An intelligent alternative to the solid modelling method

1214 0

Developing a product from scratch can be a long and drawn out process. Despite innovations in direct modelling the solid modelling approach still doesn’t lend itself to free-flowing design. So what’s the alternative? asks Martyn Day
The capture of design intent has been via many media over the last 5,000 years, from papyrus to 3D virtual models, analogue to digital. As a species we have managed to pull off many amazing feats of engineering without the aid of computer-aided-design. Just consider the Egyptian pyramids designed by architect Imothep, or marvel at the bridges, ships, tunnels and buildings that Isambard Kingdom Brunel created. You could argue that they were over-engineered but then if they had not been, they would not still be around for us to see now!

Today, we often get very wrapped up in the tools we use and, particularly from the CAD software vendors point of view, like to think that many designs wouldn’t be possible without these complex products. The truth is, despite what CAD has done for us and the improvements to manufacturing, we still have engineering disasters, designs fail and people still occasionally produce dumb products.

Yes, computer-literate designers have benefitted from our technological advancement of the CAD tools but the methodology of creating, testing and documenting the geometry isn’t particularly logical – you can tell this because they all require days of training and months of experience to gain proficiency. Engineers are still having to think about how they would model a part given the limitations of their current design system.

Jon Hirschtick, founder of SolidWorks once pointed out over lunch, that despite all of this CAD technology, there was nothing on the market that would quickly and satisfactorily allow him to model something as basic as a soup spoon. The complexity doesn’t just stop at the geometry engine and interface, all this data is stored in complex, proprietary file structures, of which there really is no reliable archive format for future generations.

In the last couple of years we have seen a resurgence of interest in more ‘flexible’ CAD tools which attempt to bypass the shackles of previous parametric systems by offering direct 3D modelling on ‘dumb’ NURBS geometry, such as PTC’s CoCreate, Siemens PLM Solutions’ Synchronous technology, SpaceClaim and Inventor Fusion. This can be very useful for making edits on legacy geometry from most CAD systems, but each attempt has its limitations and it’s all still based on solid modelling.

Is there another way?

I recently visited a company called Geomagic in Raleigh, North Carolina. The company has specialised in producing software which works with point cloud data – that’s to say 3D points from laser scanners. While primarily concentrating on software for quality inspection, Geomagic has recently developed some of the most exciting technology the CAD market has seen in the last five years.

Called Fashion, the software is capable of taking laser-scanned parts and with a little user-assistance, building editable solid models from dumb point clouds. This enables real objects to be imported into a CAD environment to be modified and re-manufactured.


The company’s enigmatic and visionary CEO, Ping Fu, believes that we are on the cusp of a dramatic change in the way we work and the way software can apply intelligence to geometry – by recognising blends filets, holes, rounds etc. With the second release of Fashion almost ready to be shipped, designers can now scan virtually any part or object and be able to use that as a base point for a product design. This will means there will be no need to start from scratch, to sketch, apply parametric constraints or add features. In addition, at any stage, the model can be saved as a dumb point cloud and the software will always be able to make sense and apply editing intelligence to the geometry. It is like magic.

Creating products based on natural shapes would be much easier in CAD land

Our discussion moved to the real sticking point of this technology, which isn’t the software, but the cost of laser scanning. While this has dropped from the hundreds of thousands of dollars to the twenty five thousands, it still inhibits the usage and keeps the price of point cloud data software high. Ping Fu believes that with the growing number of 3D still cameras and high quality DSLRs, it will soon be possible to just take a handful of flat or stereo digital images of an object and get a fully editable CAD friendly 3D model almost automatically as an output. This could bypass the existing expensive laser scanners but possibly drive development of lower cost scanning devices. I was told the software technology to do this is here, but the hardware and development would take about 18 months to two years.

With this new approach the whole argument of data reuse would become very interesting as you could capture and edit any 3D object with such a simple procedure. Creating products based on natural shapes would be much easier, although copyright issues would start getting quite complex.

Geomagic isn’t a massive player in the core product design tool market but with companies like Dassault Systèmes talking about ‘remixing designs’, it’s not such a leap to think that this technology could be embedded in a future generation of a popular CAD system.

Scale is also an issue, with Geomagic concentrating on small (teeth) to medium sized components (car parts). There is also a growing laser scanning market for big infrastructure projects likes roads, buildings and process plant (all nuclear power stations are scanned to compare the as-built to the plans). Ping Fu sees that there’s even potential to add ultrasonic and thermal scanning technology to trace in-wall hidden components too.


It might sound ‘dotty’ but i think there’s something to this technology vision. It’s bound to get easier to capture real-life objects and bring them into the computer, texture mapped and ready for use. The intelligence as seen in Geomagic’s Fashion product is still a demonstration of things to come but we are going to see radical changes to how we can capture 3D inspiration without clicking a mouse button.

{encode=” ” title=”Martyn Day”} is Consulting Editor of DEVELOP3D. He is currently trying to make his own 3D scanner out of two Nikon D90s, a slide rule and several rolls of gaffer tape. 

Martyn Day
Martyn Day asks: What’s best for free-flowing design?

Leave a comment