Conflicting sphere target distances Faro Scene 2022.1

[lsf]

Hi,

There is also one thing to know: scanning points are less accurate at the bottom & zenith.
This is something I tested a few years ago, but it was with a Focus 120. I had differences up to 5mm.

I can see that you have spheres right after the bottom disk.

Try to place your spheres further.

However, a difference of 1-2mm is something common.

@yankoch Thank you for your input. Wow, this is really discouraging but I hope it is not true for Focus S70, because its specs say +-1mm error (2mm total). And since all the spheres are within 10 meters, then my understanding is that the max error would be 1/7 of 2mm = 0.28mm. Man, I really hope that this is not true for S70 otherwise this instrument cannot be used where high precision is required.

About placing spheres further away from the scanner. I also hope that this is not true for S70 and I wonder if someone who has S70 can share his/her input. It is not as simple as it appears regarding the sphere placement. I used cones to help position scan positions to make sure that from each position I see everything that I need to see. From interior scans this includes seeing the interior spheres, the exterior spheres, make sure that exterior spheres are not visible through the glass, and to see the edge of the wood panel which is close to the yellow densglass panel. It was really challenging to position everything to see the right amount of everything (I attached some more images for visualization of this project). I was really under the impression that what ever it captures, is sub mm precision and that I can trust everything that I see in the cloud even if it is right next to the hollow disk of the scan position.

I appreciate your input but I am still curious what is the experience of S70 owners.

[VXGrid]

The Scan 1 (the left side of the image is the summary of everything, it shows all the targets, the distances and the discrepancies. Ignore the images on the right side.

EDIT: See the diagram attached, I hope that helps

I take that in the diagramm the name of the scan is corresponding to the scan name of the files.

I put the scans into PointCab Origins and registered them there, so the results will be definetly be different to Scene, but should be somewhat in the same range.

So I checked one or two spheres in regards to found accuracy (middle point of the sphere), and this is submm - on sphere diameter = 200 mm.
Meaning your spheres should be fine.

Regarding registration accuracy this is not great, because the constellation of spheres is not ideal at all.
All spheres are in the middle, building more or less a triangle with a maximum distance of 7 metres.

SphereDistribution.png

Reminder: The best constellation possible is when three spheres are building a triangle (more spheres are better, as long as the spheres are not just placed on a line) with the biggest possible area size, while the scan positions are placed inside this triangle, not outside.
Takeaway: Increase the area size of the bounding polygon of all visible spheres and place scans inside this polygon.

But what really brought the values down was removing the sphere in scan Santana003, which was scanned through the window.
(The nearly visible sphere in scan Santana005 was not used in any test)

Find attached two PDFs, RegistrationReport_allSpheres.pdf, where I used all spheres and RegistrationReport.pdf, where I didn't use the window sphere of scan Santana003

Here the values as text (since I can't upload txt here as it seems)
Identical sphere name in different scans means this sphere was matched.

RegistrationReport_allSpheres.pdf
ID Scan X[m] Y[m] Z[m] Omega[°] Phi[°] Kappa[°] S_X[m] S_Y[m] S_Z[m] S_Omega[°] S_Phi[°] S_Kappa[°]

{936de2dd-e197-4986-ae45-156e964e22d4} Santana005 -0.00000 -0.00000 -0.00000 -0.1285266040 0.2342635592 -0.0002567512 0.000001 0.000001 0.000001 0.01293542877 0.01175427813 0.00005219362
{da788a06-de46-40d1-bc97-4046f76e5c00} Santana001 1.80604 -9.77894 0.02966 -0.0272748116 -0.0440173647 132.2897666453 0.000962 0.001034 0.001955 0.01257789296 0.01073049904 0.01158594124
{a6b3a5bc-0e74-48bc-bcb8-691bb5c822c8} Santana003 -1.09567 -9.77543 0.03254 -0.1484749027 0.2011595085 -155.4450751083 0.000962 0.000756 0.001884 0.01105193471 0.01229645249 0.01158592110
{d8d8ea40-3fe3-4b04-9ecb-52a807f5c8b0} Santana002 1.78820 -6.73226 0.02968 0.1766324776 -0.1385414508 130.0653299815 0.000789 0.001175 0.001952 0.01560822837 0.01259204436 0.01322502293
{2de00a3e-185b-45dc-a896-06d5ba3438b0} Santana004 -5.47047 -3.21630 -0.03066 0.4025435496 0.3069958556 47.6689759973 0.000999 0.001057 0.001797 0.01070316498 0.01260125522 0.01158600636
{aecdd49a-6974-4178-8c4f-f957f103147c} Santana006 -8.07106 -9.26950 0.01156 -0.5543395928 0.3410949416 2.5274999142 0.001364 0.001793 0.002382 0.01471709155 0.01566498299 0.01716894830


