Since I started learning to make shoes in 2004, there has always been talk of the promise of 3D imaging – the scanning of feet – and how this would greatly simplify the work of making shoes, mainly by facilitating quick and easy last customization. The idea was that a scan of a foot could be used to derive a last and make a shoe that would fit perfectly.
While “scan-to-shoe” hasn’t really panned out, thanks to a Make|Learn|Build! grant from the Oregon Regional Arts and Culture Council (RACC), I have been able explore first hand over the past year how 3D imaging and CAD software might fit in a custom shoe making workflow.
Grant funds were used to purchase a Structure Sensor Pro and an iPad, since the Structure Sensor Pro requires an iOS device to run. This scanner is based on infrared light, similar to the Kinect and other structured light scanners. With this handheld scanning setup, I acquired scans of 5 pairs of feet from clients who consented to participate in my pilot study. I also scanned lasts I modified.
Scans of feet must be manipulated in some way to make them fit into the digital last models. Most feet are scanned on a flat surface, but most lasts are not flat. A lift of the heel is needed and some lifting of the toes as well. This already requires some assumptions about how a foot might behave when manipulated in this way. Even though I could manipulate the scan of the foot enough to make it fit inside the bounds of the last, the scan data do not provide the right kind of information to know if this is actually possible for the given foot, or if it would be a comfortable and pleasant fit.
In practical terms scanning feet using a handheld infrared scanner is pretty fussy. It can take quite some time and the results vary in accuracy. It is difficult to capture the feet from every perspective in order to obtain a complete 3D scan. I used the same scanner to scan lasts, but the edge detection is not good enough to get clearly defined edges especially around the feather line.
My original plan was to use a customized CAD program offered by the German last manufacturer Spenle. They offered a stripped down last-only CAD environment and access to a library of their lasts, however this was not available to me despite my efforts to purchase a license. As an alternative, I used grant funds to purchase a copy of Rhinoceros 7, a very capable CAD program that is also well represented in the footwear world. This also meant I had to learn CAD for real in order to make a usable model of a last to send out for production.
I manipulated and adjusted the foot scans using Rhino’s robust mesh editing tools. I also used Rhino to align and orient the images in a way that might approximate the way a foot might sit in a physical last. I built last models with a combination of parametric and subdivision (SubD) modeling.
Typically lasts are purchased from a manufacturer’s catalog of existing shapes and selected on the basis of simple length and width measurements, which in themselves do not account for many of the idiosyncrasies of individuals’ real feet. In order to streamline the custom shoe fitting process, I wanted to see if lasts carved from a modified digital model of a last, with help from the scans of feet, could offer a better starting point. I did indeed do this, but only for one individual. Because I did not have access to the library of lasts from the last manufacturers, I had to buy scans of lasts and build my own models. For this individual, I built the model in Rhino based on a standard length and width and then modified it significantly using the scan of the feet as a guide. I sent my model out for carving and the lasts look good. Final fitting is in progress.
I also modified the physical lasts to fit and then compared the modified last with the scan of the feet to visualize feet in a last that fits according to the subjective judgment of the wearer.
Visualizing the Fit
For each person, trial shoes were made and the lasts adjusted as needed before the final shoes were made. The real test of whether the last fits is whether the shoe fits. Since I have some experience fitting feet, lasts and shoes, I thought it would be interesting to see how this fit is visualized.
The visualization of feet and lasts is not intuitive. While most people might imagine that a foot fits inside a shoe, and therefore should fit inside a last, when visualized on the computer the foot protrudes in all sorts of ways. An untrained observer trying to assess these superimposed images might assume a particular combination of foot and last does not fit when in fact it does. This is a limitation of computer modeling of elastic forms. The real world is elastic in ways that a surface scan is not. This is not to say that the computer can’t model elasticity, but that the wild variation of mobility and flexibility in human feet can’t be captured by a surface scan and modeled using only that. It’s important to set realistic expectations and know the limitations of the medium.
Below are images of feet and lasts and the shoes made on those lasts. Can you tell which of these fit to the satisfaction of the wearer?
These images are of works in progress. The trial shoes have been made and the results are good, but the final shoes have not yet been made.
Many Kinds of Fit
To state the obvious, shoe size alone does not determine the fit. The fit really depends on the shoe. Not just the shape of the shoe, but what the shoe is made of and how it is put together. Shoes of the same size but different materials and/or different construction techniques will fit differently. One person may also have many fit preferences. Sometimes they may want a loose fit and other times a tight fit. Shoes are designed and used for many types of activities and occasions. For some occasions the fit is as socially constructed as the look. None of this is readily visible even in a really good 3D scan of a person’s foot.
3D Imaging in Custom Shoe Making
With the examples I’ve provided here, it is possible to see how the digital images of feet and lasts can be combined to visualize the fit. With some insight and knowledge of anatomy and biomechanics, feet, and shoes, it’s possible to make sense of these images and discern the ways in which a particular fit might be displayed. The difference between a shoe that fits really well and one that does not fit all that well can be very small, and may not even be visible in the digital images.
Facial recognition software can do a decent job of identifying landmarks of a face, but it can’t tell you anything about the personality behind the face. Likewise, 3D images of feet on their own are not enough to tell the whole story. In many ways the blueprint, or Harris Mat print, provides more useful information than a surface scan of a foot. To get a complete picture of fit, you have to know more about the feet, the person, and the shoes they want.
In the vast array of information needed to successfully make custom shoes, 3D imaging of feet and last can play a part. Given the affordances and limitations of surface scans of feet, infrared scanners might be serviceable enough. Improved resolution can’t provide more information about what is happening beneath the surface. For capturing images of lasts, however, a more accurate scanner is needed. The most promising role for 3D imaging, in my opinion, is the ability to capture and reproduce lasts that have been modified to a wearer’s preference. To make adjustments to digital last models and to be able to compare these changes is valuable indeed.