
01 / 07 · THE MARK, POURED · CLIP
One mark, taken from vector to copper.
A Houdini simulation pours the monogram on; a diffusion pass finishes the surface.
The simulation decides where everything goes. The model only decides how it looks.

02 / 07 · THE OBJECT · PLATE
The business card is a circuit board.
88.9 × 50.8 mm, black substrate, bare copper, gold finish. Two fabrication runs so far.
Rendered from the files the fab house received. The cards themselves are in circulation.

03 / 07 · CODE TO FAB · PLATE
Generated from code, checked before it ships.
The artwork is written as code, converted to board files, and held to 363 automated checks before an order.
Re-deriving the chip pinout from the datasheet caught a power pin on the wrong pad. A passing suite had missed it.

04 / 07 · REVISION · PLATE
Next revision: the card carries its own antenna.
A three-turn perimeter coil, an NFC tag, and an LED lit by the phone’s field.
Boards are fabricated and on the bench. Assembly and antenna tuning come next; nothing here taps yet.

05 / 07 · DIRECTED, NOT FOUND · CLIP
The gray render is the instruction.
Left to itself the model invents a scene. Given the clay render as control and one reference frame, it holds the geometry and paints the metal.
25 variations were run and ranked blind before one was kept.

06 / 07 · LOOP DISCIPLINE · CLIP
63 loops against a written spec. Not converged.
A second shot, a drill breaking through the board, runs unattended against an acceptance spec with a stop rule.
The reference reads as punched through; this still reads as fragments pushed through a hole. The loop stopped itself and is waiting on direction.
- 363automated checks before a fab order
- 25directed runs, judged blind
- 63loops against one written spec
| Item | Reading | State |
|---|---|---|
| Card, run 1 | 6 boards, black substrate, gold finish | In circulation |
| Card, run 2 | 6 boards, 363 / 363 checks, corrected URL | In hand |
| NFC revision | 3-turn coil, 0 design-rule violations | Assembly next |
| Pour simulation | 144 frames, Houdini | Done |
| Diffusion finish | 25 runs, ranked blind | Holds on the mark |
| Finish on the full card | small type garbles on the last frames | Open |
| Drill-through shot | 63 loops, 3 locked lenses | Not converged |
07 / 07 · BENCH LOG · LOG
What works, and what does not yet.
The method is the experiment: a simulation or a spec gives direction, a model or a loop does the volume, and a check decides what is kept.