Type Name Scan X[m] Y[m] Z[m] R_d[m] R_X[m] R_Y[m] R_Z[m] R_X'[m] R_Y'[m] R_Z'[m] X~[m] Y~[m] Z~[m]

Sphere S1029 Santana005 -1.65554 -3.25053 -1.65243 0.00270 0.00222 -0.00153 -0.00017 0.00223 -0.00153 -0.00016 -1.65331 -3.25206 -1.65259
Sphere S1030 Santana005 -4.83877 -6.97985 -1.66114 0.00186 -0.00183 0.00035 -0.00010 -0.00183 0.00035 -0.00011 -4.84059 -6.97950 -1.66124
Sphere S1031 Santana005 -2.82647 -6.86795 -1.64786 0.00141 -0.00084 -0.00113 -0.00007 -0.00084 -0.00113 -0.00007 -2.82731 -6.86908 -1.64793
Sphere S1032 Santana005 1.55329 -8.31958 -1.61415 0.00237 0.00044 0.00231 0.00033 0.00044 0.00231 0.00034 1.55373 -8.31728 -1.61382
Sphere S1032 Santana001 1.55412 -8.31595 -1.61372 0.00138 -0.00039 -0.00132 -0.00010 -0.00071 0.00118 -0.00010 1.55373 -8.31728 -1.61382
Sphere S1029 Santana001 -1.65183 -3.25190 -1.65312 0.00158 -0.00149 -0.00015 0.00052 0.00089 0.00120 0.00052 -1.65331 -3.25206 -1.65259
Sphere S1005 Santana001 0.53125 -9.80051 -1.61171 0.00086 0.00074 0.00004 0.00044 -0.00046 -0.00058 0.00044 0.53198 -9.80046 -1.61127
Sphere S1030 Santana001 -4.84025 -6.98075 -1.66052 0.00149 -0.00034 0.00125 -0.00072 0.00116 -0.00059 -0.00072 -4.84059 -6.97950 -1.66124
Sphere S1007 Santana001 -0.57993 -8.73504 -1.61130 0.00078 0.00078 -0.00001 -0.00004 -0.00053 -0.00056 -0.00004 -0.57915 -8.73505 -1.61134
Sphere S1031 Santana001 -2.82802 -6.86927 -1.64783 0.00074 0.00071 0.00019 -0.00010 -0.00034 -0.00065 -0.00010 -2.82731 -6.86908 -1.64793
Sphere S1030 Santana003 -4.84452 -6.97731 -1.66263 0.00470 0.00393 -0.00218 0.00139 -0.00267 0.00361 0.00139 -4.84059 -6.97950 -1.66124
Sphere S1032 Santana003 1.55435 -8.31694 -1.61381 0.00070 -0.00061 -0.00034 -0.00001 0.00070 0.00006 -0.00001 1.55373 -8.31728 -1.61382
Sphere S1007 Santana003 -0.57839 -8.73538 -1.61125 0.00084 -0.00077 0.00033 -0.00009 0.00056 -0.00062 -0.00009 -0.57915 -8.73505 -1.61134
Sphere S1031 Santana003 -2.82701 -6.86983 -1.64689 0.00131 -0.00030 0.00074 -0.00104 -0.00003 -0.00080 -0.00104 -2.82731 -6.86908 -1.64793
Sphere S1005 Santana003 0.53373 -9.80039 -1.61097 0.00177 -0.00175 -0.00007 -0.00030 0.00162 -0.00066 -0.00030 0.53198 -9.80046 -1.61127
