short very simple .
The meshes created from the merged point cloud in RS5 need to be remeshed /Simplify to equalize the density of the points, the future version of RS5 will have this feature already , so the meshes will have a proper equalization focusing on the finer details of the mesh with proper density to avoid seams if merging 2 scans with different pitch point ( resolution )
So yes you are right about the global merging , it is simple but most of other advanced programs do have 2 steps regarding merging , and RS5 only one what is not optimal . It was designed to merge scans at the same pitch point resolution , and not multiple resolution scans , there is way around , but cost some work to get there .
That why you need to equalize your mesh after and that feature is not yet available but coming soon .
You mean you use it to make base mesh at the resolution of 0.1mm
And use MINI for the finer details at 0.02mm .
However no matter what program you use , you still need to prepare the scans the same way to get perfect results and equalize the pitch point distance to avoid seams .
I know the idea in theory is great , but it take some work to complete it , it is not so quick and simple as you may imagine , not in RS5 or CLoud Compare .
hello,
i have made multiple scan with MINI to scan the engine bloc ( intake side only ).
each scan have around 500.000points … i have merge then the 9 scans… one after one… the final global merge of the 9 scans done less than 450.000points… why the accurancy fall done like that ? where is the parameter to avoid cleanning during merge … in fact the global scan with RANGE is better to finish!!.
i have try a merging of the 9 scan in one time …1 500.000 point , this is better … but the overlapping of scans are not so good… so accurancy is bad …
thank’s
Tangi
There are no parameters like cleaning while merge , the overlapped areas was probably partially removed what is a good thing .
I hope you don’t expected 13.5 millions points after merge , that is not how it works.
Exactly , you counting points in place to create simple clean point cloud .
I prefer 450K clean proper point cloud than 1.5 milions points of overlap mess that is not good for anything .
I see other users scans with MINi producing overlapped scans of almost 50 millions points , where they can have clean point cloud of only 10 millions with precise accuracy and details , more is not always better . Especially if it is overlapped scans .
