Description
Key Learnings
- Learn about photogrammetry workflow in ReCap
- Discover the powerful point cloud tools within InfraWorks
- Learn how to prep and manipulate surface data inside of InfraWorks
- Learn how to transfer data into AutoCAD Civil 3D
Speaker
- JCJosh ClawsonJosh Clawson is a Applications Engineer (Civil Industry) for MasterGraphics, Inc. He has been in this position for the last year. Prior he had spent 5 years as acting CAD/Standards manager for the City of Grand Forks Engineering Department where he maintained templates, instilled standards and trained staff on AutoCAD Civil 3D, Map3D, Infraworks and AutoCAD. Josh is an Autodesk Certified Instructor and a Civil 3D Certified Professional.
JOSH CLAWSON: Thanks everyone for attending this presentation. It's going to be utilizing drone workflows for civil workflows. So before we get going, just a little bit of housekeeping. Turn your phones on silent. Hold questions until the end. We'll have a question section at the end and then just do the surveys. That's how they give us feedback, if we're doing a good job presenting or not.
So who am I? My name's Josh Clawson. I work for a company called MasterGraphics out of Wisconsin. I'm originally from North Dakota. I guess I've been at MasterGraphics for just over a year, about 16 months now. Grand Forks is actually where I was before. So I was a CAD manager for five years prior to that for the engineering department for the city of Grand Forks, North Dakota. And then before that, I kind of cut my teeth with the Autodesk stuff doing production drafting for a precast company, just production drawing.
So a fun fact about myself, just figured I'd break the ice with this. Three years ago, me and a buddy decided to start a fishing TV show and I still periodically co-host this TV show on mainly ice fishing. We go all over the upper Midwest. So we're just getting ice in North Dakota now, don't have it in Wisconsin yet but something fun to do. Anyway, I thought I'd throw it in there.
AUDIENCE: [INAUDIBLE]
JOSH CLAWSON: Yeah. So anyways, enough about me. What we're here for. Drone workflows utilizing the AEC collection for civil workflows. What does that really mean? This entire process stems around one word, and that word is photogrammetry. So to be able to utilize a drone for any type of data collection, for the most part, we're using digital cameras. And we get images and we've got to turn those images into usable data. So photogrammetry, what does it mean? Well as it is defined, it's the process of making surveys and maps through the use of photographs, especially aerial photographs. So let's break down the word. Photo means light, gram means draw, and metry means to measure. So we're going to draw and measure with light or photos, photogrammetry.
So how long do you think photogrammetry has been around? Going to take kind of a look back in history here, and I'm going to really not pronounce these names correctly, but-- So the theory of perspective, Leonardo da Vinci came up with that theory. That from a single point, you can draw lines to measure things. Then we had what was called projective geometry or drawing by perspective. And from there, the first practical photograph was taken by that guy. Before that, it took photos like a four-hour exposure time. It wasn't just a [CLICK] go. It'd need four hours of exposure time, so the first practical photograph.
From there, the term photogrammetry was adopted in 1840. And then we had the first topographic map from terrestrial photos in 1849, so the process of photogrammetry was originated from map making. And then we had the first aerial photograph by hot air balloon. Anybody ever heard of Nadar correction like when you're processing survey data? Same guy. And then finally in 1862, photogrammetry is an accepted form of map making by the Science Academy of Madrid.
So there's some development cycles with photogrammetry. It's grown, it's matured over time. First we had plane table photogrammetry. Now that was usually terrestrial photographs from a single point, take a bunch of pictures and then put them together. From there, we went to analog photogrammetry. Now, there's one invention in the early 1900s that really stemmed this. Anybody know what that is? The airplane. So now we can get a camera in the air and fly grids. And also they utilized the stereoscope for arranging the photos after the fact. Now we have analytical photogrammetry. This is where we start utilizing ground control so we're putting some survey data to it, actually geolocating where this set of images take place, and we still use that today.
But what we're going to talk about today is digital photogrammetry. This, we're just getting started. It's fairly new technology. Digital cameras haven't been around like forever, but they've been around for 20 years. But we use digital photos, we use drones, we use pixel to pixel mapping engines, and we get our results fairly quickly. So digital photogrammetry today. Our most common aircraft's going to be a UAV or a drone. You can also do terrestrial photogrammetry, something you'd see more in 3D printing where you're replicating an item. So we're using smaller, more affordable, higher resolution cameras where we can get them up on a small aircraft, something more affordable. We're utilizing flight control software which is kind of amazing in its own right. You send your path or you map out your mission and you send it to the drone and you say go. You don't touch it. It goes up, flies its mission, comes back, and lands exactly where it took off. I think it's pretty amazing. Then we also utilize software too that automates a stitching process. It's automatic. You submit it, it goes up into the cloud, and you get back your results quickly.
So overview and goals for the workflow I'm going to show you today. Now the goals, we're going to use a consumer grade drone to collect surface data. Consumer grade, nothing-- I mean this is under $1,000. We're going to generate an orthomosaic or a georeferenced image of our site, so we have extremely up-to-date aerial photos for use in our design or any exhibits. But the most important one is we're going to generate a usable TIN surface that we can put into Civil 3D and it will function just like any other surface that you've ever used.
