An early CAD file amphibian or a GIS mammal? Dale Lutz reveals his true identity as he takes time out of his FME World Tour to explain how the two compare, who the new kid on the mapping block is, and where his allegiance lies.
Dale is the co-founder of Safe Software. Since 1988, his bread and butter has been spatial data. Along with his co-founder Don Murray, they built a translation software for a then-emerging format called SAIF, in many ways the grandfather to the modern GML. It was a popular solution at the time of the evolution of data formats, with people primarily using it to move data in and out in CAD files.
Historically, CAD was used to draw buildings and facilities. The Intergraph Design System (IGDS) that was run on PDP-11 was its predecessor. To build a translator for older design files, people had to know PDP-11, integers, and floating-point, amongst others. The process was tedious and slow. On top of it, someone then had to optimize the storage of the drawing so that people could open it and see it on a screen quickly. These ancient systems were never meant to store complex attribution we take for granted in the GIS world today. Later down the line, when people started to use technology to do more complex mapping, they used CAD as their starting point.
MicroStation was a PC clone of IGDS using the same file format as PDP-11, only it could run on a PC independently. Early mapping was done mostly in MicroStation while utility companies, on the other hand, were using AutoCAD to design facilities. They realized they could start to map with the same tools that they already knew and used, but that required attaching meaning to things.
CAD was designed to produce drawings that humans could read. For instance, floor plans for a building project. An engineer would understand what is implied on the plan, and they’ll know what to do with it. That’s not the same thing as a computer having to translate what’s going on. These early CAD drawings could not embed attributes or properties. Even if they did, it was clunky.
MicroStation files had provision for a Z value, such as a Z low and Z high, that users realized were not utilized for their 2D drawing files. They started to shoehorn information like database key and parcel identifier in there manually, things that were never meant to belong there. This ended up messy for users, but it was great news for those who wanted to make a living designing a flexible tool to clean things up.
Over in AutoCAD land, users were also being creative inputting crazy things inside of blocks. Those blocks used symbols to communicate information, and users could add labels and block attributes to their projects. Did you need to draw a forest? No problem. You just added a label that said “spruce tree.” A human could understand that this was a forest, but a computer had to connect the boundary lines into a polygon, then do a point in polygon test to fish out the labels.
We can clearly see the mismatches between the GIS approach where you have rich information attached to each piece of geometry, and the CAD approach where the information is implied, or worse, encoded.
In CAD, symbols can end up being aggregations of complex collections of multiple types of geometry. Also, polygons are hardly a native concept, and the meaning is only implied by where the lines are. Take a leader line that needs a label, but the label can’t quite fit where it’s supposed to go because it would be on top of things. You place the label on the side with a line that points to where it belongs. When you then try translating this in GIS, you need to go and slide the label along the leader line to figure out where it should be, and you derive the implied relationship. CAD geometry is more straightforward than GIS because things don’t have a relationship with anything else around them other than where they physically are. At the same time, CAD is more complex because it has splines, arcs, cones, and 3D objects. It’s truly a mixed bag.
GIS and CAD are fundamentally different.
GIS is more structured. A bit like Canadians. You know, law, order, and good government. It’s data management with precise specifications and schemas defined. It would be hard to go rogue and abuse it.
CAD, on the other hand, can be a backdrop for a Wild West scenario. CAD technicians can apply their artistic license and suggest how it should be used and design anything they want. Nothing in the system will stop them. CAD is like Americans. You can’t tell them what they can’t do. Their founding three words are life, liberty, and the pursuit of happiness after all.
Based on these fundamental differences, data migration between the two can be a nightmare. Even if users followed CAD specifications, they probably applied them loosely. A street can be labeled as a boulevard as well as a road. They are the same thing. Once you scan a thousand CAD files, you’ll see how people just named things on the whim. During the transformation from CAD to GIS, you read the CAD file and check against a spreadsheet of what’s standard and should be allowed. Colors, line styles, levels, and layer names. Anything that doesn’t match the specs is handed over to a technician for cleaning up. In the case of our misnamed streets, once the different names you encountered are collected and put in an Excel sheet, someone can manually go through the list and group them. Doing this can reduce 800 variations of the layer names down to 30, so that the translation can take place.
Rules are often interpreted and enforced differently depending on how rigorous the mindset of the users or the organization is. More recently, systems like Maximo and IBM require CAD standards as well as CAD files to conform. This opened up a new market for CAD add-on tools that enforce specifications, resulting in tighter data models. NewCAD now has certain GIS elements, a little bit like Bentley’s XFM that stores rich attribution with drawing elements such as AutoCAD map or Civil 3D, providing structure. If an organization prefers to go down the CAD route, it can do so with the help of these newer developments focusing on stricter specs.
