The craft of a corrected cast, held to a tolerance.
Before CAD, an orthotic was made over a plaster positive that a technician balanced by hand. We wrote our own software to do that job digitally rather than approximate it.
What balancing a cast actually was
The negative cast was poured and the positive came out of it. A technician then read the forefoot-to-rearfoot relationship off that positive and added plaster until the foot sat correctly. The device was formed over the corrected positive. The quality of the orthotic was the quality of that balance.
That operation is what our software reproduces. The geometry it produces was developed against an annotated plaster positive from this lab, and matched to it — it was not designed from a template and checked against a cast afterwards. The cast was the specification.
Who wrote it
Our own podiatrist did, over hundreds of hours and several hundred revisions, and he is still writing it. Not a software company with a podiatry consultant on call, and not a general CAD package with an orthotics menu bolted to the side. The person who decided what the geometry should do is the person who treats the feet it is for.
That is the whole reason it behaves the way it does. Every operation in it is there because the plaster workflow had an equivalent, and the test of whether an operation was finished was whether the surface it produced matched the balanced positive this lab had already been making by hand for years. The old method was not something to modernise past. It was the specification to hit.
What that buys you is a tool with no general-purpose controls in it. Heel width, forefoot width, arch height, skive, flange, apex position and trim line — the things a prescription actually says — each applied to the left and the right independently, each held in millimetres and degrees against the foot it was measured from. Nothing is averaged across the pair, because feet are not a pair of averages.
The device that comes out is made to that patient's foot, rather than a stock shape adjusted until it fits.
Four things it does
The scan becomes a corrected digital cast
Not a shell with corrections layered onto it. The scan is turned into the digital equivalent of a corrected plaster positive, and the orthotic is designed against that surface — the same order of operations the plaster workflow used.
The forefoot is measured, not assumed
Forefoot varus and valgus are measured from the patient's own scan, in millimetres and degrees, from the true height difference between the first and fifth metatarsal heads.
The real forefoot tilt survives alignment
Alignment is yaw-only. The patient's anatomical tilt is preserved through it rather than flattened to a plane — which matters, because the relationship you need in order to balance the foot is the one a flattening alignment destroys before design starts.
The post and the arch fill are one surface
The forefoot post sits at the height of the lower metatarsal head, and the post and medial arch fill are built as a single continuous surface unioned into the cast — the way plaster is added to a positive, rather than a wedge placed under a shell.
Half a millimetre
The corrected cast is held to 0.5 mm of the source scan in every axis. Smoothing is forbidden on the plantar surface, and any surface the reconstruction extrapolated beyond the real scan is trimmed away rather than kept.
Accuracy figures in this industry describe scanners and printers. This one describes the design: how closely the corrected cast follows the foot that was scanned.
Craft and repeatability are not opposites
This is the method a good technician performed by hand, written down precisely enough to run. The clinical judgement is still yours, and the finishing is still done by hand. What changes is that the operation is defined, so the second pair is geometrically the same as the first.
Measured from the scan
Forefoot varus and valgus in millimetres and degrees, from the real first-to-fifth metatarsal head height difference.
0.5 mm, every axis
The corrected cast's tolerance against the source scan. No smoothing on the plantar surface.
Stored, not redrawn
The design file is kept, so a repeat pair is the same device, not a second attempt at it.