So the workflow, we're going to collect the images with a drone. And this is why we're doing the AEC collection, we're going through four different softwares. We're going to process those images in a software called ReCap Photo. That's fairly new, it's only been out for six weeks, maybe two months. We're going to take the exports that we get from ReCap Photo, in this instance, it's going to be an RCS, that's a ReCap scan. We're going to bring it in to ReCap Pro if we need to do any editing, it's an optional step. In this particular scan, we had some water and it reflected really bad so we need to clean it up. Then we're going to take that edited RCS and dump it into InfraWorks. Inside of InfraWorks, we can take a point cloud and generate terrain data out of it. So once we have that terrain data, we'll retrieve that terrain data in the form of a TIN surface inside of Civil 3D.
So it all starts with images. This is just kind of an overview of recommended things on site when you're going to collect with a drone. What is needed? At least three visible ground targets with survey data for each of those ground targets to get a proper stitch. You want to use flight control software. I've ran into a handful of people that they try and fly manually and take their images. You're not going to get a proper overlap and it's going to be inconsistent. You're not going to get a desirable result essentially. 80% minimum overlap on the photos, that's just what I do. I've heard some go in as low as 65% and I've not had as good of results. Now in this example, we're using a 12 megapixel camera, nothing super high fancy. So we went with 80% overlap and we've got really good results with an 80% overlap. And if you want less images, fly higher. You get a bigger footprint on your actual image capture.
So some things to consider about this if you are going to fly a site. Vegetation's a big one. We're drawing and measuring with light or photos so we can't penetrate whatever's on the top. So if we've got bushes or trees or some sort of canopy, we're not going to be able to get the ground below it with photos. Wind. I mean, I'm from North Dakota. 45 mile an hour winds are not uncommon. Not a good day to fly. Sun and sky. If you've got varied cloud cover going over, you're going to get a little bit of an inconsistency in your orthomosaic where you've got lighter and darker areas. Just something to consider. But the legality is a big one. I'm not going to dive into that. There's enough information on the web about the legalities, but just proximity to airports and residential areas are two things you've got to kind of really watch out for. You're not supposed to fly over anyone that doesn't know about it, and then airports is a big one. I think it's within five miles of an airport tower.
So let's see if this plays. This is the drone we used. It's a DJI Phantom 3 Standard, 12 megapixel camera. I think it retails for $499 now so incredibly cheap. And like I said, we used DroneDeploy on an iPad. And with just your finger, draw where you want and go through a couple of settings. Tell it you want this much overlap and to fly at this altitude. We flew, I believe at 110 feet for this one. And you hit the button and say go. It takes off on its own and flies its grid, comes back, and lands. I mean, there's no touching the controller for the drone. This is all just point and click, go. And it comes back and lands exactly where it took off from. That's pretty cool.
AUDIENCE: [INAUDIBLE]
JOSH CLAWSON: We did not. The question was, in this, did we do an overlap so we would run the grid, let's say north-south back and forth across the site and then go east-west? We did not on this one. Depending on your sensor and everything, there's different ways to go about it. I've had good results just doing an 80% overlap. If you wanted to spread that out and then also do a transverse grid pattern, yeah. I mean, there's different ways.
So the images. It all starts with images. The image results-- With that particular drone, that DJI Phantom, the photos have embedded GPS coordinates and altitude, just an x and y. Actually I think it's a lat long, but that helps once we get it into Recap Photo. So our ReCap Photo workflow, anybody recognize that little landing page? That's ReCap Photo. It used to be ReMake, prior to that it was Project Memento. It's actually been around for a while, but they added-- whoops. Laser. They added that button. That's the aerial one that you can do now and that's ReCap Photo.
So we're going to import the images from our drone flight. We're going to geolocate those images from our ground control points or our GCPs, which is our survey data. We're going to select a data format that we want to retrieve. And we're going to submit it to the cloud for processing, freeing up your machine. You can submit as many as you want. It's all processing in the cloud. From there, well actually, yeah. Where do you get ReCap Photo? If anyone has ReCap Pro right now and they're on subscription, you have ReCap Photo. On the landing page for ReCap Pro, you'll get this little instructions on how to retrieve ReCap Photo.
So the ReCap Pro workflow is we're going to import that point cloud file that we get out of ReCap Photo, and that's going to be that RCS file. We have the option also to combine multiples. So if we reach our photo limit and we want to have a larger scan area, we can combine those scans inside of ReCap Pro, or we can also edit that point cloud. And then the last step inside of there is to export out your edited RCS, and then we will push that into InfraWorks. So another note, has anybody noticed you can't get LiDAR into Civil 3D now, after I think '17 release? ReCap Pro will accept many different point cloud file formats. Like the list is huge when you go to the files of type. Bring them into ReCap Pro, export them out as an RCS, then you can push them into Civil 3D or InfraWorks.