Sphere S1029 Santana003 -1.65281 -3.25358 -1.65264 0.00160 -0.00050 0.00152 0.00005 -0.00018 -0.00159 0.00004 -1.65331 -3.25206 -1.65259
Sphere S1005 Santana002 0.53197 -9.80080 -1.61100 0.00043 0.00002 0.00033 -0.00028 0.00024 -0.00023 -0.00028 0.53198 -9.80046 -1.61127
Sphere S1032 Santana002 1.55400 -8.31703 -1.61407 0.00044 -0.00027 -0.00024 0.00024 -0.00001 0.00037 0.00024 1.55373 -8.31728 -1.61382
Sphere S1007 Santana002 -0.57947 -8.73507 -1.61156 0.00039 0.00032 0.00002 0.00023 -0.00019 -0.00026 0.00023 -0.57915 -8.73505 -1.61134
Sphere S1030 Santana002 -4.83996 -6.97963 -1.66064 0.00088 -0.00063 0.00014 -0.00060 0.00051 0.00039 -0.00060 -4.84059 -6.97950 -1.66124
Sphere S1031 Santana002 -2.82788 -6.86884 -1.64834 0.00074 0.00057 -0.00024 0.00041 -0.00055 -0.00028 0.00041 -2.82731 -6.86908 -1.64793
Sphere S1030 Santana004 -4.83983 -6.98025 -1.66089 0.00113 -0.00076 0.00075 -0.00036 0.00004 0.00107 -0.00036 -4.84059 -6.97950 -1.66124
Sphere S1005 Santana004 0.53099 -9.80016 -1.61142 0.00105 0.00099 -0.00030 0.00015 0.00044 -0.00094 0.00015 0.53198 -9.80046 -1.61127
Sphere S1007 Santana004 -0.57883 -8.73472 -1.61124 0.00048 -0.00033 -0.00034 -0.00010 -0.00047 0.00001 -0.00010 -0.57915 -8.73505 -1.61134
Sphere S1031 Santana004 -2.82707 -6.86950 -1.64863 0.00085 -0.00024 0.00042 0.00070 0.00014 0.00047 0.00070 -2.82731 -6.86908 -1.64793
Sphere S1032 Santana004 1.55290 -8.31688 -1.61336 0.00103 0.00083 -0.00040 -0.00046 0.00027 -0.00089 -0.00046 1.55373 -8.31728 -1.61382
Sphere S1029 Santana004 -1.65282 -3.25193 -1.65267 0.00051 -0.00049 -0.00013 0.00008 -0.00042 0.00028 0.00007 -1.65331 -3.25206 -1.65259
Sphere S1030 Santana006 -4.84023 -6.97918 -1.66163 0.00061 -0.00036 -0.00031 0.00038 -0.00038 -0.00030 0.00038 -4.84059 -6.97950 -1.66124
Sphere S1029 Santana006 -1.65357 -3.25234 -1.65211 0.00061 0.00025 0.00028 -0.00048 0.00027 0.00028 -0.00048 -1.65331 -3.25206 -1.65259
Sphere S1031 Santana006 -2.82742 -6.86911 -1.64803 0.00015 0.00011 0.00003 0.00010 0.00011 0.00002 0.00010 -2.82731 -6.86908 -1.64793


