Lesson 0.4Lesson 0.4 · Foundations of Digital Fabrication
Reading Fabrication Data
What a machine actually needs - vector, mesh or solid; layers as cut-vs-engrave; units, origin and machine limits; and the preflight before you press go
The machine does exactly what the file says - which is wonderful right up until the file quietly says the wrong thing.
A machine has no common sense. It will not notice that your part is in inches, that the origin is off the bed, that a stray line will cut where you meant to engrave, or that the drawing is twice the size of the material. It reads the data and it executes - flawlessly, into the wrong result.
So the last foundational skill, before we meet any specific machine, is reading fabrication data: knowing what a machine actually needs from a file, in what form, and how to sanity-check it before you commit material and time. This is the preflight that separates makers who waste one sheet learning from those who waste twenty.
The machine has no common sense. The preflight is the common sense you bring to it.
Three kinds of geometry a machine reads
Machines do not read your beautiful rendered model; they read one of three kinds of geometry, and using the wrong kind is a common first mistake. 2D vector geometry - closed curves and lines, usually from DXF or SVG - is what laser cutters and 2D CNC need: paths for the tool to follow across a flat sheet. What matters here is that curves are truly closed and not accidentally doubled; an open path or two lines stacked on top of each other confuses the machine into skipping or double-cutting.
3D mesh - a skin of triangles, from STL - is what 3D printers slice into layers. A mesh must be watertight (manifold): no holes, no flipped normals, no self-intersections, or the slicer cannot tell inside from outside and the print fails. Solid geometry - a true B-rep with faces and volume, from STEP or native files - is what 3-axis and multi-axis CNC CAM prefers, because CAM needs real surfaces to compute toolpaths against. The rule: match the geometry type to the process before anything else. Sending a mesh to a CNC shop, or an open polyline to a laser, creates work and errors that no amount of careful machining can fix later. A useful way to remember it is by what each machine wants to be told: the laser wants a path to follow, the printer wants a volume to fill, and the mill wants a surface to carve down to. Give it the geometry that answers its question, and everything downstream gets easier.
Vector for the laser. Watertight mesh for the printer. Solid for the mill. Match first.
Layers and colours are instructions
On sheet machines, the file does more than describe shape - it encodes what to do with each line, and it does so through layers and colours. The universal laser-cutting convention is that different colours or layers mean different operations: commonly red for cut-through, blue for score or engrave outlines, black for raster engraving. The machine software maps each colour or layer to a power, speed and mode. Put a line on the wrong layer and the laser will engrave what you meant to cut, or cut a fold line you meant only to score.
The same idea runs through CNC and printing in different clothes: a toolpath is the concrete instruction the machine follows, and its ordering matters - you cut internal holes before the outer profile, otherwise the part comes loose and shifts while its details are still being cut. Getting this right means treating your drawing as a set of labelled instructions, not just an image. Before export, audit every layer: is each line assigned to the operation you intend, in the order you intend, with nothing stray on a live cutting layer? A single forgotten construction line on the cut layer has ended many a promising job with a slice straight through the middle of a finished panel.
Red = cut, blue = score, black = engrave. The colour IS the command.
Units, origin and coordinates
Two invisible properties ruin more parts than any tool ever will: units and origin. A machine works in a coordinate system with a defined zero - the origin or work offset - and every move is measured from it. If your file assumes the origin is the bottom-left corner of the sheet but the machine is homed to the centre, the toolpath runs off the material. Always know where your zero is, and set the machine's work offset to match the file.
Units are the classic silent killer. Machines read numbers, and G-code even has an explicit units mode - G20 for inch, G21 for millimetre. A part modelled in millimetres but interpreted as inches comes out 25.4 times too big; the reverse makes it vanishingly small. A concrete example: a 100 mm bracket sent as inches becomes a 2.54 metre monster that instantly exceeds the bed. Before export, confirm the document units, the CAM units and the machine expectation all agree. The third coordinate to respect is the Z origin on machines that have depth - the top of the material versus the bed - because a wrong Z zero either cuts air or plunges the tool through the workpiece and into the table. Units, XY origin, Z origin: three numbers, checked every single time.