So the InfraWorks workflow is we're going to create a terrain model. We're going to import-- this is optional, the ortho for use. So we could do in our InfraWorks model, utilize the aerial photography so the aerial photo that we collected with our drone. We're going to classify a point cloud. So InfraWorks has-- it's a type of classification that does on your point cloud. It will recognize if it's ground or if it's a sign or trees. It's kind of neat. It's not perfect. It's not an end-all fix but it is very beneficial and I'll show that. But then we'll generate our terrain surface from the point cloud. And it's going to parse out just those ground points inside the point cloud. And that's going to replace the existing ground inside of your InfraWorks model with your new ground data from your point cloud scan from your photos that were processed.
And from there, we can retrieve that InfraWorks terrain model inside of Civil 3D. There is an option just to open your InfraWorks model inside of Civil 3D and then you can also pick what items you want to push into your drawing from your InfraWorks model. In this instance, we're only after that terrain surface. We can define boundaries on it, but this is another big one. Being that it's from a point cloud, a lot of data points in it. More data points are more accurate but it doesn't look like a traditional survey or traditional surface. It's a very tight triangulation on the surface. If you do want to reduce it down, we'll run through how to do that as well.
And then we can create data shortcuts for downstream use just like we would any other surface that you're collecting with traditional survey. So the results, what we're going to get. Up-to-date aerial imagery of the site. We're going to get that orthomosaic, georeference, orthomosaic of our site. We're going to get a point cloud. We can do other things with a point cloud than just create terrain data out of it, and the big one is we're going to get that TIN surface. Now this snapshot is after it's been reduced down, the triangulation on it is going to be much higher prior to that. So we're going to go live here and hopefully everything works.
So first, we're going to take a look at the photos. So this is the photos that we captured on that particular site. It's an active construction site. If I go in and take a look at one of these photo's properties-- Actually, one second. And we go into the details of these photos. You come down and actually see that there is GPS coordinates embedded into these photos. Now this is going to make more sense why this is desirable in the next step here. So these are the photos. And we're going to jump into ReCap Photo which is this software. This is the newer one, used to be ReMake and used to be Memento prior to that. So what we want to do is we-- Excuse me, I've got a frog in my throat.
We want to start an aerial project. Two different types of projects, aerial or object. And again, that object is for things like the space mouse that they have there. So select to create an aerial scan or aerial project. You really only have two options inside of here. So we select to add photos and we're going to grab that set of photos we just looked at and actually just select them all and hit open. It's going to populate them. Notice I have 760 photos in this scan. Anyone familiar with ReCap 360? In the past, 250 photo limit. Now it's 1,000 but it does cost cloud credits if you see that up there. And the amount of cloud credits that it costs is relative to the number of photos you have. The less photos, the less cloud credits. I don't know what the exact correlation is but I know it increases the more you get.
So we've got the photos in here, now we need to geolocate them. Before I do that, I'm going to jump into Civil 3D. This is just the Bing imagery of a site and we have some ground control shots, just some control points on site that we are going to use. This is just to show the x, y, and z. Now I made-- this is a custom label type x, y, and z, easting northing elevation. The plug-in inside of ReCap Photo is x, y, and z, so this is just something I did to make it a little easier. So what I'm going to do is go back into ReCap Photo and we're going to apply some ground control points to this.
So up at the top, you can set your coordinate system. This is big. We didn't have that before in ReCap 360. You had to convert everything to metric. So we can set our coordinate system and I'm going to add a couple ground control points. And just so I can see this on one screen, I'm going to make this a little bit smaller and go into my properties. Whoops. I don't need to go into my properties. So I want to add in that x, y, and z for each of these. And I'm not going to do this for every one of them, I'm just going to show the process. So our x, 834105.07, and then tab over to get to the next field. Our y is going to be 540441.99. And then notice as soon as I hit enter off of the x and the y, my number of photos reduce down. This is why it is a benefit to have that GPS data embedded into your photos. So now I've only got 10 photos to choose from instead of 760 to try amd locate this particular GCP. And I'll enter in the elevation here real quick and then we'll continue on. 922.347.
So from there, I've got this GCP selected. I'll come out and select a photo. Let me make this a little bigger here. And I'll select a photo and what I'm looking at or what I'm looking for is this control point, just on a pavement, a little triangle. I didn't actually place ground targets on this one. So select that control point, I can refine it from there if I want and I'll say done. All my photos disappeared. This is the one step I think they need to improve on. I've got to go back out to my entire set of photos the first time. After I select one, it'll keep me in that area. You need four photos for each ground target, so with the three ground targets I had on this site, it was 12 photos that I needed to locate ground targets in. I'm just going to find this one again here real quick. That's that same point. Done. Go in and do a couple more, same point. Done. I'll do one more. Like I said, you need four. Done and then I'll say done.
So here it shows me the four photos that coincide with this GCP. I'm not going to go through that process three more times, I think you guys get it. So we'll say done, and then we'll say create, and this is going to kick this up to the cloud. What we have some options here for is we can-- you want to name your project, something recognizable obviously. And then the 3D textured mesh, you have to have that. That's what the initial or the base product out of this is going to be is that RCM. Can't utilize that RCM in any of our design software yet, it's mainly used for 3D printing. But we want to select that RCS which is our 3D point cloud as well as our orthophoto and that's going to be in the form of a GeoTIFF.