RegistrationReport.pdf
ID Scan X[m] Y[m] Z[m] Omega[°] Phi[°] Kappa[°] S_X[m] S_Y[m] S_Z[m] S_Omega[°] S_Phi[°] S_Kappa[°]

{936de2dd-e197-4986-ae45-156e964e22d4} Santana005 -0.00000 -0.00000 -0.00000 -0.1301389886 0.2377342738 -0.0002567512 0.000001 0.000001 0.000001 0.00869292891 0.00794528430 0.00003501035
{da788a06-de46-40d1-bc97-4046f76e5c00} Santana001 1.80567 -9.77931 0.02972 -0.0236881577 -0.0458431294 132.2848609491 0.000646 0.000694 0.001312 0.00849327964 0.00721701096 0.00777869955
{a6b3a5bc-0e74-48bc-bcb8-691bb5c822c8} Santana003 -1.09826 -9.77504 0.03201 -0.1449485350 0.2201553161 -155.4644281753 0.000703 0.000529 0.001270 0.00747718092 0.00978424600 0.00854184741
{d8d8ea40-3fe3-4b04-9ecb-52a807f5c8b0} Santana002 1.78806 -6.73269 0.02976 0.1787605798 -0.1413163567 130.0608689182 0.000530 0.000789 0.001310 0.01048618423 0.00848217027 0.00889230828
{2de00a3e-185b-45dc-a896-06d5ba3438b0} Santana004 -5.47027 -3.21605 -0.03024 0.4046971655 0.3103953658 47.6640840393 0.000670 0.000709 0.001213 0.00720581132 0.00850284812 0.00777874566
{aecdd49a-6974-4178-8c4f-f957f103147c} Santana006 -8.07051 -9.26988 0.01235 -0.5558599955 0.3452668571 2.5316571549 0.000917 0.001205 0.001614 0.00987991505 0.01056972988 0.01154256552


