Reality Capture Round-Trip for Construction: Physical to Digital and Back to Physical

DANIEL CHAPEK: That should help, huh? All right, now I got to watch what I say. Yeah, no problem. Thank you.

Any questions on who I am, what I do? No. OK, good.

So next, I want to talk a little bit about-- you have an introduction to reality capture and reality computing. So when I'm talking about reality capture, for those of you guys that are scanning in here, if you were to describe or define reality capture, how would you define it?

Oh, by the way, this is an interactive class.

[CHUCKLING]

Go ahead and throw something out. You're not going to be wrong. How would you define reality capture?

AUDIENCE: Collecting existing conditions.

DANIEL CHAPEK: Yeah, collecting existing conditions-- certainly. Anything else?

AUDIENCE: [INAUDIBLE]

DANIEL CHAPEK: It is. Yeah, collecting existing conditions at a certain point in time. What reality capture does is it's creating a digital replica of the physical world. And we use that so that we can design in the full contextual understanding of the environment around us.

Now that's a pretty broad statement. And we have to understand that, when we're talking about it that way, we're not talking about a specific technology. Reality capture is a lot like BIM. BIM is not a technology. It's a workflow, it's a process, it's a mentality. Reality capture is similar. We're not going to buy a reality capture system. We're going to buy technology that falls under the reality capture umbrella.

So when we start understanding the technologies within the reality capture realm, we're looking at things like 3D laser scanning, which is predominately what we're going to look at today. But we can look at other things, like mobile LiDAR and aerial LiDAR. We talked about that before class a little bit. And all three of these are technically LiDAR.

What does LiDAR stand for? If I had prizes, I'd give them out. What was that?

AUDIENCE: Light Detection and Ranging.

DANIEL CHAPEK: Light Detection and Ranging, using light or lasers to measure distances. So when we're looking at aerial LiDAR, mobile LiDAR, or terrestrial LiDAR, which is 3D laser scanning, we're simply using different mounting platforms to capture distances.

When we look at completely different technologies, like photogrammetry, ground penetrating radar, Sonar mapping-- again, all of these are designed to do the same thing. We're taking a snapshot in time. We're measuring what's out there in the real world. So all of our traditional survey equipment fall under this banner. All of our GPS systems, all of our total stations. So there's variables obviously. Even if we go down to the tape measure, the pencil and paper. The sketch pad, when you walk out onto the site, and you take a measurement from your length and width of the room. We're technically doing reality capture. But there's a lot of differences in the technology, from your cost of equipment, time of collection, accuracies attained, so on and so forth. The tape measure from Home Depot's got a little bit different price point than an aerial LiDAR setup. I would think that's fair, right?

[CHUCKLING]

But if I wanted to use that same tape measure to go measure the entire topography of a county, that's not the right tool.

So we have to understand that as we look at the different technologies that are out there under the reality capture banner, we have to deploy the correct technology to the correct project purpose. There is not a right or wrong technology. There's not a better or worse technology. We have to understand the project purpose and deploy the correct technology to it.

Now when we look at the more advanced forms of reality capture, we're really approaching the way we measure the world differently. When we're looking at traditional-- tape measure and paper, traditional total station-- we're going out there and taking individual measurements-- Length and width of the room, height from floor to ceiling, or back of curb, edge of sidewalks, center line of road, all these individual measurements. And we've got to go back to the office and interpret what's in between, assume that based on the length and width of the room, this room might be square, or those walls might be plumb. We've got to make those assumptions. As we're surveying a cross-section of a road, and we survey it every 25, 50 feet, however far you want to walk, then we're assuming that that road is completely planar in between our cross sections. And again, depending on our project purpose, that may be absolutely fine.

When we're looking at more advanced reality capture, it's a 180-degree flip of how we traditionally look at it. Traditionally, we go out there and say, what measurements do I need? When I'm using scanning, or aerial photogrammetry, or something like that, I'm really capturing everything I can possibly see. And then I take everything I can possibly see back to the office, and I extract the measurements I need when I need them. So I gather everything I can see. I'm not focused on, I need to measure this, or that, or the back of the curb, or whatever. I'm not focused on that. I'm focused on capturing everything I can see. Make sense? Cool.

Now the product of a lot of these things is a point cloud. All right, so who wants to offer a definition of a point cloud?

