What's the point?

(0)

After the AU keynote presentation this year, one of the things I resolved to look into in a lot more depth was laser scanning and point clouds.  After all, we’ve decided to move all of our work completely to BIM, so it seems pointless to entertain the idea of 2D surveys any more, and besides, they look awesome…

So awesome in fact, that after I got back I talked about very little else whenever BIM came up in conversation.  In a weird twist of fate, it turns out that a friend of mine is in fact a real life professor in the subject of geomatics, which involves a lot of work with laser scanners and their outputs.  This conversation culminated in a visit to the University of Newcastle for a demo of the hardware and software, and put us in touch with Leica, who also took the time to visit our offices and give us a demo of their HDS6100 scanner and importantly for us, the results.

I’m certainly no expert in the field, but I thought it would be interesting to share what I’ve picked up so far from an architects point of view.  We’ve been considering what to do with the data, how to manipulate it, how to specify a survey and importantly what we can do with it once we’ve got it.  We don’t have a tie-in with Leica, but they were good enough to spend some time demonstrating and explaining the process, so much of my knowledge comes from that meeting and hence so does much of my point of reference.  If you read anything you know is wrong, or have a different idea, please feel free to use the comments – I’m learning all the time…

What is it?

The technology behind laser scanning is fairly complex, and I’m not going to embarrass myself by trying to explain it and then get it wrong.  Needless to say, the survey needs to be completed by someone trained to carry it out (this apparently is fairly simple to undertake – Leica offer a 4 day course for beginners) and involves some seriously expensive kit. (However you sell it, £70-80k is a big investment, but you can rent for a lot less).  Most survey companies now offer scans – just ask around. 

Close range work (internal and detailed survey) is best done using a phase scanner.  My understanding is that these scanners fire a continuous laser beam to and measure the intensity of return.  They are accurate to approximately 60m.

For longer range scans (ie. External facades and civil works), a time of flight (or time-pulse) scanner is your best bet.  These fire individual pulses of laser and perform a similar measurement to produce the point cloud.  These however are accurate to a distance of approximately 300m.

 The main difference between these two types of scan is the time to produce.  Whilst at the University, we ran a scan of the room using a time of flight scanner.  This produced a point cloud of the room, but took more or less the whole time we were there to capture – approximately an hour.  When Leica visited our office, they used a phase scanner, taking two scans of the space in less than ten minutes.  It certainly seems to me that for architectural internal surveys (and really most externals if its not massive), the phase scanner has the most potential.  Interestingly Leica have now brought out a kind of hybrid that sits somewhere between the two – I’m not going to try and explain that one, but look here if you’re interested.

 Both types of scanner can also capture photographic data to combine with the point cloud data.  This enables each point to be given colour data and therefore materialistic appearance – great for presentation of the data.  If this isn’t available, the intensity of the return can be used to give some idea of material finish – what gives a lot of point cloud images a ‘thermal’ appearance.

 What do we get?

The output of a laser scan is a ‘point-cloud’.  This is a file made up of points in 3D space representing the solid surface returned to the scanner by the laser.  These have x, y and z co-ordinates and depending on the scanner also carry intensity values (which can reveal materiality differences) and image data (if the scanner is fitted with a camera to capture picture data).  A typical scan will contain millions of these, so the level of detail possible is exceptional.  Interestingly, Autodesk have just confirmed that their 2011 release of AutoCAD will manage point clouds containing up to 2 Billion points.

 The raw data files returned from the scanner are not always ready for manipulation immediately.  Typically these require some coordination, or registration, to establish connections between more than one scan, or to a local or national coordinate system (newer scanners can do this on-site in a total station set up, I believe).  If the scan is one of a series (normally you would expect this to get more accuracy and less ‘shadows’ or simply to cover more area), it may be ‘point to point’ or ‘cloud to cloud’.  ‘Point to point’ clouds require pre-determined reference points used to coordinate the data.  ‘Cloud to cloud’ clouds rely on on-the-fly coordination of the data, based off a set starting point.  The difference between the two is not in the outcome (you still get the same point cloud), but in the set up.  With a point-to-point registration, more set up is required in the field but the results are then ready straight away, but with cloud to cloud, this needs to be done afterwards using software.  Worth thinking about if you are commissioning a survey yourself – time spent up front may save it afterwards…

 What do I do with it?

