For me, one of the most intriguing areas of development in 3D modelling technology has been its recent use in digital heritage. Specifically, the use of digital scanning to create detailed records of ancient monuments and artefacts. After the disaster at Notre-Dame de Paris nearly one year ago, much was written about the existence of a detailed record of the building after a digital scan completed by pioneering researcher Andrew Tallon. That survey may prove invaluable in restoring this great work of architecture. Institutions such as CyArk are doing similar work documenting other important sites around the globe, safeguarding valuable treasures in a digital vault for future generations.
Photogrammetry received a quick mention in the Noroff course material, and I jumped at the chance to try it out at a very basic level. The concept is that an accurate 3D mesh or point cloud model can be created from a series of 2D digital photographs. The potential applications for this extend from digital heritage to complex topological surveying, scene creation for 3D animation and a great deal more.
Autodesk have a piece of software called ReCap, which is what I’ll be using to create a scanned model. This is compatible with other Autodesk products, meaning scanned models can be imported into the likes of AutoCAD or Revit. I’m eager to test out the possible applications for architectural design here, but thought to get started with a simple object first.
Below is a little rubber frog, a replica of the one owned and carried as a lucky charm by composer Edvard Grieg whilst performing (and incidentally was bought from the gift shop at his former home here in Bergen). I chose this because of its rough texture and organic shape, meaning it should be a suitable object to scan with a regular iPhone 6 camera (which has seen better days!). Smooth surfaces are apparently difficult to capture, whilst an object based on simple geometry rather defeats the point of wanting to create a scan in the first place. So Grieg’s frog it is.
These are the set of images I took on a window ledge (mostly to get good natural lighting on the object), which were then imported into ReCap photo (a side module to ReCap specifically to handle the photographing/stitching aspect).
It took a little while to stitch the images together, and things seemed to be moving slowly. I left the software with a ‘waiting in cue 1% complete’ message and came back a day later to find a completed model. I then had various options for patching up the model, and various formats for exporting including video/png.
The model itself came back with a few incomplete areas, and very much of the window ledge which I trimmed away to just the rectangle you see. The white areas on the frog shows how important it is to get photographs from all possible angles. I made that difficult by having it on the window ledge, and therefore the back and underside of the frog are not shown accurately. The photo points are shown in the screenshot below by the triangles, mostly from above the model.
I had previously read about others having problems due to moving the object in relation to the background. For the scan to work, either the object or the camera has to remain still. In my case, having the window and the ledge there made getting a full orbit difficult. Trying again with the object elevated on a small platform of some sort would perhaps produce a better result.
For a first attempt, this was quite an exciting experiment! Room for improvement of course, but the applications for this technology on even a very low budget scale could be very useful. If one begins to look at the scan below as a potential building form, the potential uses for application in design process become more apparent. It allows a designer to get hands on with a prototype or physical model, and then scan the geometries to reproduce them digitally. Until this, the exchange of CAD data to the real world had mostly been one way but now it seems there is a very accessible, affordable and simple way to input complex geometries back into the digital modelspace and become part of the workflow.
Nick Walkley 