AUDIENCE: A 3D photograph.

DANIEL CHAPEK: No. Normally I'd spin that and say, well, you're kind of right. But, no. Anybody else? Point cloud.

AUDIENCE: Is that taking a summary of all of millions of measurements?

DANIEL CHAPEK: Yes, absolutely. And I'm sorry to jump on you with the 3D photograph. But the photography is-- we're looking at images. And we're not looking at images with a point cloud. I'll get to it specifically.

But yeah, you're right. It's a summary. It's a collection of millions or billions of individual points. As the laser scan is going out, as the photogrammetric processes the photos into a point cloud, the point cloud is simply a collection of multiple individual measurements. So when I understand that, we have to know that what's on the screen here is a point cloud. Any idea what this point cloud is of?

AUDIENCE: [INAUDIBLE]

DANIEL CHAPEK: Well, it's not a black screen. It's a black screen on the projector. It's not on my screen. That's kind of hard to see.

AUDIENCE: [INAUDIBLE]

DANIEL CHAPEK: Yeah. Well what I'm doing with this slide that you guys can't see, but if you could see it, it would be wonderful, is that I'm upping the density of the point cloud. So in the first couple series, there's a point but they're very, very sparse. You can't really see what's going on. As I up density in the point cloud, you get a better and better understanding of exactly what's there.

So we have to understand that not all point clouds are created equal by any means. Again, there's a lot of variables in some of these different technologies we talked about. You know, a 3D laser scanning point cloud will be of a different quality than a photogrammetric point cloud, and that'll be of a different quality than an aerial LiDAR point cloud. So those variables of density, amount of data, again, the time of collection, the overall size of the point cloud, all those variables are going to come into play.

So when we look at the quality of a point cloud, there's kind of two factors that make that up. There's accuracy and density.

ADMINISTRATOR: Sorry, the fire marshal in the house, so everybody has to be in [INAUDIBLE]

DANIEL CHAPEK: Oh, OK.

ADMINISTRATOR: That's the issue.

DANIEL CHAPEK: That's fine.

ADMINISTRATOR: Yeah, sorry about that.

DANIEL CHAPEK: Hey, better safe than sorry. As a volunteer firefighter, thank you guys for taking your seats.

So again, the quality level of a point cloud is based on its accuracy as well as its density. And those two factors are not tied to each other whatsoever.

Accuracy is going to be the relative error of each individual point to the real world. Now that's going to be variable, but it's fairly constant, based on the quality of equipment or the type of equipment that's used to collect the data. So using photogrammetric data, I have a lot of error variables that go into that, as far as how often I've taken shots, what overlap my photos have, the height of my flight, things along those lines.

With 3D laser scanning, it's based on, am I using an imaging class laser scanner, or am I using a survey grade laser scanner? How does it approach tilt compensation for outside-- your environmental variables? All of those things are going to go to come into play. So accuracy is going to be based on the quality of equipment that I'm using to collect the data.

Density is a user-defined variable. When I'm out there with my laser scanner, I'm setting that I want this density. I'm telling the instrument the density that I need. And that can be variable on what I want to do with the data later.

Remember your job, Troy.

TROY: It's just one job.

DANIEL CHAPEK: Troy has one job. That's to make sure nobody kicks the tripod.

[CHUCKLING]

So again, density is based on a user-defined variable. What am I intending to use this data for? Can I do a quick scan of this room, and get length and width of the room? Sure. Do I need to accurately measure the trim work in the back of the room to replicate it? OK, well that's going to require a different density, and potentially a different accuracy. Make sense?

AUDIENCE: [INAUDIBLE]

DANIEL CHAPEK: I know. I was about to point to it, and then I'm like, wait, that's printed.

[CHUCKLING]

That'll still show up, but not geometrically.

So anyway, again, we have to understand that not all point clouds are created equal by any means. So as we take a look at different reality capture technologies, as well as different instrumentation within the same class-- different UAVs, different 3D laser scanners-- they will all have these variables of accuracies, and densities are going to be the user-defined variable that we can set. Make sense? Cool.

So how it works specifically with laser scanning is that it's going to pulse the laser if it's a time of flight laser. It pulses the laser by rotating horizontally and vertically. And with every vertical rotation, it moves slightly horizontally.