 This is really where we were coming from in this thread of research.  Whilst the concepts above are fairly complex, typically as architects we pay someone else to worry about that and we get the data.  With point clouds though, the biggest question for us was “What do we do with this?”.  For instance does it go into Revit (short answer – no)?

There is a massive amount of software out and about for point-cloud manipulation, and in an emerging market I’m not going to try and guess where Autodesk are going to take it.  The fact that its already in AutoCAD 2011 says a lot though – how long before Revit can use them?

At the moment though, Navisworks Manage seems to be the best format, and certainly one we’re investigating heavily right now.  The scan that Leica completed for us is of our central atrium, and we have that modelled in Revit, so we’re currently investigating the clash detection capabilities and onward potential for rendering and walkarounds.

Leica have a very sophisticated bit of software in the form of Cyclone, which they used to manipulate and register the model I’ve shown in this blog for us.  It has settings that enable you to export to a variety of formats, and for manipulating the cloud in terms of level of detail, and for stripping out information such as people and furniture.  Interestingly, you can apply layers that Navisworks will pick up, so it’s already possible to manipulate the data to exclude furniture and the like from first glance.  This is Leica specific software though, I’m sure there’s other out there, but I’m not aware of a point-cloud standard file format, so surely some post-survey manipulation will be necessary (comments here welcomed)?

We also discussed geometry creation, and it is evident that this is of most interest to architects and the construction industry (think of ‘as-built models’ and even a BIM model from scan – sorry, did I say think, I meant dream!).  3Dreshaper and Polymagic offer some opportunity for this, but we haven’t got into that as yet.  Its worth noting right here though, that the geometry generated is just that, not a smart BIM model.  We briefly discussed the necessary workflow to get solid geometry models, and it involves quite a lot of steps (outlined below), personally I think we’ll be taking a watching brief on how this area develops, as its evidently the next step.

• Scan
• registration
• unification
• export to 3^rd party app
• algorithm
• mesh
• solid

Other applications such as the Alice Labs plug in for 3DS Max and point tools offer point cloud rendering capability, which is certainly of interest for visualisations.  Nodal Ninja kit also develops upon the photographic rendering capabilities, enabling hi-resolution photographs to be further integrated with the point cloud data regardless of the type of scanner used through a mechanical device mounted onto the scanner tripod.  Leica have developed an excellent output called Truview, which is a fully intuitive 3D web based output from the point cloud – enabling measurement and visualisation of the scan, via the web, for all members of the project team.  You’ll need to download the plugin from their website, but the results of our scan in truview speak for themselves, and apparently English Heritage now list it as a deliverable for their surveys…

You can view thie trueview file, in your browser by clicking HERE.  You will then need to select one of the stations that we set up (Yellow triangles on the plan) and you may need to install a small plugin.  If the model looks like a low resolution photograph give it a few minutes to load.  This is not a panoramic photograph, you can take real 3D dimentions, areas and coordanates straight from the model, try taking a dimention between 2 surfaces.

Laser Scanning and BIM?

I’ve been slow in writing this blog (there’s a lot to get your head around!), but from what I’ve seen, we’re convinced that point-clouds are the surveying tool of the future for BIM applications.  Judging by the pace at which Autodesk and Bentley are developing their ‘point cloud engines’, we’ll surely not be far away from seeing the results embedded within our software applications.

In day-to-day terms of commissioning a survey, the fact is that costs presently remain the primary obstacle to their widespread application across the construction industry.  Another obstacle remains the outputs – its difficult to visualise the benefits when you can’t put your hands a set of survey plans directly from the commission (though this is eminently possible with a minimum of additional work, especially when you think of the alternative).

However, we’ve moved into new territory with BIM.  We don’t work in a 2D world any more and neither should our surveys.  The opportunity that 3D laser scanning offers us is the chance to combine real, measurable context within our virtual models, both in presentation and in production information.  That’s one huge benefit that we’ve already seen with BIM models, and it also applies to laser scans – the presentation flythrough and visualisation data is the same as the construction survey – one dataset of information for the duration of the project in true BIM fashion.

I just said I was slow – so slow that someone has already beaten me to an article on this very subject!  Have a read what Martyn Day has to say about the same subject, I’m sure you’ll find it as interesting as I did and I hope you’ll find it supplements what we’re discovering and prompts you to understand my point.

Thanks go to Paul from Leica, who has an awesome youtube channel - check that out if you want to see more, and the laser scanning forum have a constantly updated twitter feed. 

• < Prev
• Next >