Type Name Scan X[m] Y[m] Z[m] R_d[m] R_X[m] R_Y[m] R_Z[m] R_X'[m] R_Y'[m] R_Z'[m] X~[m] Y~[m] Z~[m]

Sphere S1067 Santana005 -4.83886 -6.97990 -1.66065 0.00107 -0.00106 0.00000 0.00016 -0.00106 0.00000 0.00016 -4.83992 -6.97989 -1.66049
Sphere S1068 Santana005 -2.82657 -6.86800 -1.64750 0.00140 -0.00101 -0.00093 -0.00029 -0.00101 -0.00093 -0.00030 -2.82758 -6.86893 -1.64779
Sphere S1069 Santana005 1.55319 -8.31963 -1.61401 0.00207 -0.00000 0.00204 0.00034 -0.00000 0.00204 0.00035 1.55319 -8.31759 -1.61367
Sphere S1070 Santana005 -1.65527 -3.25091 -1.65228 0.00236 0.00207 -0.00111 -0.00022 0.00207 -0.00111 -0.00021 -1.65320 -3.25203 -1.65250
Sphere S1069 Santana001 1.55377 -8.31634 -1.61368 0.00138 -0.00058 -0.00125 0.00001 -0.00054 0.00128 0.00001 1.55319 -8.31759 -1.61367
Sphere S1070 Santana001 -1.65175 -3.25201 -1.65295 0.00151 -0.00145 -0.00002 0.00045 0.00096 0.00108 0.00045 -1.65320 -3.25203 -1.65250
Sphere S1073 Santana001 0.53077 -9.80080 -1.61157 0.00054 0.00034 0.00011 0.00040 -0.00015 -0.00033 0.00040 0.53111 -9.80069 -1.61117
Sphere S1067 Santana001 -4.84049 -6.98059 -1.66007 0.00099 0.00057 0.00069 -0.00042 0.00013 -0.00088 -0.00042 -4.83992 -6.97989 -1.66049
Sphere S1075 Santana001 -0.58032 -8.73524 -1.61110 0.00049 0.00045 0.00012 -0.00015 -0.00021 -0.00041 -0.00015 -0.57987 -8.73512 -1.61126
Sphere S1068 Santana001 -2.82825 -6.86928 -1.64751 0.00081 0.00067 0.00036 -0.00028 -0.00019 -0.00074 -0.00028 -2.82758 -6.86893 -1.64779
Sphere S1077 Santana003 -4.84561 -6.97552 -1.66415
Sphere S1069 Santana003 1.55279 -8.31731 -1.61335 0.00058 0.00040 -0.00028 -0.00032 -0.00025 0.00042 -0.00032 1.55319 -8.31759 -1.61367
Sphere S1075 Santana003 -0.58009 -8.73504 -1.61152 0.00035 0.00022 -0.00008 0.00026 -0.00016 0.00016 0.00026 -0.57987 -8.73512 -1.61126
Sphere S1068 Santana003 -2.82807 -6.86872 -1.64774 0.00053 0.00049 -0.00021 -0.00005 -0.00036 0.00039 -0.00005 -2.82758 -6.86893 -1.64779
Sphere S1073 Santana003 0.53167 -9.80042 -1.61096 0.00066 -0.00056 -0.00027 -0.00021 0.00062 0.00002 -0.00021 0.53111 -9.80069 -1.61117
Sphere S1070 Santana003 -1.65265 -3.25287 -1.65281 0.00105 -0.00055 0.00084 0.00031 0.00015 -0.00099 0.00031 -1.65320 -3.25203 -1.65250
Sphere S1073 Santana002 0.53149 -9.80110 -1.61088 0.00063 -0.00038 0.00041 -0.00029 0.00056 0.00003 -0.00030 0.53111 -9.80069 -1.61117
Sphere S1069 Santana002 1.55364 -8.31742 -1.61399 0.00058 -0.00046 -0.00017 0.00032 0.00016 0.00046 0.00032 1.55319 -8.31759 -1.61367
Sphere S1075 Santana002 -0.57986 -8.73529 -1.61136 0.00020 -0.00001 0.00017 0.00011 0.00014 -0.00011 0.00011 -0.57987 -8.73512 -1.61126
Sphere S1067 Santana002 -4.84022 -6.97952 -1.66016 0.00058 0.00030 -0.00037 -0.00033 -0.00048 0.00001 -0.00033 -4.83992 -6.97989 -1.66049
Sphere S1068 Santana002 -2.82813 -6.86889 -1.64798 0.00058 0.00055 -0.00004 0.00019 -0.00038 -0.00039 0.00019 -2.82758 -6.86893 -1.64779
Sphere S1067 Santana004 -4.84007 -6.98008 -1.66043 0.00025 0.00015 0.00019 -0.00006 0.00024 0.00002 -0.00006 -4.83992 -6.97989 -1.66049
Sphere S1073 Santana004 0.53051 -9.80045 -1.61128 0.00065 0.00060 -0.00024 0.00011 0.00022 -0.00060 0.00011 0.53111 -9.80069 -1.61117
Sphere S1075 Santana004 -0.57921 -8.73491 -1.61104 0.00072 -0.00066 -0.00021 -0.00022 -0.00059 0.00034 -0.00022 -0.57987 -8.73512 -1.61126
Sphere S1068 Santana004 -2.82730 -6.86951 -1.64831 0.00083 -0.00028 0.00058 0.00052 0.00024 0.00060 0.00052 -2.82758 -6.86893 -1.64779
Sphere S1069 Santana004 1.55254 -8.31726 -1.61331 0.00081 0.00064 -0.00033 -0.00036 0.00019 -0.00070 -0.00035 1.55319 -8.31759 -1.61367
Sphere S1070 Santana004 -1.65274 -3.25204 -1.65250 0.00045 -0.00045 0.00001 0.00000 -0.00030 0.00034 -0.00000 -1.65320 -3.25203 -1.65250
Sphere S1067 Santana006 -4.83996 -6.97938 -1.66114 0.00082 0.00004 -0.00051 0.00064 0.00001 -0.00052 0.00064 -4.83992 -6.97989 -1.66049
Sphere S1070 Santana006 -1.65357 -3.25231 -1.65196 0.00071 0.00038 0.00028 -0.00054 0.00039 0.00027 -0.00053 -1.65320 -3.25203 -1.65250
Sphere S1068 Santana006 -2.82716 -6.86916 -1.64768 0.00049 -0.00042 0.00023 -0.00011 -0.00041 0.00025 -0.00011 -2.82758 -6.86893 -1.64779