So think about again the density portion. You guys might have a peg board hanging in your garage. It's got holes drilled in it. Those holes are drilled in 1 inch horizontal, 1 inch vertical arrays. You guys understand what's I'm talking about? OK, cool. Peg board, check-- base level of understanding.

So we've got this peg board with 1 inch holes. If you put that peg board five feet from you, and you take your eyes from hole to hole to hole, your eyes are making an angular shift to get that density at that distance. You take that exact same peg board, put it 500 feet away, and if you can still see the holes, and you take your eyes hole to hole to hole, you have a much smaller angular shift.

So when we're looking at how the laser scannings work with that vertical rotation, it's essentially surveying a vertical plane. So every rotation is scanning the plane of that rotation. Then it rotates horizontally, then it shoots another plane, rotates horizontally, shoots another plane. So while it looks like the scanner is continuously rotating, it's not. It's rotating, stopping, shooting the plane, rotating, stopping, shooting the plane.

So the speed of that vertical rotation is based on how fast the mirror can spin. But the speed of the horizontal rotation is based on what density we ask for. That's setting that horizontal angular shift between points.

So as it goes around and scans all those planes, it's capturing everything it can see within the range of the instrument that we're using. And every pulse gets us a measurement, the distance measurement. The three-dimensional point is calculated based off the horizontal turned angle and the vertical angle, the direction the laser is being shot. This is crucially important because this idea of controlling the direction the laser is being shot is the difference between a survey-grade instrument and an imaging class laser scanner. Because every laser scanner is going to understand I'm shooting the point in this direction. And in this direction, I just measured the corresponding distance.

But the different laser scanners on the market will approach tilt compensation differently. Tilt compensation is how we take into account all of the different environmental factors that are enacting on the scanner. There's vibrations in the floor. The semi truck drove by. It's a little bit breezy. All of those things are enacting on the scanner, making the scanner move. And any movement of the scanner is going to adjust the actual direction the laser is being shot.

So imaging class laser scanners like the BLK 360 use electronic tilt compensation, or tilt monitoring, depending on, again, which manufacturer we're talking about. But the BLK is using an IMU, an inertial movement unit, which has basically a circuit board of sensors with an accelerometer, a tilt sensor, things along those lines, to measure how it's moving in the real world. And it's looking at those measurements to adjust these values of horizontal vertical angle to correspond correctly with each individual point. When we get to survey-grade instrumentation, it gets a lot more accurate. Because a survey-grade laser scanner is going to use a fully liquid dual-axis compensator, where you've got a little puck of oil that's rigidly mounted to the cast metal frame of the scanner. If you look at a Leica P30 or P40, for instance, that dual axis compensator is accurate to one arc second. You guys familiar with arc seconds?

AUDIENCE: [INAUDIBLE]

DANIEL CHAPEK: Some. In other words, circles.

AUDIENCE: Like the angular seconds.

DANIEL CHAPEK: Right, yeah. So if you're not familiar with arc seconds, it is just a unit of measure. If you take a circle that's 360 degrees, if you take one degree, divide it by 60, you have an arc minute. If you take that arc minute, divide it by 60, you have an arc second. So we're really just talking about a very, very small angle.

To give you an idea, if you extrapolated it out, the degree of one arc second is about the diameter of a dime a mile and a half away. So we're looking at very, very accurate.

So when we're looking at a professional unit with a dual-axis compensator like that, it's looking at those angles that accurately. So any breeze or vibration in the floor if we're in that manufacturing plant and the press is stomping away, anything that's going to shake the direction of that laser is going to be tracked, and these angular measurements are going to be updated to make sure that the direction value the laser is being shot is correctly corresponding to the distance measurement that was taken in that direction.

And the horizontal angle, the vertical angle, and the distance combine to give me my actual coordinate. So that coordinate that's calculated, when we talk about accuracies of scanners, we're really taking a look at the accuracy of that coordinate versus what's truly out there in the real world.

So we can talk about these individual accuracies. Scanners will have specs of horizontal angle tolerance, vertical angle tolerance, ranging tolerance, range noise, all those different accuracy levels on there. But they all need to be summed up to 3D positional accuracy. How accurate is that coordinate versus the true real world?

