BLK360 NOT leveled scans

Discuss all Leica Geosystems related issues here.
User avatar
topogeo
I have made 10-20 posts
I have made 10-20 posts
Posts: 14
Joined: Fri May 02, 2014 1:38 pm
Full Name: piercarlo roasio
Company Details: studio roasio
Company Position Title: titolare
Country: italy
Linkedin Profile: No

BLK360 NOT leveled scans

Post by topogeo » Mon Nov 06, 2017 2:48 pm

I know the BLK360 does not have an internal compensator.
I thought that the scans were, however, in some way approximated.
I disagree with balconies that are horizontal (verified with the total station).
From the scanning is a difference of 32 cm at 13.50 mt !!
Is it a characteristic of the BLK360 or is my unit having any problems?
Precisely I used a Leica TR100 tripod that has a small spherical bubble (and common BLK360's original tripod does not even have that :-)

Piercarlo
You do not have the required permissions to view the files attached to this post.

admad1st
I have made <0 posts
I have made <0 posts
Posts: 2
Joined: Thu Aug 10, 2017 4:43 pm
Full Name: Adam Leslie
Company Details: Global Energy Group
Company Position Title: Survey Manager
Country: United Kingdom
Linkedin Profile: Yes

Re: BLK360 NOT leveled scans

Post by admad1st » Fri Nov 10, 2017 12:35 pm

We have run various tests with the BLK360 and this was one of them. We deliberately set the first scan off level then registered cloud to cloud through RecapPro desktop. Outcome is the BLK360/Recap does not automatically compensate for scans being off level. Everything will become relative to how level the primary scan is. If being completely level is critical for your workflow then surveyed targets are required (as far as i can determine).

tbwester
V.I.P Member
V.I.P Member
Posts: 776
Joined: Fri Apr 20, 2012 1:49 pm
Full Name: Thad Wester
Company Details: WeWork
Company Position Title: Head of Reality Capture
Country: USA
Linkedin Profile: Yes
Location: SC/NYC

Re: BLK360 NOT leveled scans

Post by tbwester » Fri Nov 10, 2017 3:47 pm

I think this is an issue with recap?

My scans come in level using cyclone. Even when not setup perfectly level, they come in very level.

IMO all levels are very good, even our 5k matterport scans are as level as our p20 over 100 feet when we checked it.

JimC323
I have made <0 posts
I have made <0 posts
Posts: 3
Joined: Wed Oct 25, 2017 5:27 am
Full Name: Jim Caldwell
Company Details: Jim Caldwell
Company Position Title: ---
Country: United States
Linkedin Profile: No
Location: Northern California

Re: BLK360 NOT leveled scans

Post by JimC323 » Fri Nov 10, 2017 8:46 pm

I've had the same problem with out-of-level results -- sometimes grossly out of level) after bringing my BLK360 scans into ReCap Pro. In my experiments, this rendered my project useless, since I wanted to use my registered, unified point cloud to measure elevations at various points on sloped outdoor concrete surfaces.

This caused me to investigate whether I could use targets (for which I provide separately-measured coordinates that I trust) to force ReCap to orient the Z-axis correctly -- thereby making the X-Y plane perfectly horizontal. I think this is essentially the same idea as Admad1st alludes to in his earlier post on this topic.

By trial and error, I discovered that ReCap only allows me to enter the measured coordinates of a given target if it has first detected that a target is present in the region where I've said there is one (i.e., ReCap requires one to supply a "hint" as to where it will find a target, by clicking close to the target in a RealView of the scan).

The problem I've encountered with this seemingly reasonable strategy is that I can't get ReCap to recognize most of my targets. It DOES recognize SOME of them, but refuses to recognize others. This prevents me from getting the necessary minimum of three targets to match across any two scans that I'm trying to align. If this minimum of three is never reached, ReCap does not take into account any target data when it performs a registration.

