Photogrammetry anyone?

Has anyone had any experience using Photogrammetry to try to capture design data. I have been looking at one called Photomodeler and it seems to allow you to capture some pretty decent data with nothing more than a digital camera. I know it isn't going to be comparable to a laser scanner but for a rapid way to capture photographic measurement data it looks interesting. We have to go out to measure customers plants and equipment and quite often there is a surprise gotcha that can be annoying or quite expensive. We take lots of pictures as well as measurements with tape measures but sometimes when we get back to the office we see that there is a beam or pipe that we forgot to measure.for interference purposes.







Re: Photogrammetry anyone?

our company is currently evaluating laserscaning and Recap application. Hopefull,upcoming ST10 can also proceed with point cloud based files...

Re: Photogrammetry anyone?

The output file type is the important thing here. Many of these apps are sophisticated enough to generate .stl files or even surfaces from the scanned data. Solid Edge ST10 not only imports .stl data but also allows significant use of the geometry during later modeling operations. The simple answer is that if the app outputs a compatible file format, Solid Edge can use the data.
With ST10 we do have some capabilities to edit mesh data that comes with stl output.
Mark D. Thompson
Application Engineer, Global Technical Business Development
Mainstream Engineering

Re: Photogrammetry anyone?

Turning laser scans into 3D models is a very BAD way to do things.


Using laser scanning to inspect and compare manufacturing to CAD is great.


Photogrammetry is better than laser scanning if you just want basic dimensions.


If you do gown the road of turning laser scans into models, your software choice and resultant product need a very narrow definition. (Primitives, Nurbs, Polygons, Tins, etc...)

Every combination of scanner and product use a different set of software, each seat costing $30,000 each.


For example: Lets say you are using a Cyrax and Cyclone. You go to the field and scan a steel beam. If the bream is straight, that software and scanner will do just fine. But if the beam is curved, you better be prepared to move the point to another software and start tracing and modeling nurb curves. Lot's of procedures and file format issues to work out. The time to reverse engineer the beam just went from 5 min to 5 hours because it was curved. To do anything, you will been about 5 different $30,000 pieces of software each with a 6 months learning curve. How much do you want to spend?


Here are the big boys of point clouds

Imageware (NX add on): Nurb curves from point, the only software that can project curves through couds

Geomagic: Organic closed polygons

Cyclone: Extruded shapes (Cyrax scanner only)

Rapid form-ok at many things

Rhino - for very small uner 2K points


Most scanners generate 50 Meg files that are 1 million point. Most of the time you need 4 to 16 scans to get 90% coverage of an opject. Can your computer rotate 16 million point while viewing them all? Yes you will want this!


Point clouds want computers faster than can be purchased. I would not recommend anything less than a top of the line BOXX and 2 gig optical connections to each computer.


Re: Photogrammetry anyone?

I have been trying to do some tutorials in the PhotoModeler software. Still trying. I have also been asked to look at the faro laser scanner so point clouds are important. And I hear what you are saying about monster computers being needed for point clouds.


Anyone know anything about the SMAP Scan to Cad app?


The machines we want to scan or capture some design data from are sometimes on support structures up in the air on second storys so the laser scanners have some attraction. They have lots of holes in the sides that we would like to be able to measure as well as the support points and things like that. I don't know if we will ever be able, at least at this time, to be able to measure the bolt holes with enough accuracy. I believe the laser scanners would only be accurate to plus or minus 3mm. Guess its time to talk to Faro again.


We own a Faro 12 ' arm but that isn't much help in these situations.




Re: Photogrammetry anyone?

The accuracty of laser scanners is typically (1/10,000 x the range) of the scanner. So you will want a scanner that has enough range for your situations, but no more,


Cylenders are the best objects referse engineer from laser scanners. So if the holes you want to measure are round, you are in good shape. If not, you will end up tracing point in cloudworks.AutoCAD.


The gaem of Dot to Dot will never be the same.