All right, so as it does that, it captures the point cloud. And the point cloud is a three-dimensional object built up of millions and billions of individual points. All these scanners also have cameras in them as well, which will take full dome photography to then overlay-- and overlay is not the right word-- but colorize the point cloud based on those photos.

So with the horizontal and vertical angle that we were talking about, this scanner understands what direction each individual point was taken in. It also understands what direction each individual pixel is in in the photos. So then it just correlates. It says that individual pixel was taken in this given direction. We'll apply that single RGB color to this individual point.

All right, so when we're talking about roundtripping, now that we got the basics of laser scanning down-- and for those guys that have been laser scanning, thank you for being patient. But when we're talking about the roundtripping of reality capture data, typically when we talk about reality capture it's all about taking the field into the office. We're taking the reality of the physical reality, and we're bringing it in digitally for our design purposes. But when we're talking about roundtripping it, we want to take the office, we want to take our accurate design, and put it back out into the real world for construction.

How often do you guys hear-- anybody in construction? OK. This is a pretty targeted glass to that industry, apparently.

[CHUCKLING]

So yeah, when I'm looking at construction, how often do we hear that there's a break between the office and the field? The plans are beautiful, they look great, but it's not constructable. Or it's not constructed the way the plans are set up.

This kind of alleviates some of that. If I can actually take my designs, and put them in the real world, and my design is constructable, then the real world should be constructable. So that's what we're looking to do here. We're looking to bridge that gap between the field and the office. So if we initially do reality capture to understand exactly what's there, we can design based on exactly what's there. We can then take our design and put it out into the world. So that it can accurately be constructed. As we're going through that, we can scan it again, iteratively, to ensure that it was built per plan.

So as certain trades go in there, as the steel goes up, we can scan it and verify, validate that that steel was built correctly. And if not, we can get ahead of it. We can either get that steel fixed before it causes problems downstream, or we can adjust our plans, and give everybody updated things so when the HVAC goes in there, they're expecting that beam to be there. As we look at it here, we've laid it out. We've built part of it. And you can see that this part of the point cloud, that duct is in completely the wrong spot. And I can then determine if I care or not. Is that going to interfere with anything else downstream? If not, OK. If it will, then I need to get ahead of fixing that. And it's all about fixing these problems at the cheapest point possible, getting ahead of it before there's a sub on site saying, well, I guess I'll just stand here all day because I can't work now because that's in my way. They're still billing. They're just sitting there, not doing anything.

So we want to be able to, again, bridge that gap between reality and digital, to be able to make sure this is designed and built accurately.

So now we want to go ahead and get into the scanning portion, the reality capture side of it. And we want to go ahead and look at the BLK 360 and ReCap. So we are going to take a quick break from PowerPoint, and we're going to talk about the BLK for a second.

So for those of you who have not seen the BLK, this is all it is. This is everything we'd have to take out to the site to go scan. And as a guy that's been lugging around tripods and larger scanners for over a decade, this is refreshing. I just take this through TSA, and I fly wherever I need to go. But in this bag, it's got a couple of things. It's got the scanner itself. And you pull the scanner out-- little case here. The scanner is incredibly small. Again, this is an imaging class laser scanner. So we're looking at about 1/4 of an inch at 30 feet error. We're looking at 5/16 at 60 feet or so.

So while it can see 60 meters, its happy spot is going to be under that 10, 20 meter range. So it really is intended to be internal-- architectural interiors, things like that. It can go outside. It's IP54-rated. With the range that it has, it's much more intended to be more of an interior scanner.

The tripod just folds out like so. Quarter turn, expand it. So we'll do that. And then the center post will just compression fit in there. And this center post is great because it's just a quick-connects onto the scanner like so. And I can put it upright. I can also hang it upside down. The IMU inside of it is going to understand whether it's right side up or upside down.

So we'll set it there. Oh, it needs a battery. That's important. It needs more power, Scotty.

So in here, we've got the battery door. Just put the battery in there. And as we were joking before, there's only one button. So I'm going to hit the button, and it will go ahead and boot up. Now as it's booting up, it's going to go ahead and turn on.

And part of that is that it's going to be its own Wi-Fi hotspot. Right so it's going to emit its own SSID where I can go ahead and connect to it. Once it goes solid on the LED, than it's booted up. There we go. Jump into my settings. Just want to make sure that my Wi-Fi is on. You can see I'm connected to the BLK.