I'm looking for insights as to the factors that determine whether ReCap recognizes a target. I can think of a few things: 1) the resolution of my BLK360 scans is too low (I'm using medium resolution right now); 2) a given target could be too far away from the BLK360 in a particular scan, such that the pertinent cloud points are too far apart; 3) the geometry or size of my targets doesn't correspond to whatever ReCap is looking for (I'm using hand-made checkerboard targets in which there are two blue 2"x2" squares on an off-white background -- the targets are numbered, but ReCap probably can't read the numbers); 4) the blue color I've used in my targets might be insufficiently distinct from the off-white background; 5) the surface normal of a target might be at too great of an angle relative to a line between the target and the scanner, causing the target image to be distorted too much; 6) there might not be enough ambient light. Regarding the target design, ReCap DOES detect some of these targets, suggesting that target design may not be the problem.

Any suggestions based on your experiences?

Ryan
V.I.P Member
V.I.P Member
Posts: 189
Joined: Thu Aug 14, 2008 3:29 pm
Full Name: Ryan Frenz
Company Details: Autodesk
Company Position Title: Engineering Manager - ReCap
Country: USA
Linkedin Profile: No

Re: BLK360 NOT leveled scans

Post by Ryan » Sat Nov 11, 2017 1:56 am

ReCap Pro Mobile is built to use the measured inclination as reported by the BLK firmware.

If you're seeing different results between the various compatible apps it would be great to get a look at a scan.

-Ryan

JimC323
I have made <0 posts
I have made <0 posts
Posts: 3
Joined: Wed Oct 25, 2017 5:27 am
Full Name: Jim Caldwell
Company Details: Jim Caldwell
Company Position Title: ---
Country: United States
Linkedin Profile: No
Location: Northern California

Re: BLK360 NOT leveled scans

Post by JimC323 » Mon Nov 13, 2017 9:56 pm

Thanks for responding, Ryan. Unfortunately, the particular scans upon which my earlier post was based contain personally-identifiable information which I don't want to publicize, for security reasons. I can probably reproduce the problem in another setting -- and assuming I'm able to do so, I'd be happy to share my data. So, we'll have to put this conversational branch on hold for now.

In the meantime, there are several factors (that might be affecting ReCap's ability to recognize my targets) that I can and will investigate further.

JimC323
I have made <0 posts
I have made <0 posts
Posts: 3
Joined: Wed Oct 25, 2017 5:27 am
Full Name: Jim Caldwell
Company Details: Jim Caldwell
Company Position Title: ---
Country: United States
Linkedin Profile: No
Location: Northern California

Re: BLK360 NOT leveled scans

Post by JimC323 » Wed Nov 15, 2017 2:23 am

OK [referring to my earlier posts of 11/10/17 and 11/13/17] ... I've reproduced the failure of ReCap Pro to recognize a target, using a different scene that doesn't reveal any personally-identifiable information. This is a ONE-SCAN project, and the scan used is from a BLK360 operating in its default medium-resolution mode. I've attached two screen captures which I took from the pertinent ReCap Pro session. The first shows an image (RealView) of the target that the software seems unable to detect. The second shows the cloud points (3D View) in the region of the same target. I include the cloud points snapshot so that one can see that, in the point cloud, the target is represented by many cloud points -- yet, ReCap Pro doesn't detect it as a target.

NonRecogTarget_Image.JPG
NonRecogTarget_Points.JPG

Oddly enough, ReCap Pro DOES detect other targets of more or less identical design (but of poorer point cloud quality) at other locations in this same project. By poorer point cloud quality, I mean that because these other targets are further away from the BLK360, the point cloud density in the pertinent regions is presumably LOWER than in the region of the target shown in the screen snapshots. In addition, recognition of the other targets occurs despite the fact that the BLK360 sees them from angles that are LARGER relative to the target normals, distorting the target image.

By the way, I made another scan without moving the BLK360, but with the BLK360 set to high-resolution mode instead of medium resolution. The same target is not detected by ReCap Pro.

So -- my question remains: is there something about the targets I'm using that doesn't suit the algorithm ReCap Pro uses to detect targets? For example, could it be the blue color I've used?
You do not have the required permissions to view the files attached to this post.

Return to “Leica Geosystems”