[VXGrid]

[...]

However, a difference of 1-2mm is something common.

@yankoch Thank you for your input. Wow, this is really discouraging but I hope it is not true for Focus S70, because its specs say +-1mm error (2mm total). And since all the spheres are within 10 meters, then my understanding is that the max error would be 1/7 of 2mm = 0.28mm. Man, I really hope that this is not true for S70 otherwise this instrument cannot be used where high precision is required.

[...]

I am not going to check the data sheet, but if I remember correctly FARO has something like: ranging error + angle error.
The ranging error is normally X mm + Y ppm, meaning your X is there whatever (and this being 1mm might be right), which is the minimum error + the distance factor.
Then there is the angle error, which becomes bigger with greater distance.
The calculation is not: Max error / Max range.

But this is the single point of a single scan accuarcy.
If you measure anything within this one scan point cloud you will have this accuracy.

Now you are using multiple scans, so the calculation is single scan error + single scan error + registration error.
Registration error is depending on how many scans are connected to each other - Error propagation.
To make a very simple comparison: If you measure a distance of around 50 meter with a tape which is max 20 meter long you need to measure three times. If your tape placements are not perfectly parallel your measurement will be off by an error.

Please check the Taming errors series by Daniel Wujanz:

search.php?keywords=Taming+errors&terms ... mit=Search

[yankoch]

@yankoch Thank you for your input. Wow, this is really discouraging but I hope it is not true for Focus S70, because its specs say +-1mm error (2mm total). And since all the spheres are within 10 meters, then my understanding is that the max error would be 1/7 of 2mm = 0.28mm. Man, I really hope that this is not true for S70 otherwise this instrument cannot be used where high precision is required.

About placing spheres further away from the scanner. I also hope that this is not true for S70 and I wonder if someone who has S70 can share his/her input. It is not as simple as it appears regarding the sphere placement. I used cones to help position scan positions to make sure that from each position I see everything that I need to see. From interior scans this includes seeing the interior spheres, the exterior spheres, make sure that exterior spheres are not visible through the glass, and to see the edge of the wood panel which is close to the yellow densglass panel. It was really challenging to position everything to see the right amount of everything (I attached some more images for visualization of this project). I was really under the impression that what ever it captures, is sub mm precision and that I can trust everything that I see in the cloud even if it is right next to the hollow disk of the scan position.

I appreciate your input but I am still curious what is the experience of S70 owners.

Please do the following test at your office:
- do the test in a room with 2 walls 3-4 meters apart ,
- place 2 checkerboards at the top of the walls,
- place 2 checkerboards at scanner height,
- place 2 checkerboards at the bottom.

For each height, the checkerboard must be approximately opposite the other.
Use a laser range finder (Leica disto for example) to measure the distance between the checkerboards at each height.
Then, place the scanner in the middle of the two walls. Orient the scanner head so that the screen is in front of one wall. This way you'll make sure that the scanner will scan the checkerboards on both sides.
Do a 1/5 resolution scan, quality 3.
Then, check the distances in Scene.
Tell me if you have differences between the Disto and the measurement in Scene.

[landmeterbeuckx]

The Scan 1 (the left side of the image is the summary of everything, it shows all the targets, the distances and the discrepancies. Ignore the images on the right side.

EDIT: See the diagram attached, I hope that helps

I take that in the diagramm the name of the scan is corresponding to the scan name of the files.

I put the scans into PointCab Origins and registered them there, so the results will be definetly be different to Scene, but should be somewhat in the same range.

So I checked one or two spheres in regards to found accuracy (middle point of the sphere), and this is submm - on sphere diameter = 200 mm.
Meaning your spheres should be fine.

Regarding registration accuracy this is not great, because the constellation of spheres is not ideal at all.
All spheres are in the middle, building more or less a triangle with a maximum distance of 7 metres.
SphereDistribution.png

Reminder: The best constellation possible is when three spheres are building a triangle (more spheres are better, as long as the spheres are not just placed on a line) with the biggest possible area size, while the scan positions are placed inside this triangle, not outside.
Takeaway: Increase the area size of the bounding polygon of all visible spheres and place scans inside this polygon.

Bad sphere placement, scanned all exterior the spheres circle radius, no scan in the door treshold.

It has nothing to do with your scanner, it's the use of the scanner that's wrong.

[lsf]

Hi, thank you so much for your input!

I want to emphasize that I am not going into the scan registration step. I am simply comparing the distances between known targets between different scans.

@landmeterbeuckx
You said "Bad sphere placement, scanned all exterior the spheres circle radius, no scan in the door threshold." The registration part is not of concern yet because I am just comparing the distances between the known targets, therefore it should not matter where the scan positions are as long as I can see all the targets directly and not through the glass, and this requirement was satisfied. Though while on this subject, I am curios.. If I did C2C registration, then absolutely a scan at the door threshold would be a must, but since I am using target registration, and since my center interior scan sees all all exterior targets and my center exterior scan sees all interior scans (see the attached picture), then why would my scan locations be considered bad? I think I read in the Faro paper book/manual that they want the spheres to be placed at different heights (see Note 1) but for the purpose of this test it should not matter.

@yankoch
Thank you! I will do it and get back with the results.

@VXGrid
Thank you so much for your extensive information. Good comment about the 'angle error'. My point was that all the spheres in my case are within 7m which is 10x less than the maximum range, so I would expect a negligible error of any kind in one singe scan.

