COFA PhD candidate Hamish Dunlop and I have been sharing an enthusiasm for facets. His research involves building a 6-axis mill, with the target of cutting his own component parts for faceted 3D models.
Here’s a work he produced for his final MFA installation, with construction detail:
Wolfie on the left side of the image is made up of 432 unique 8-sided polygons. To convert him from polygon mesh to solid polygons, I wrote a plug-in for Rhino. A CNC company in NZ cut him out.
Hamish Dunlop
I’ve really fallen in love with the aesthetic and tactile qualities of faceted objects. I created a beautifully faceted object accidentally when I sent a 3D reconstruction to be printed, and discovered the backside of the model:
Detail from the Victoria and Albert Museum Cherub Reconstruction, Josh Harle 2010
From here, I was inspired to explore ways of achieving larger scale, cheap, faceted 3D models. An obvious avenue to investigate was papercraft; unwrapping 3D models into a 2D, foldable net/mesh and then folding it together into the final model:
Harry Portrait, Bert Simons
I made a number of experiments using papercraft meshes as molds for cast plaster objects. At the time, I was mainly interested in the possibility of producing the models with photorealitic colour. I used colour transfer paper, with an inverted mesh (the detail printed on the inside and folded with the tabs outside), painted the finished model in latex, and filled with plaster. This gave varying degrees of success in terms of colour transfer, but always gave great results as interesting objects. The intersection between planes in these papercraft models are emphasised by the folds in the paper, so they look and feel great:
Concept exploration, Josh Harle
I’m also interested in using the papercraft unwrapping software to take large-scale 3D reconstructions of environments, built using photogrammetry, and cut them into 5mm MDF, to create spaces. I’m keen on looking at something like this wonderful work, but as en enveloping space:
Fragments of memory, Twana Sivan
Finally, here’s a look at other forms of interesting reduced/simplified geometry. These are taken from an exemplary student of course I am tutoring called Experimental Modelling. The final stage of the course is the processes of fabrication either with 3D printing or laser cutting – Good Fun!
Project renderings, Matt O’Brien 2011
This is incredibly interesting experiment! We have been trying different but close solutions at http://www.my3dscanner.com but were not that successful. We actually used http://www.tamasoft.co.jp/pepakura-en/ It’s useful software which automatically creates paper craft from 3D models which we generated with photobased scanning. The problem was a large number of parts to glue to keep 3D model more or less detailed. I am sure you know Pepacura but just in case.
Hi Paul,
I’d recommend using either PhotoSynthToolkit of SFMtoolkit provided by Henri Astre (http://visual-experiments.com) to reconstruct your 3D models.
Pepakura is actually the software I was using once I had a 3D model
A few limitations in terms of texture size and polys, but a good place to start.
Josh,
Thank you. Actually my3dscanner.com works great for me (I tried Toolkit as well).
The problem I have is big number of polygons when I want to turn mesh into something physically doable. Mesh decimation kills all the details. So my idea to start with scanning and then reduce number of facets is probably not great, it’s much easier to model from scratch if simplified geometry is needed. But I started with human faces and wanted them be at least tangentially recognizable. Probable FaceGen will work.
Paul,
I’ve used FaceGen; it’s a great program for coming up with good-looking face meshes, but it still produces a fairly high poly count, and the symmetry of the final mesh probably will look a bit odd in the real world.
What decimation are you using? Multires modifier in 3D Studio Max is pretty good, and I especially like the “Quadratic Edge Collapse Decimation” available in Meshlab for reducing poly count of organic-type objects.
Well, I have only started 3DS Max (and only because of your Camera Projection experiments).
I tend to use various Meshlab filterings (Clustered vertex subsampling more than others). I also use CloudCompare (which is very fast and efficient) and Geomagic (it has good tools for sub-sampling point cloud or decimating mesh).
My point is that going from dense point cloud (say by my3dscanner.com) to sparse point cloud (say by photosynth), or decimating dense mesh probably makes less sense than using any modeling software to create low poly model for brilliant physical reconstructions described in your post.
I agree – those computer game artists managed to breath life/realism into a handful of polys!