Again, we double check our coordinate system and then we'll say start and it'll upload. Yeah, I'm not going to start it. I just wanted to show you where you can see the progress of that. So down here, you've got your cloud drive. If I zoom in and look at this, what you'll get is photos being uploaded. You'll get a little timer wheel, photos being uploaded a percentage. Don't close the software until they're completely uploaded. After it's uploaded, then it'll say processing. Then you can exit out of the software and everything's in the cloud.
And after it's done processing-- and this set of photos, 760 photos took about 4 and 1/2 hours in the cloud. Wasn't real long in retrospect to doing it locally, it processes fairly quickly. So you'll get an email saying, hey, your stuff's done and that email will have a link out to ReCap 360. The ReCap 360 is the online version of ReCap, you can do some things there. But if you select the project that you have, it'll take you to the project landing page. And from there, the main thing we're after is view downloads so we can retrieve our OBJ or our RCS or our TIFF is what we're after. So in this instance, the RCS and the TIFF is what I'm wanting. I've already downloaded them so we're not going to do that step. You can also view your model here, which is kind of neat. Just check things out before you continue on, even before you download them.
So this is that set of photos unedited. And this is the RCM that we're viewing online, but you can see like I'd said, there's a pond in here and the reflectivity off of the water really screwed up the stitch. So that's why we're going to bring it into ReCap Pro to edit that out before we push it into InfraWorks. But it did a really good job. Just the DJI Phantom flown. So we drop this. That's the RCM triangulation anyway. So that's how you get your model from the cloud back in a usable format.
AUDIENCE: Do you get a quality report with it?
JOSH CLAWSON: A quality report?
AUDIENCE: Right. like [INAUDIBLE]
JOSH CLAWSON: Uh-uh. I don't believe so anyway, I guess I'm not familiar with it. So from there, we'll jump into ReCap Pro and this is what it's going to look like as soon as you open the project. I actually [INAUDIBLE] to show that. So here's the landing page for ReCap Pro. Wow, they moved the ReCap Photo thing a little bit. So this is where that sticker was, if you will, about how to get ReCap Photo. So I want to look at this scan again. So this is the RCS that we got from ReCap Photo, and if we look at this and just kind of see the quality of it, if you will. I mean, we've got construction equipment over here that it picked up on. We've got orange barrels, we've got a pond. Here, a dry pond, and then we've got a pond over here that has water in it. We've got some structures in here with some varying slope inside of it.
So we want to clean this up and edit it. And I'm just going to do this real quick, try and get this pond area cleaned up, though it's as simple as just selecting it, then you use a fence and go around the perimeter of the water, and then deleting it out. So now it's no longer in the definition. Another thing with photogrammetry in general is your perimeter. You're going to get some goofiness along your perimeter because there's no adjacent photo to stitch to on the perimeter. So I always recommend going at least one or two photo widths wider than what you want to collect. We can go through and trim up the exterior of this as well. So just something like that. Clip outside.
From there, after we've done all the edits that we want to our point cloud, we'll go to the export option here. And we have RCP, RCS, PTS [INAUDIBLE]. RCS Is what we want. I had this issue before, my dropdown menus start disappearing on me. Fun. Anyway, so we'll export it out as an RCS, save it somewhere as an edited RCS, and now we're done inside of ReCap Pro. From there, we'll jump into InfraWorks. Now, generally you're going to use model builder to get your model started. In this instance, I'm not. I'm going to make an empty model, has no data in it other than a coordinate system. So I'm going to click new instead of model builder. Hmm, but it looks the same up there as it does down here.
AUDIENCE: [INAUDIBLE]
JOSH CLAWSON: The what?
AUDIENCE: The graphics card is [INAUDIBLE].
JOSH CLAWSON: Yeah. Let me get out of some of this other software. OK, there we go. So, name [INAUDIBLE] maybe. Awesome. There it is. So I'm going to create this model that has no data in it. I want to name it, and then I want to go into the advanced settings and set the coordinate system for the model. We can do it after the fact or we can do it here prior to launching it if my graphics card can keep up here. I apologize about that. All right. I'm going to cancel out of that and just show what this will look like regardless, show the process anyway. There it is, graphics card. So we would have had an empty model. This is after the fact, after the finished product. Let me jump in and turn some of this stuff off. Maybe.
All right, so once inside of InfraWorks, we would have had an empty model. This is not the best example because this is the generated terrain data from that point cloud. We would have an empty model and what we want to do is bring in that RCS file and there's two different ways we can do that. We can drag and drop it right into the screen and it's going to jump in and you'll be prompted with the configure dialog box, or you can go into your data source window and navigate out. And there's my dropdown menus disappearing on me again. To point cloud and you'll browse out for an RCS file to bring in.
