I would just like talk about what Pointerra is trying to do and...

I would just like talk about what Pointerra is trying to do and the obstacles that they will have to face.

Pointerra's current claim to originality is what they imply is a unique method of slicing and dicing extreme 3D point datasets . Certainly their method is original, as they have patented a technique. Is it unique in its functionality, most definitely not. I think this is an important issue.

As was mentioned by one of us, the amount of data is huge. Even small surveys with a laser scanner will product 100s of millions of points very quickly. The files produced tend to be XYZ ASCII files, which as flat featureless files that are very big on disk.

Generally theses files are what many programs like Microstation and AutoCAD, and there are many such programmes, can display individual files. They load every point into memory and graphic memory and they spin around magically. However if you try and load a number of files the amount of data gets too large for memory in the computer and the graphics card. The speed gets slower and slower and eventually the program terminates graciously ( or crashes ungraciously). This is extremely frustrating. The idea is to get a bigger computer.

Is this a bad thing? Not really. Programs do a lot with this data. They use the point data to help in spatial modelling and analysis as well as the visualisation. What is happening is the traditional expectation of turning Data into Information into Knowledge and then Wisdom if you get that far (rarely). That is a bit buzzwordy, but it is essentially right.

So when you have data that is not in the 100s of millions, but in the 100s of Trillion points. What happens? I loaded many trillions of data points from an aerial survey of the Senoma area in Northern California. It wa free data from OpenTopography.org. The data is massive and very impressive. However it comes in 1000s of compressed XYZ files. These have to copied locally to your machine uncompressed and then you have 1000s of very large ASCII files. If you try and load a number of these files into a Microstation, maybe 2 files can be loaded at a time, and any hope of seeing the whole area is impossible without severe culling, which is an external process. The whole process took months, with many download failures. Very frustrating.

What companies like PointCloudTechnology, Euclideon and many others do, and Pointerra is trying to do, is to build all these individual files into a compressed and Octrees structured file such that it is easy to firstly have the data in one location (in one database), secondly compressed to limit the size on the server (which is not that important) and finally that it is indexed (Octrees) in such a way that there is a hierarchy of resolutions. Theses companies tend to have their own display programs because they do not do what traditionally is done, that is the data is loaded into graphics or local memory. In essence the display is really a movie view of their program dynamically parading through the data base based on your viewing commands( mouse movement). As you zoom into the area a new "frame" of data (a screen frame) is added to the computer monitor. Hence you don't need a big computer with a big graphics card. You need a multi core computer with as fast a disk drive as possible.

So dynamic graphics speed ( the immersive experience) is determined by the computers processing speed and the speed of the disk drives. But also another issue. The resolution of the screen (over the movie) if you have a 1080p monitor you need a bandwidth from the CPU to the screen memory of around 18Mbps (18 megabits per second). This is essential what a 25frames a second movie will deliver. Any slower bandwidth and you will get jerky frustrated pictures. The immersive experience is lost! Bear in mind the movie structure is probably MPEG-4 AVC or H.264 compression, which is standard today. If you try to display on a 4K screen the data required for 25fps display is huge, A very daunting task. The technology required here, which is similar to Netflix trying to run a 4K movie is in constant innovation. MPEG-4 is too slow for today's internet capacities, but within the speeds of a computers memory buses, quite acheivable.

A couple of points here.

This is working locally on your own computer.
Is it Big Data ( not really but getting close to be considered) but it is not relevant
Is there any analytics (like the capabilities in Microatation). No, only visualisation (you have outsourced analysis to your brain)
Is there any design no!
Have you created Information, no not really.
Is it internet based, a big cloud server in the sky. No!!
Does it work....yes. How big a dataset, as big as you want given disk space and speed. Just get a demonstration from Euclideon.

So that is what Euclideon etc are doing, and what Pointerra is trying to do. Euclideon describe it as Unlimited Detail (UD) and it is very impressive and it has been available to the public for maybe 5 years.

Is it what people want. Well a qualified ABSOLUTELY. Many government, survey, environmental and engineering groups are using this. It is a well established technology. Euclideon and Point Cloud Technology have made it infinite in size, in theory and to a degree in practice. It is a well known and used technology.

What is missing is the ability to do this in an analytical environment such Autocad and Microstation, for infinite datasets. A user of Euclideons Geoverse (a brilliant system) in Holland could visualise everything they wanted in Euclideon, but all their design work was happening in Autocad and ProEngineer. It was frustrating. They wanted a merging of the technologies.

There are issues in moving this technology to a cloud serve environment. Certainly adding the database into a cloud environment is trivial, and server groups like AMazon are brilliant. Initially they seem quite cheap. However when you start looking at significantly large data sizes, as espoused, when you start looking at multi user access (a technology and equipment issue), when you start looking a a multi core environment, when you start maybe using GPGPU processing power, then the costs start mounting significantly. You could go to a local server group, but why, your international footprint and sale size disappears. Costs must be past onto customers, who are already asking why.

However the most prominent problem is in bandwidth. No matter what your server is, and how quick it can generate the frames for your computer monitor, if it can't get it to you then the experience become frustrating and unusable. To take a simple case a 1080p movie at 25fps will require you to have an internet connection of 18Mbs. I don't know about you but my home connection is about 3-4, On 4G I get up to around 15Mbps sometimes more. And then there is the download volumes, at high speeds the volume costs become horrendous, and there is a lot of volume.

So it is unlikely to get very impressive performance for a 1080P let alone greater resolution. Without the use of certain dynamic level of detail techniques, (Euclideon use a dynamic level of detail technique for its solid scan product).

It does not matter what proprietary algorithm for 3D point data you are using, whether it is Euclideon's break through technique or Pointerra's untested variant. The issue becomes, if one wants to create a big server based cloud based 3D Point farm, it ain't going to be very impressive without some serious research into dramatic bandwidth reduction. To my knowledge Pointerra has not indicated any internal research, or utilising any if the research being conducted in the USA.

There are other work around options (which reduce the need for massive file downloads, but make downloads localised and fastish) but this removes the immediate immersion capability. There is nothing new or technically innovative in these ideas, but they do improve the solution.

Please watch Euclideon and PointCloudTechnogy, and others, are doing. It is out there now and moving. Euclideon have a staff of high quality and highly experienced technical innovators probably an order of magnitude bigger than Pointerra, and PointCloudTechnology has access to a world best research institute.

Just saying like...you know....do the maths..

(Again stating I have no interest in Pointerra, Euclideon or Point Cloud Technology)