Capturing the shape of the human body is surprising difficult, especially since we change our shape all the time. Whether we are sitting, standing, walking, or running, we change our shape continuously. On the inside, our muscles help us to breathe, our hearts beat, and our arteries pulse. And as we continue to learn more about life, from the human scale to the molecular scale, we find that form is a significant driver of function.
As part of the effort to build a complete human model, with bridges between scales within the human form, the Parametric Human Project partnered with Quantum Capture in Toronto to create a detailed surface scan of Andrew Hessel, a futurist and catalyst in biological technologies. A photogrammetry process was used to derive detailed 3D geometry from a vast array of photographs. This method has many advantages as it is non-invasive, extremely fast (the speed of taking a single picture), and above all, does not change the shape of the subject.
After both manual and automated processing of the myriad of photographs, the resulting 3D geometric model has some interesting properties. For example, the cornea of the eyes is not captured but the pupil seems to be represented geometrically. Wrinkles, for example on the forehead, are very thin but are visible in 3D, whereas hair and eyebrows are only roughly modeled, due to interference in the shading between strands.
Still, the results are impressive and, when the geometry is textured with colours from the original photographs, the reconstruction of the surface geometry is incredibly lifelike.
Now that we have a high resolution outer shell of Andrew, we can calculate his volume, height, and other traditional anthropometric measurements. But a more advanced set of digital measurements becomes possible and could automatically contribute to phenotype databases.
The next step is to fill this shell with the volumetric data of an MRI scan. Unfortunately, when Andrew will be laying down in the MRI scanner at York University in Toronto, he will not be the same shape as he was when the photogrammetry capture was done. To map the surface scan to the volume scan, corresponding features between the two will have to be labelled so that a registration process can deform the surface to the volume. Then, these two data sets together will form the context for more specific scans.