And then from there, to get this terrain. After you have your point cloud inside of the drawing-- Come on. One second. Yeah. Doing things live in graphics performance is always fun. So the point cloud will come in. I think the point cloud's on here. Yeah. You see some areas that have point cloud that do not have terrain, that's after that classification process. I'm just trying to see if I can replicate this without actually creating that new model. OK. So we bring in the point cloud and then we want to do point cloud terrain, which is this button here. Point cloud terrain, or you're generating terrain data from a point cloud. It's just a button inside of InfraWorks under the manage and analyze your model. Select a point cloud that you want and then we've got some options here to go through. The optimum actually does a really good job. You can do more or less. You've got three options. Optimum, more detail, less detail.
One thing about this, though, is if you are dealing with a point cloud and your point cloud's just massive and it's clunky inside of other softwares, InfraWorks does handle point clouds very well. It's very fluid and smooth once you get point clouds inside of there. But we can process this point cloud and it'll actually reduce that point cloud down keeping the accuracy, and we have the option of exporting out a file. So export processed file. We can do that, not going to in this instance. And then the type of data. Lightweight or more detail, less detail again. So we'd run this process and that's where we get the terrain data from. I'm not going to run it just out of fear of things breaking because it's going so well so far. And we get the terrain data generated out of it.
So another thing you can do is create a point cloud theme. That's what this panel is. We have point cloud themes right here, and this theme is-- graphics card's being just awesome. So the theme that I made, you have an option called classification. Any of you notice that list of like 31 different colors on the side of the screen? Those are the classifications for your point clouds. So if I turn this back on, we'll see that there's different colors in here now. So I've got some black, I've got yellow, I've got blue. The brown is what it's recognizing is ground. And does a pan around, notice like the machinery here. So it's not going to process those areas into the terrain data. Like in this instance, it clips it out. It recognizes that is-- I guess I'm not sure what the light blue is, reserved TIN. But it's fairly automatic when it does this. If you do have a site with a lot of trees around the edge and you did not trim them out inside of ReCap Pro prior to getting it into InfraWorks, this does a pretty good job of recognizing that's probably not the ground and it will clip it out of your ground surface.
So alternatively, we could have used model builder and created our model from OpenStreetMap, data sources, and USGS topo data. Then we can bring our point cloud in and then process that point cloud, and it's going to replace the footprint of that point cloud. The footprint that that point cloud takes of the existing ground data that it has, it's going to replace that with this newly generated terrain data. You get some really goofy looking stuff around the perimeter where it's tying in, so we don't have like daylight settings inside of InfraWorks for that particular instance. But you can do that and you could also bring in your othomosaic just like any other aerial imagery inside of InfraWorks to drape over the Bing map aerial imagery that you get. This instance, we're just after that surface so I'm not even messing with the aerial imagery inside of InfraWorks, but after we generate the terrain data and as long as we have our coordinate system set and our coordinate system matches that of our survey data and also that of our Civil 3D drawing once we get into there, that's the important thing.
So we're done inside of InfraWorks so I'm going to actually exit out of InfraWorks. Now we're back inside of Civil 3D. I'm going to jump out to-- Actually, I'm just going to start a new drawing. Yeah, I'm real scared with my graphics card here. So this is just the NCS Imperial National CAD Standard Civil 3D template, nothing fancy inside of here. First step. What I want to do is set my coordinate system first and foremost. And this one is in Dane County, Wisconsin. If I can grab it. Wisconsin. OK. Just making sure that the coordinate systems match. You can save your drawing out to where you want, name it something. I'm not going to in this instance. The next step I'm real curious about if this is going to work or not. On the insert tab of the ribbon, we have an InfraWorks button and also we have an InfraWorks tab on the ribbon now. I don't know that this dropdown is going to work for me. It's not. That's my graphics card acting up because I'm projecting here. So I'm going to try my Autodesk InfraWorks tab on the ribbon if this will work. All right. I'm going to revert to a video that I got of this. Just, this isn't working real well.
OK, so this is a Civil 3D workflow. I made this video as a backup just in case. But the video here is from a InfraWorks model that had model builder in it, so I'm going to throw a border on it. So essentially what I'm doing here is saving the drawing out. So drone surface. And we'll see if I can just talk through this as we go. It's at two times speed so I might have to stop it. So setting the coordinate system, we just went through that. Dane County, county coordinates. Wisconsin has like three different county coordinate systems. It's not fun. From there, we'll go up to the insert tab on the ribbon, I think. Maybe. There we go, InfraWorks button there. On the dropdown for the InfraWorks button, you have open InfraWorks model and that's what we want to do. We want to open up that InfraWorks model inside of Civil 3D.
Now the one that we created with nothing else in it but that point cloud, we would not have to go through and refine our selection set, what we're bringing in. This will run through that. Just to note where your InfraWorks models are stored out of the box. It's in your documents, so C, users, my name in this instance, documents, InfraWorks 360 models, Autodesk 360, and then some crazy number like that. Inside that file, there's this SQL file that's named what your model's named, but they've always got some goofy number on them. The easiest way to navigate for them is just sort by date modified. Put the most recent one at the top. Open that one, that's the one you're after.
AUDIENCE: Can you change the location?