A missing coordinate system is a menace, really. Even back in the days of MapInfo, people realized that not having a coordinate system at all was a problem. They relied on standards and metadata. Yes, there is always the option of tricking the local coordinate systems as everyone did early on, inspired by ESRI. A CAD file could be shifted by plopping down a coordinate system and a world file, then the feature manipulation engine (FME) would pick it up.
NewCAD is said to have a coordinate system that’s going to be written and read properly. Civil 3D and AutoCAD maps have also progressed in this regard.
You can already see a meeting in the middle of modern tools as a result of the progression in the last 15 years. Just look at ArcGIS. Sophisticated geometries are a possibility. Rich data collection that CAD technicians love and sophisticated production of drawing is already available in ArcGIS’s editing tool.
Take a look at the most recent network and utilities work that EPA EMI is doing for modeling transmission lines. It’s a powerful data model that’s deep, connected and has elements of old CAD files with a familiar look.
IMDF 4 is an indoor mapping data format that’s used by the Apple ecosystem. At the moment, it’s a maze due to the varying degree of rigor in how facilities such as shopping malls or airports collect their data. Some did it robustly with high spec and quality. Some did it in a Wild West manner. Indoor mapping and navigation are a wave that washes the causes of change. Now there is more reason to give maps a meaning other than just for the purposes of facility maintenance, and as a result, people are beginning to pay more attention.
GIS and CAD have managed to run an exclusive show as a duet, so far. With the arrival of BIM or Building Information Models, it’s going to be a stellar trio. BIM is the unholy marriage of GIS and CAD for the purpose of building information modeling. BIM takes a drawing from CAD and attaches meaning to it to make explicit relationships. Technicians can now walk around and change lightbulbs in the right places and people can get to the emergency exit quicker.
Unlikely. They coexist peacefully in the geospatial realm with their new friend, BIM.
BIM will produce CAD files for downstream use. Some users will prefer collecting information in CAD because they are unlikely to migrate to BIM. The letter A in CAD is not for analysis, it’s for aided. The current system lacks analytics.
GIS is the analytical toolbag in this mix. It can take the data collected in CAD or designed in BIM and lets you reason, ask hard questions and use it for analysis. What is the time that it takes to exit from each spot in this building? Can we draw diagrams or make labels on different things? Can we make heat maps of where people are going? Can we compute the most efficient routes to and from various spots?
CAD is more human. It has been making beautiful drawings for the last 40 years. Drawings that people can look at, understand and, based on them, make quick decisions.
CAD has nailed speedy and efficient data collection, and their user interfaces are efficient. GIS can learn a lot from it. Minimal mouse clicking and allowing users to sit down and benefit from whatever they’re doing within the first minute without slowing them down.
With so many of the latest software solutions out there that don’t require an in-depth knowledge of Geographic Information Systems and its specialist tools is it a wonder that GIS technicians and analysts are wondering what the future has in store for them? The same questions must be running through the minds of CAD technicians too.
The role of CAD technicians and experts is likely to evolve by progressing to BIM. They will be using extensions to the NewCAD system that will allow them to collect and attribute information at the same time as well as mobile data collection. A CAD technician isn’t just sitting there identifying red lines. Using plug-ins like XFM or Civil 3D with a structured data model lets them put more intelligence into the data they’re collecting. They know that it’s a red line because it’s a certain kind of road. It’s planned, and it has a surface. It’s explicit.
AutoDesk and Bentley are also moving towards online data and away from tiled files that are local which would also go hand in hand with BIM. We don’t need to store data locally anymore, as we did on AutoCAD and MicroStation, for speed of display. We’ll be using JSON, a flexible way to have as many different properties, attributes, and geometries as we want. The division between GIS, CAD, and BIM will get blurrier whether we use Bentley’s iModel 2.0, AutoDesk’s Project Quantum, or whatever it may end up being named when it becomes commercial. It’s going to take a while but watch this space. These two are where the puck is going, as Wayne Gretzky says.
Sign-up for the weekly news digest for those who are interested in more relevant information, actionable advice my podcast guests shared, and snippets interesting geospatial stuff I discovered during the week. You can sign up HERE.
To put it simply, point clouds are a collection of XYZ points that represent some real world object of nearly any scale.They can be generated in a few ways. As geospatial scientists, we mostly work with LAS/LAZ data collected by aerial LiDAR (light detection and ranging) scanners at varying scales, from landscapes, down to project sites. We may also derive point clouds from highly detailed orthoimagery of an area, such as from the products of a drone flight.
As a data scientist, you don’t just go in and solve problems. You make recommendations to multi-faceted issues so that you get a fantastic model in the end. You’ll also be advocating a better use and understanding of the data while you do that.