And I'll just go ahead and jump into ReCap Pro. So on the iPad, I get an interactive interface with how I control that. The iPod is not necessary to run the BLK. I can walk over there and hit the power button once, and it would go ahead and scan. And I can scan onboard all day long, get back in the office, and pull the data off then. And depending on how experienced I am with scanning, I may prefer that, I may not. If I'm new to scanning, then the feedback I'm going to get on the iPad is really helpful to make sure that I've got good coverage, I've captured everything I need to see, all that kind of good stuff.

So in this case, I'm going to hit New Project. We'll name it Class, because I'm feeling creative this morning. I'll hit Capture. And what that does is that kicks me into the actual project. And we can see that I'm connected to my BLK. See the firmware I'm running. There's 53 gigs left of storage on the BLK itself. As well as there's-- it's going to sync over. If I look at the settings, it's going to sync over to my iPad as well.

When I look at the settings, you can see that I've got scan quality. Again, the scan quality is going to be talking about density. So again, what is going to be the angular difference between every vertical rotation of the laser? Now I've got low, medium, and high. Part of the BLK is that it's incredibly easy to use. The ease of use is phenomenal with the BLK. But a lot of times, when we up ease of use, we drop applicable settings. So again, as we take a look at an imaging class scanner versus a survey-grade scanner, when I'm working with a survey-grade scanner, I can set up, I can occupy known points, I can traverse, I can do all of that stuff. I can have specific settings on what I want my density to be. I can do all of those types of things. The BLK does not allow me to do those things. The BLK's ease of use is simply made for me to go ahead and turn it on, and hit Go.

So I can set the different-- low, medium, or high. I can tell it whether I want photos or not. And then I'll just go over here to the right, and I'll hit New Scan. And as I hit New Scan, it goes ahead and starts.

So it's first going to start by doing a 360-degree rotation to calibrate its tilt sensor, just to identify how out-of-level it is. And then it's going to start by taking photos. Now as it takes photos, you're going to see-- smile at him, because it's looking at you. You're going to see there's three lenses. And each one of those lenses does have an LED on it. So if we were in low-light situations, it would brighten it up, and the photos are pretty decent.

Those photos are being populated in real time over here. They should be.

AUDIENCE: [INAUDIBLE]

DANIEL CHAPEK: Yeah, it did. Way to go.

AUDIENCE: [INAUDIBLE] best used low light [INAUDIBLE]

DANIEL CHAPEK: Ambient light really doesn't make a difference. So the laser scanning side-- the laser is its own light source. So it can be used in complete darkness. The ambient light portion is only going to affect the photography side of it. So the point cloud that's captured isn't dependent on the ambient light whatsoever. But the photography, to colorize the point cloud, make the color real, that's going to be where ambient light comes into play.

We'll just kill it and we'll try it again. There we go. New Project, and New Scan.

AUDIENCE: [INAUDIBLE]

DANIEL CHAPEK: Yeah, it gets mad at me. That's the technical term.

So I've checked the Wi-Fi. And we will go in here. Class 3.

[CHUCKLING]

Third time's the charm. There we go. New scan.

So again, it's going to do that tilt calibration, start taking photos, and then it's going to start to scan. So again, some of the variables we have in there with the BLK, again, the ease of use part of that is just low, medium, or high, what density are you looking at. And for the most part, I do everything in medium. If I go outside, or if I'm doing a large foyer or something like that, I might kick it up to high. But medium is typically where I'm going to be. If I go into an electrical closet, I'll kick it down to low. But medium, for the most part, is where it's going to land.

And again, the accuracy is a constant based on the quality instrument. So again, we're looking at 6 millimeters at 10 meters-- 1/4 inch at 30 feet-- and 8 millimeters at 20 meters, so 5/16 at 60 feet. So that's going to be a standard, regardless of what we're scanning or where we're scanning. Again, because of the electronic tilt compensation, if I do get into an area that has a lot of vibration or it's really windy, then I need to make other considerations to it. But in an environment like this, that's kind of the accuracy I can expect.

AUDIENCE: [INAUDIBLE]

DANIEL CHAPEK: I'm told it can. I've seen some guys that make different rigs to strap it to the side of a column, something like that. I've not done it personally. But in theory, the IMU is supposed to track that, and it's supposed to be able to do it. But I haven't done it.