JOSH CLAWSON: Can you change it? Yeah, you can absolutely change where you--
AUDIENCE: [INAUDIBLE]
JOSH CLAWSON: In the settings inside of InfraWorks, you can tell it where you want it to go, but that's the default out-of-the-box one. So from there, double check our coordinate systems, shows that they match, good deal. You will get a warning if they don't match here and you can't move on any further so you do need to make sure you have your coordinate systems set properly. From there, again in this instance, I'm going to actually refine my selection set. So inside of an InfraWorks model, you can bring in many different items that push over into Civil 3D. Alignments, profile, surfaces, a bunch of different stuff. In this instance though, like here's a list of all the different types of things that you can bring over. The surface that we are after is going to be that AIW existing ground. That's all we want. That's the one that's going to hold the data for our terrain surface that we generated from our point cloud inside of InfraWorks. And then from there, we'll say open model. And again, I had a bigger footprint. Same exact data set in this instance, but you can see that I've got contours there because this is an out-of-the-box template. And let's let that tip a little bit. But notice how tight that triangulation is. It's because it's a mesh surface, but it's still a TIN surface coming out of InfraWorks.
But you can see the level of detail that you get out of that. Now this is way tighter of a mesh than we're ever going to see in a traditional survey. So what we want to do is-- If we can leave it like this, it's still functional. And then if we data reference it out, it's going to function better than a very heavyweight surface would downstream. So what I'm doing here is I'm just putting a boundary on it. And this one didn't have the ballooned area cleaned up on the water. So I'm just drawing with the polyline, putting a boundary on this. I'm going to put a clipping boundary on my surface so we don't have everything else outside of it. Add an outer boundary. And then from there, we have the option to reduce or simplify our surface. I'll just zoom in and take a look at what we got. And there is a couple little spots there that I should have cleaned up, but I guess I'll adjust that now. Rebuild the surface. So to simplify that surface, you select the surface. And then on your ribbon, you have your edit surface. Down at the bottom is simplify surface. There's two different methods, edge contraction or point removal. Being that where this surface is built off of point cloud, we're going to use point removal. Let me slow this down just so I can keep up with it.
So we've got edge contraction or we've got point removal. We want to use point removal. And then we're going to use the existing boundary for our surface if we want. You really just skip this page. But now that we're going to actually do the reduction, reduction options. This total points inside that region is 140,049. Traditional survey, you're going to have 700, 800 in that area. Same footprint. So we've got way too many data points to replicate a traditional boots-on-the-ground survey. That's one thing to keep in mind. We're not doing a boots-on-the-ground survey. So what we have right now, this is going to be the most accurate. As we start to reduce this down to look like something more familiar, we're pulling data out of it is what we're doing. So if we're looking for something to be as accurate as we can get and relative to the elevation of it, don't reduce it down. Just leave it as-is. It's not going to look familiar though, so it's just kind of getting over that.
So we want to reduce this down. In this instance, I'm actually going to reduce it down by 95%. And I'm going to do that twice. So 95%, then I'm going to run in and reduce it down again by 95%. So it's drastically reducing the amount of data inside of here. We can watch the contours kind of shift a little bit. We'll take a look at it in object viewer. Now we're getting something. This is still a lot. We've reduced 95% of those points out and we still have something that's not real recognizable. That looks like a texture mesh, it doesn't look like a TIN surface. So we'll run that reduction again or simplification again. Running through, 95% finished, and it'll jump. Now we're getting something that looks a little more familiar. So if we go into object viewer, that's a little more typical of what we're going to see from a traditional boots-on-the-ground survey.
So from there, it is a surface. I mean, in every way, shape, or form, it's a surface. Oh yeah. Our othomosaic, map by insert. We'll either use raster tools or the MAPIInsert command to drop in your orthomosaic, and then navigate out to it. The TIFF file, open it. It's georeference, it's going to automatically drop in where it needs to be. It works pretty well. Send it to the back, put a screening on it if you want, or a fade. I'm putting a clipping boundary on here just to match the footprint of the surface, and then draw order essentially to push it to the back, or display order. And I think I just went and changed the style to something a little brighter so you could see the contours on top. And we'll drop the fade down. And now we have a very up-to-date aerial imagery of our site that we just flew, as well as a functioning TIN surface inside of Civil 3D that's georeferenced and geolocated.
And then here I just go through and cut a couple of alignments and show in profile view, which we'll go ahead and let it go. So just make a note, your basic alignment down this path. That's just kind of showing some of the benefit of having that up-to-date aerial imagery. I'm running right down the center line of a bike path and then down the center line of a sidewalk. I don't know where those are without that photo. And then I'm going to do a cross section of the pond just to see how the slope goes with that. And notice right here in that photo, it's cut out over the water. That's just how it came out because it couldn't stitch that area together. And that's why there's a strip out there and that's why the water looked like it was blowing up on the edges.
And then we get the alignments done and then go through and do a couple profile views just to show what this looks like now in profile view. I'm sure everyone knows how to make a profile, right? Go ahead.
AUDIENCE: [INAUDIBLE]
JOSH CLAWSON: The question was, how do you handle the trees?