Regarding the registration error. You are mentioning the registration error, and this is exactly why I am conducting this test, at this phase I am not concerned with the registration yet, I am only comparing the same 'triangular' distances between each separate scan, and I am expecting those to have submillimeter variance. All the distances match between left and right scans, but some distances between the left or right and center scans don't match, so this is how I have simplified the problem. I don't know what the angular error is, but if the range error is only +-1mm over 70m, then I have t think that regardless of where my spherical targets are located in the 3D (next to the hollow scan station disk, sideways, at the ceiling, on the wall etc) the distances between those targets should be the same in any of the scans that sees all 3 targets, right?

Therefore my instinct is that this is not the problem with the point cloud acquisition process, this is the problem with the sphere matching algorithm but I am not sure because I don't have any other software where I can find all the 8" spheres automatically.

I see that in your diagram the exterior sphere distance in your software shows as 4906mm by scene shows 4907.5mm. Your ext to int sphere distance is 6004mm by Scene shows 5994.5mm. This is puzzling.


Note 1:
I think I read in the Faro paper book/manual that they want the spheres to be placed at different heights:
This does not make sense to me from the mathematical perspective, because all the 3d points are somewhere in the 3D coordinate system from the scanner perspective so it should not matter whether the target sphere is on the floor or on the table. I discussed this with the the support, but the person was not able to answer the question. As long as we have 3 known points in 3D in each scan, then we should be able to position and rotate these scans to match these 3 known points, the math is as simple as that. The support told me that they are projecting the points from one coordinate system to another, or to a plane, I don't recall, but the fundamental math is simple.

[VXGrid]

[...]

@VXGrid
Thank you so much for your extensive information. Good comment about the 'angle error'. My point was that all the spheres in my case are within 7m which is 10x less than the maximum range, so I would expect a negligible error of any kind in one singe scan.

Regarding the registration error. You are mentioning the registration error, and this is exactly why I am conducting this test, at this phase I am not concerned with the registration yet, I am only comparing the same 'triangular' distances between each separate scan, and I am expecting those to have submillimeter variance. All the distances match between left and right scans, but some distances between the left or right and center scans don't match, so this is how I have simplified the problem. I don't know what the angular error is, but if the range error is only +-1mm over 70m, then I have t think that regardless of where my spherical targets are located in the 3D (next to the hollow scan station disk, sideways, at the ceiling, on the wall etc) the distances between those targets should be the same in any of the scans that sees all 3 targets, right?

Therefore my instinct is that this is not the problem with the point cloud acquisition process, this is the problem with the sphere matching algorithm but I am not sure because I don't have any other software where I can find all the 8" spheres automatically.

I see that in your diagram the exterior sphere distance in your software shows as 4906mm by scene shows 4907.5mm. Your ext to int sphere distance is 6004mm by Scene shows 5994.5mm. This is puzzling.
[...]

I see you are interested in the distance between the spheres in the local scans. These should indeed match with a distance divergation of around 3 to 4 mm I guess (again sphere detection error *2 + point accuarcy error of the scanner)

Feel free to compare these with the PointCab Origins results, here are the spheres in local coordiantes, identical sphere name should mean identical sphere in the Scans.
Order is Name X Y Z W, W should be 1.0

Scan Santana001
S1069 1.2504931248114119 -0.7972756617158745 -1.6447238118270593 1.0000000000000000
S1070 7.1538185072061733 -1.8331499679852370 -1.6891465953710738 1.0000000000000000
S1073 0.8405690480986864 0.9583236098376381 -1.6415610477324540 1.0000000000000000
S1067 6.5408205073682941 3.0345584767661822 -1.6937658565288913 1.0000000000000000
S1075 2.3764378526575771 1.0633843837147712 -1.6422815655141478 1.0000000000000000
S1068 5.2693057049981515 1.4709936729962423 -1.6808365376931327 1.0000000000000000

Scan Santana002
S1073 -1.5437610604604817 2.9314453076200824 -1.6459921493785539 1.0000000000000000
S1069 -1.0660745260040538 1.1942240863810745 -1.6448631774370146 1.0000000000000000
S1075 -0.0127699223328024 3.0960623341616431 -1.6507694339337182 1.0000000000000000
S1067 4.0728682132282037 5.2265626139722539 -1.7162939635487233 1.0000000000000000
S1068 2.8625924487124550 3.6154335338688104 -1.6960982162558658 1.0000000000000000