AUDIENCE: With the tilt compensator, do you have to set the [INAUDIBLE] level at all, or will the tilt compensator just detect how much you're not level?

DANIEL CHAPEK: So this doesn't have a tilt compensator. So this has the IMU, the Inertial Movement Unit. So being that that's all digital, then no, you don't have to put it level at all. If I was using a professional-grade scanner, then I would. Because I want to level it within reason so that it knows where plumb is.

AUDIENCE: [INAUDIBLE]

DANIEL CHAPEK: Right. So that dual-axis compensator is accurate to one arc second, but it has five minutes of play. So it can move around quite a bit. When that semi drives by, and the gust of wind almost knocks it over, it's going to check that. And it's actively compensating within 5 minutes of play, to an accuracy of 1 second.

I was in a mine down in Springfield where, in this mechanical room, there were five 5-foot diameter turbines that were sucking all the air out of the mine to cool this mechanical room, and then go out in a shaft. And I set up on the backside of one of those turbines. And in that shaft room, they said it was sustained 30 to 40-mile-an-hour winds. And my survey-grade scanner didn't care. It compensated fully for all of it. And I was right, directly, lined up with it. It wasn't ideal.

AUDIENCE: [INAUDIBLE]

DANIEL CHAPEK: Yeah, exactly. But that's one of those considerations where it's the correct technology for the job at hand. I would not take this guy into that scenario. I mean, it's great that it's only 2 pounds, but that also means a lighter wind can blow it over. There's a reason that the professional-grade scanners are bigger and bulkier than they are. They can make total stations smaller than they do, but there is a reason that they're not. And that's because it has that mass, and it can maintain that accuracy with those outside conditions on it.

OK, so it's done. So you can see it. It was populating those photos in here. Smile, man. So serious. Oh yeah, you guys all signed the form of consent before you walked in, right?

AUDIENCE: No.

[CHUCKLING]

DANIEL CHAPEK: So what it's doing now, once it's scanned, it transfers the data over. So you can see now it's saying Processing. But once it's done scanning, I can go ahead and pick it up and move it. And the thing about laser scanning is-- a lot of guys know that have been scanning already-- is this all about point of view. It's all about line of sight. So as the scanner was here, you could see everything from that viewpoint location. I'll just hit New Scan again. So we could see everything from that point of view, which means as it was sitting here, it couldn't see on the backside of the table. You guys were all blocking my view of this wall down here at the bottom. It saw up there completely. So as I'm scanning throughout the room I need to take that into consideration.

Now in this case, we're not going to get a clean data set because we're in here. But if we were out in the hallway, and a few people were walking by, I can clean those people out. I can look around people. I can look around traffic. I don't need to shut down the interstate to be able to survey it. I don't have to make sure that that hospital facility is evacuated before I scan it. I can do all of those things while these environments are in use.

So again, it's going to scan this setup. And once it does that, it's going to again transfer the data over, and go through the registration process. Now the registration process is the alignment of the data. So again, every individual setup location is a confined set of data. As it's capturing each individual point, we understand that each point was calculated by the horizontal and vertical angle from the eye of the scanner. So each setup has an origin, that 0, 0 point, at the eye of the scanner. So setup 1 has an origin, 2 has an origin, 3 has-- each individual setup has its own origin. And the registration process aligns all of those individual setup locations so that I have one overall point cloud at the end of the day.

And that registration process is crucially important. The alignment of this data is probably the most important step in this entire process. It doesn't mean it needs to be difficult, but it does mean it needs to be right. Because any misalignment in the point cloud is going to give me error in the results. So if I've got points on this wall from setup number 1, and I've got points on this wall from setup number 2, and they're slightly misaligned, then when I look at the point cloud, if I slice through the point cloud, I'll have two parallel lines of points-- some points from setup 1, some points from setup 2-- and they're offset each other because the two point clouds aren't aligned to each other.

So I need to be able to cut that slice, and have one crystal clear line of points for exactly where the wall is. And again, that is going to go into everything else that I need to do with this point cloud. All of my model extraction, going in there and having the software draw the 2D floor plan for me, for it to take a look at that point cloud, and model that piece of structural steel for me, extract the piping for me.