AUDIENCE: [INAUDIBLE]
JOSH CLAWSON: That's one of the things you really have to watch out for. I mean, photogrammetry we're doing by photos or by light. So if you have trees, you're not going to get anything underneath them. It's the first thing that it hits. If we were able to get a LiDAR on a drone, which we're very close to being able to do that, which would penetrate that vegetation, then we're down to mineral soil as opposed to just an image. So yeah, vegetation's a big thing to watch out for with this process, just something to keep in mind as you move forward. Yeah. It's just showing the contour, the profile of that surface. OK. So that's really it. Let me jump back into my PowerPoint here.
So in review, we'll just kind of go through what we did. We flew the site, we utilized ground control, we had flight control software 80% overlap. And then we processed the photos inside of ReCap Photo. We assigned our ground control points and we selected the output that we wanted. We retrieved it in ReCap 360 online. We have the option to edit the point cloud inside of ReCap Pro and then also export that out as a new RCS file. From there, inside of InfraWorks, and I apologize for my graphics issues there. We processed that point cloud so we could categorize it as well as generate ground data from it. And then we went into Civil 3D and accessed that InfraWorks model to get our TIN surface. And then we created our surface file for downstream use. I want to open it up for questions for a few minutes and then I got like an interactive quiz to do at the end and I got a couple prizes to give away. Yes?
AUDIENCE: Do we have to [INAUDIBLE]
JOSH CLAWSON: OK. The question is, can we skip InfraWorks in this process? To get a TIN surface, no. You can put a point cloud inside of InfraWorks absolutely, and then you can create a surface out of that point cloud inside of Civil 3D. But you're stuck with that point cloud definition inside your Civil 3D drawing which gets really clunky.
AUDIENCE: Oh, so the [INAUDIBLE]
JOSH CLAWSON: Yep, so now we don't have a point cloud in our Civil 3D drawing. We just have that TIN surface definition. Yep, go ahead.
AUDIENCE: What kind of flight control software were you using to collect all your images?
JOSH CLAWSON: So what kind of flight control software did we use in this particular flight? DroneDeploy on an iPad, that's what we used. It wirelessly connects to the drone and you just push the mission to it and hit go.
AUDIENCE: Well see, [INAUDIBLE]
JOSH CLAWSON: You said, what's the resolution?
AUDIENCE: [INAUDIBLE]
JOSH CLAWSON: That I'm not sure. It depends on how high it is. That was a 12 megapixel camera and the bigger the footprint, the bigger area a pixel's going to take up. So the lower you fly, the more accurate, but also the more photos so you've got to find that balance of what you're looking for.
AUDIENCE: So all the imagery is Bing in InfraWorks. Any way to do Google Earth Pro imagery?
JOSH CLAWSON: The imagery that's always been in InfraWorks, is there any way to bring Google Earth imagery into InfraWorks?
AUDIENCE: Google Earth Pro.
JOSH CLAWSON: Google Earth Pro imagery? It depends on what file type format. I don't know. Paul, do you know? You can?
PAUL: You can extract but I can't remember [INAUDIBLE]. It's a KML I think.
JOSH CLAWSON: It's a KML or a KMZ.
AUDIENCE: KMZ.
JOSH CLAWSON: OK.
AUDIENCE: Yeah, but will it bring into InfraWorks?
PAUL: Yeah, you can bring that into InfraWorks.
AUDIENCE: Yeah, so how much-- you said there was a cost per image, [INAUDIBLE] Do you have any idea how much that is?
JOSH CLAWSON: Each cloud credit is $1.
AUDIENCE: $1 and--
JOSH CLAWSON: That one for 760 photos was 55 credits, so $55. And that's just Autodesk.
AUDIENCE: So to that same point, if you're using something like Pix4D to do your generation of your point cloud, you can bring that into ReCap and the rest of the workflow would be the same?
JOSH CLAWSON: Yep, yep. Any point cloud you can get into ReCap, you can do. So if you do have a Pix4D or some other photogrammetry stitching engine that you're getting an export from in a cloud format, you can push that into ReCap Photo, convert it into an RCS and then continue on and still get that TIN surface.
AUDIENCE: So you did some of the basic editing for like the point cloud in ReCap. Is there any way or benefit to do like more fine-tuning in InfraWorks at all? Like if you wanted to get rid of some-- let's say some vegetation area that [INAUDIBLE]
JOSH CLAWSON: So can you edit the point cloud inside of InfraWorks? No, not really. You can't do any editing to it. Once it's in InfraWorks, it's just a point cloud. There's not a set of point cloud editing tools. There's a set of processing tools, but not editing. We can utilize the data once it's inside of InfraWorks in a few different ways. Like one with laser scans so you get a much higher density point cloud inside of InfraWorks or you can extract linear features from it and get 3D polylines for like your curb and gutter line that you can push into Civil 3D. But to edit it, no. You've got to do that inside of ReCap Pro or something else that will edit an RCS file, but I think that's the only option is ReCap Pro.
AUDIENCE: So just do it there?
JOSH CLAWSON: Yeah, yeah.