Scan Santana003
S1077 2.2529560115219942 -4.0985661023555942 -1.6978942151879266 1.0000000000000000
S1069 -3.0106559497682546 -0.2210408627331164 -1.6575075135639592 1.0000000000000000
S1075 -0.8969301547582760 -0.7267490172401221 -1.6488349458586777 1.0000000000000000
S1068 0.3731900386846381 -3.3579509063419919 -1.6868363732892884 1.0000000000000000
S1073 -1.4658863843994336 0.7040976146556431 -1.6468451713191001 1.0000000000000000
S1070 -2.1975683794879592 -6.1591529199525210 -1.7088720330700060 1.0000000000000000

Scan Santana004
S1067 -2.3491122290188158 -3.0123642467630840 -1.6217078466850530 1.0000000000000000
S1073 -0.8172846565395343 -8.8812474508414017 -1.5227980040524363 1.0000000000000000
S1075 -0.7770047001715157 -7.3433488556719277 -1.5332055131087647 1.0000000000000000
S1068 -0.9119059002923549 -4.4255805625345159 -1.5918198384580580 1.0000000000000000
S1069 0.9674062723968652 -8.6377995169673412 -1.5168838257177439 1.0000000000000000
S1070 2.5531658312675791 -2.8574797207226608 -1.5883136618984921 1.0000000000000000

Scan Santana005
S1067 -4.8319015157869307 -6.9760844218062372 -1.6965661214979821 1.0000000000000000
S1068 -2.8196750877892929 -6.8642218193171054 -1.6748086612100563 1.0000000000000000
S1070 -1.6483826345085606 -3.2471449875445146 -1.6665175437235609 1.0000000000000000
S1097 1.5600364462622913 -8.3163683103687482 -1.6254140258145917 1.0000000000000000

Scan Santana006
S1067 3.3385951564251548 2.1615083726026407 -1.6325001692689802 1.0000000000000000
S1070 6.6863901910860140 5.7437442426867822 -1.5683979623510922 1.0000000000000000
S1068 5.3541857140039317 2.1824638184747167 -1.6066971675526536 1.0000000000000000


While pressing a thousand times Ctrl + C Ctrl + V to copy the data I thought about what you mentioned, that this is not correlated to the registration, but it is indeed.
See when you do registration from all matched features you will calculate a virtual best fit sphere, from which you real measured spheres diverge. I uploaded these values as well.
In the values after
Type Name Scan X[m] Y[m] Z[m] R_d[m] R_X[m] R_Y[m] R_Z[m] R_X'[m] R_Y'[m] R_Z'[m] X~[m] Y~[m] Z~[m]
You should check the R_d[m] R_X[m] R_Y[m] R_Z[m] values because these are the distance between the estimated and real sphere.

And there you can directly see if it correlates with distance or with angle. Since you device is measuring two angles, a horizontal and a vertical one.

[Scott]

I did some tests with Tad Laird from Koppa Targets in December 2014. We wanted to get an idea of the general range limits of 150mm spheres (and a few larger diameter spheres)

[lsf]

@VXGrid
Again thank you so much for all the data. I will make a Google Spreadsheet and calculate the distances between your point coordinates.

I see you are interested in the distance between the spheres in the local scans. These should indeed match with a distance divergation of around 3 to 4 mm I guess (again sphere detection error *2 + point accuarcy error of the scanner)

Why do you think the variance would be so bad - 3 to 4 mm on such close (max 7m) distance? Regardless, this is a good point, I need to find some Faro formula to figure out what could be the max error at some specific distance and angle from the scanner position.

While pressing a thousand times Ctrl + C Ctrl + V to copy the data I thought about what you mentioned, that this is not correlated to the registration, but it is indeed.

No + no, I did not mean that it is not correlated to the registration, they are, I absolutely agree with you. What I tried to say is that I am not yet at the registration step. I am currently trying to understand if I can trust 100% to the automatic sphere detection tool in Scene. You see, if each scan shows different distances between the same spheres in separate scans, then there is no point of hoping to get good registration results. So when I compare left and right scans, the results match and I can trust that, but the problem are the center scans, the distances between targets don't match ones on the left and right scan. So I am at the very first step where I inspect the raw scan data, so nothing else should matter at this early step, right?

[lsf]

@Scott

Did you place the targets using totalstation and were you expecting all the distances to match exactly? Or what does the "Test 1.jpg" tell us?