Well, if these point clouds aren't aligned correctly, it can't do that extraction because it doesn't know where the wall is because there's two lines of points instead of one. So the registration process, again, is crucially important. But it doesn't have to be difficult, and it doesn't have to take a lot of time. And what you'll find is there's a lot of different ways to do the same thing when we're talking about laser scanning. There's different pathways, different parallel workflows, depending on what you're doing and how you want to do it.

So we talked about the BLK, and I'm using the iPad with ReCap here. And in this workflow, I get feedback from the BLK. I can see-- and I'll show you here shortly what we can do with this data while we're here in the field using the iPad, but that's not the only way to do it. Again, all of this data is being written directly on the BLK. So I could not use the iPad, and I could just run around, hit the button, and it would scan each individual setup, take it back to the office, and use more advanced registration software if I'd like to.

There's a lot of different options to use depending on how big my project is, how many setup locations I have. This process here, where you can see it's scanning setup number 2. It's done scanning. It's processing it now. It's going to register it on the fly here. Well, if I was going to move on, I would been paying attention, and as soon as it got done scanning, I would have gone out the doors and put the scanner out there. I always want the laser firing. If I'm out there in the field, I want to keep actively scanning. If the laser is not firing, I'm wasting time and I'm wasting money.

So what you'll find is that this process of copying the data over, doing registration in real time, if you have a large project, you're going to outpace it. I'm going to basically be out there-- if I'm on setup number 35, 40, 60, then by the time I'm done scanning setup number 60, it may only be done transferring and registering setup 35, 40. So if it can't keep up, then I may want to just not worry about it at all, and do the registration back in the office.

So again, it's going to change based on my level of comfort with the laser scanning process, and what other software I have available to me. So you can see, it's going through registration now on setup number 2.

AUDIENCE: [INAUDIBLE]

DANIEL CHAPEK: I could have done that a while ago. As soon as it's done scanning, I would move it. So when it's done scanning the photos and then the scan, then it'll say Transferring, then it'll say Processing, and when there's more than one it, goes through Registration. As soon as it's done scanning, when it goes over to transfer, I can move it.

AUDIENCE: [INAUDIBLE] to have it register on [INAUDIBLE]

DANIEL CHAPEK: Right, yeah-- no, if it's coming over with the ReCap app, it's going to register on the fly. So what it's done here is it says that it's registered. I can click on Confirm Results, and I can see this preview of the room, where again, setup number 1, setup number 2, one's red, one's bluish. And I can just visually make sure that they're aligned. I've got my standard ReCap reporting that 99.9% of them are within a quarter of an inch of each other, which is what I would expect. And a lot of times, why aren't they 100%? Well, because Adam moved. And it's all Adam in both setups.

So the more vegetation that I may have, the more it might think that that tree is the same in both setups. But that tree was blowing in the wind. So it thinks they're overlapping points, but because it was slightly moving, it may be causing a higher error than I think it is.

Go ahead.

AUDIENCE: [INAUDIBLE]

DANIEL CHAPEK: Not notable-- not notable that I have identified.

AUDIENCE: I've been told not to use the iPad for the registration process.

DANIEL CHAPEK: What you're going to see is that when I connect to it, and bring the data over, the desktop app is going to check that registration, and tighten it if it can. Because the iPad is less of a powerhouse than my laptop, so it is doing the registration with the same ReCap routines. But as far as accuracy or iterative nudging to make sure they're tighter, that is looked at on the desktop as well. Go ahead.

AUDIENCE: So when I move the scanner [INAUDIBLE]

DANIEL CHAPEK: Yeah, you could do that. And again, that's actually a pretty good workflow because of file size limitations that you might experience with ReCap. Normally I would go through and I would scan the entire building. I'd scan up down the stairwells. I'd have all the project in one set of data. Because with other software, I can. I can have files that are incredibly large, and I don't have any performance issues.

AUDIENCE: [INAUDIBLE]

DANIEL CHAPEK: You could. You could. I mean, I could take it through the stairwell. And as long as I've got overlap from setup 1, to setup 2, to setup 3 through the stairwell, I could do it.

Yep.

AUDIENCE: [INAUDIBLE]

DANIEL CHAPEK: The thermal imaging? Right. That should be out, with another firmware, first quarter next year.