AUDIENCE: [INAUDIBLE] accuracies of that? So [INAUDIBLE]
--what the accuracy is?
JOSH CLAWSON: So the sense of what you're getting at is vertical accuracy of your end result.
AUDIENCE: [INAUDIBLE]
JOSH CLAWSON: And horizontal's spot on. Horizontal's extremely accurate, like down to the thousandth. Go ahead, sorry.
AUDIENCE: [INAUDIBLE]
JOSH CLAWSON: Right, and that was just for demonstration purposes. You can zoom way in on that to where those little triangles are about that big on the ground. That'll fill up your entire computer screen. I didn't go through the precision there. But yes, you can zoom farther in than you need to.
AUDIENCE: And then what was the [INAUDIBLE]
JOSH CLAWSON: 2/10 is the worst I've seen. 2/10 off is the worst I've seen. Usually it's under a tenth. I mean, we're talking like 5/100 or 6/100 off vertically. It depends on-- Your overlap is a big thing, the quality of the camera. So a higher resolution camera would have been better, it gets you more points because your pixels essentially turn into your point cloud. Other things, to fly lower is going to give you more dense per square footage, more pixel density per square footage. And it's, again-- Yeah.
AUDIENCE: [INAUDIBLE] Just to recap that, that's the same question. You're trying to say that you're getting 2/10 of a foot error on these--
JOSH CLAWSON: In certain areas, yeah. Closer to the control points, it's way better. But it depends on-- A big part of it is what the ground's like. If it's pavement, we get really close. If it's like a gravel field like that was, you get a little bit of variance. It depends on what you're taking the pictures of. And then reflectivity is another big thing. If you've got areas that reflect, then it's not going to be as accurate. There was-- We'll go with you.
AUDIENCE: [INAUDIBLE]
JOSH CLAWSON: Say that again.
AUDIENCE: [INAUDIBLE]
JOSH CLAWSON: I think cloudy is better because you get less reflectivity. The question was, is cloudy or sunny better conditions to fly in? It depends on what you're flying. If you've got a lot of high-shine surfaces, a cloudy day is going to give you a better result. But in the most part, I don't think it really matters as much as if you get broken clouds and you get varied cloud cover going over the site, then you have different illumination levels in your photos that they're trying to stitch together. That can sometimes result in not a proper stitch. You had another one?
AUDIENCE: Yeah, just-- With the classification in InfraWorks, is there anything to that or is that just an automated process and is it OK? Or how--
JOSH CLAWSON: The classification process inside of InfraWorks. How does it work or how do you go about doing it? Essentially, when you process your point cloud into terrain data, that's when it classifies it. It's doing it itself for its own purposes to create that terrain data. A byproduct of that is you can go and put a theme on those classified points to view them, then you can also go through and that classification. So we've got a group like a street sign, a small chunk of points going vertically. It'll say, hey, that's a sign. You can go through and model your point cloud so all of those things that are not the ground or it doesn't recognize as that, it'll go through them one by one and you can replace that colored chunk with a model from your model library inside of InfraWorks. So you can go through real quickly and correct all those goofy areas, replacing them with a 3D model.
AUDIENCE: So taking this one step further, do you have experience [INAUDIBLE] the duration of the project?
JOSH CLAWSON: What are you asking? You're looking for volume surfaces or--
AUDIENCE: [INAUDIBLE]
JOSH CLAWSON: Yeah. OK. So doing multiple scans at the same site throughout the project lifecycle. As long as you use the same ground control points, it's going to be in the same spot essentially. Common areas for that'd be like mines where you're doing volume surfaces. You fly an area, you fly weekly or every other week and you need to compare volumes. Construction sites, stockpiles, things of that nature. Is that what you were getting at?
AUDIENCE: [INAUDIBLE]
JOSH CLAWSON: That part inside of InfraWorks where the point cloud is processed, it reduces the point cloud once there as well. And you can also reduce it more than that, but again, you're losing accuracy so-- Yeah. We had a question back here?
AUDIENCE: [INAUDIBLE]
JOSH CLAWSON: So I'm assuming Pix4D didn't import your CSV file? The question is, do you have to manually input your x, y, and z for all your ground control points inside ReCap Photo? Right now, yes. They're supposed to be changing it very soon. I have an update that I haven't updated yet just for fear that it wasn't going to function for today, but there is an update I need to update on that and it may be in there.
AUDIENCE: [INAUDIBLE]
JOSH CLAWSON: So is there a limit to how far you can reduce your surface down? No, not really. I mean, you need three points to make a surface, to make a plane, but it has no accuracy to it. You know what I mean? So it's really up to what the purpose of that is. What's your intended purpose for the surface and how much accuracy do I need? Another thing would be putting a boundary on it for just a specific area would be another way to minimize that.
AUDIENCE: Did you find that reducing it by 95 twice works the most?
JOSH CLAWSON: That's pretty common to get down to something that looks more like traditional boots-on-the-ground survey. That being said though, not reducing it at all is going to be more accurate. It just depends on what you're using it for. So any other questions? OK.
Downloads
Tags
Product | |
Industries | |
Topics |