There was a lot of questions here. You were next.

AUDIENCE: [INAUDIBLE]

PRESENTER: Yeah, cumulative air, you mean? Well, again, it depends on what we're doing and why. If I'm worried about cumulative air and I've got a long way to go, maybe this isn't the correct instrument. And I think I was mentioning earlier, if I was doing a hospital or a school-- I think I was talking to you about it-- I would take a survey grade instrument and do the hallways, the main corridors, and drop a BLK in each room. Because cumulative air is going to be there.

Every time I registered, there is going to be a misalignment. It should be minimal, but there's going to be that. And yeah, we do have the potential of cumulative air if all that air is in the same direction the whole way through. It's--

AUDIENCE: [INAUDIBLE]

PRESENTER: Certainly. Yeah, because at the end of the day, a point cloud is a point cloud. So there's variables and accuracy and time of collection and costs, all that kind of stuff. But they're all just collections of millions or billions of individual points. So using different softwares, we would be able to basically combine data from any source. UAV photogrammetric data with terrestrial LiDAR data-- all of it goes together, we just combine it and work with it then. OK, other questions right here? Go ahead.

AUDIENCE: [INAUDIBLE]

PRESENTER: It is track-- right now, on this one, the IMU is not activated. Again, that'll be in the firmware here shortly. But it is going to be trying to track your movement lightly. But it's still using the overlap for registration. The IMU is going to get your initial alignment direction you went and the level of it. And then, cloud to cloud registration is what's going to tighten it together.

AUDIENCE: [INAUDIBLE]

PRESENTER: Yeah. Yep, absolutely, especially if we have setup 1 and 2 like we did, and then I invert it for setup 3, the IMU is going to tell it that. So the IMU is going to understand, you flipped it upside down. So the data will register automatically even if it's inverted. OK, go ahead.

AUDIENCE: So how much overlap [INAUDIBLE]

PRESENTER: Well, the overlap between the two setups is based on you as a user. I had this scanner here. I picked it up and I walked over there. The further I walk, the less overlap there'll be. So it's a matter of where I put the setup locations. And the different environments that I'm working in will be variables there as well. If I'm in a very complicated mechanical room, I need to stick it in back corners and around different equipment. And I may have less overlap in those areas.

But that complexity will just drive me to have more setup locations, so that I have more overlap in areas that are blocked-- things along those lines. Go ahead.

AUDIENCE: [INAUDIBLE]

PRESENTER: Absolutely.

AUDIENCE: [INAUDIBLE]

PRESENTER: Absolutely. Yeah, so the question was, do I see different registration results between ReCap and Cyclone-- like a Cyclone. And yes, there is a number of different results, inaccuracies and things. And a lot of the big ones-- it just comes down to reporting. And the way that ReCap reports the accuracy is more of a go-no-go switch, where 100% of your points are within 6 millimeters. That doesn't tell me how off it is, that tells me how on it is, right? Does that make sense? Right? So in Cyclone, I actually have reporting mechanisms that tell me, this is your error, not how much percent is better than a threshold, right?

Also your registration techniques in cyclone are much stronger, where ReCap is going to focus solely on cloud to cloud, where target based is going to strengthen that cloud to cloud. In Cyclone, I can do traversing, I can do resections, I can do all that normal survey stuff, as well as automated targeting, and then, cloud to cloud to strengthen it up. I mean, at the end of the day, if I was in Cyclone, I could model this plane of the wall in two setups and say that those modeled planes were coplanar. And that would do the registration.

So I've got a lot more flexibility in Cyclone versus ReCap as far as registration goes. Also time wise. So when I'm looking at things like ReCap versus Cyclone register versus Register 360, I did some benchmark testing. I had 20 setups out of a P40 that I put into ReCap, and it took 4 and 1/2 hours to import-- not register, just import. I brought it into Cyclone-- that was imported and registered in an hour and a half.

I brought it into Register 360, and that was imported and registered in 13 minutes. So there are distinct variables or differences in some of these software's. To give you another, I can show you a different project. It's a mechanical room with a BLK. It was four setups. In ReCap, the four setups was 2.4 gigs. In Cyclone, it was 1 and 1/2 gigs. In Jet Stream, it was 400 megs. So there are significant variables as we take a look at some of these different software's that apply to laser scanning.