Units, XY origin, Z origin. Three numbers. A wrong one scales or crashes the whole job.
Machine limits - bed, travel and nesting
A file that is geometrically perfect can still be unmakeable because it ignores the machine's physical envelope. Every machine has a finite bed size (a laser might be 600 by 400 mm, a router 1220 by 2440 mm to suit a standard sheet) and a maximum travel in each axis, including a limited Z height - the tallest thing a printer can build or the deepest a spindle can reach. Geometry outside that envelope is simply not reachable; the machine hits a limit switch and stops, or clips the part.
Sheet work adds two more envelope questions. Nesting is how you arrange parts on the stock to fit within the bed and waste the least material - good nesting can be the difference between one sheet and three, and it is worth rotating and interlocking parts to close the gaps. And you must leave room for clamps, hold-downs and the machine's own margins, since the full bed is rarely usable edge to edge; a router in particular needs its work held down clear of the toolpaths, or the cutter meets a screw. On a 3D printer, the equivalents are staying within the build volume, keeping a first-layer footprint that adheres, and respecting overhang limits from the last lesson. If a part is genuinely bigger than the bed, the envelope forces a design decision: split it into pieces with joints, or move to a larger machine - better discovered now than after a failed run. The habit to build is to always place your geometry inside a drawn rectangle of the real bed, with margins, before you believe the file is ready. Fitting the envelope is a design constraint, not an afterthought.
Draw the real bed as a rectangle and place parts inside it - with margins for clamps.
The preflight - sanity-check before you press go
Pilots do not take off on faith, and neither should you press start on faith. A preflight is a short, repeatable checklist you run on every file, and it turns most fabrication disasters into a ten-second catch. Run it in order. Geometry type: is this the right kind - vector, watertight mesh, or solid - for the process? Units: do document, CAM and machine agree (mm or inch)? Origin: is the XY zero where the machine expects, and is the Z zero the top of the material? Layers and colours: is every line on the correct operation, holes before profiles, nothing stray on the cut layer? Envelope: does everything sit inside the real bed with margins for clamps, and within travel and Z limits?
Then the physical checks the file cannot show: is the material the size and thickness you assumed, secured flat, and the right side up? Is the tool or nozzle the one the toolpath expects? Many machines let you run a frame or dry run - the head traces the bounding box with the laser off or the spindle up - which is the single best final check that the job lands on the material. None of this replaces proper training and supervision, especially with lasers, spindles and moving robots, which are genuinely dangerous and must be run under the rules of your fablab. But a disciplined preflight is the cheapest insurance in all of fabrication: run the list, trace the frame, then press go.
Preflight: geometry, units, origin, layers, envelope, material, tool, dry-run. Then go.
Watertight (manifold) mesh
A closed STL with no holes or flipped normals
A slicer needs a watertight mesh to tell inside from outside. Non-manifold geometry is the most common 3D-print failure - repair before slicing.
Cut / score / engrave layers
Colour or layer conventions on a laser file
Commonly red cut, blue score, black raster engrave. The colour maps to a power and speed; a line on the wrong layer does the wrong operation.
Origin / work offset
The zero point every machine move is measured from
Set the machine offset to match what the file assumes. A wrong XY or Z origin runs the toolpath off the material or crashes the tool.
G20 / G21
The G-code inch / millimetre units mode
A mm part read as inches is 25.4 times too big. Confirm document, CAM and machine units all agree before export.
Bed size / travel
The physical envelope the machine can reach
Geometry outside the bed or beyond axis travel simply cannot be made. Place parts inside a drawn rectangle of the real bed, with margins for clamps.
Frame / dry run
Tracing the job bounds with the tool disengaged
The head outlines the bounding box, laser off or spindle up, to confirm the job lands on the material. The best final check before you press go.
Workshop — write and run a fabrication preflight
The deliverable of this lesson is a habit, so the workshop is to build and rehearse it. Turn the checks into a checklist you can run in under a minute on any file, then pressure-test it against a deliberately broken example.
Pen and paper are enough for the checklist and nesting. Optional and free: a 2D editor (Inkscape) to practise layers, or a slicer (Cura, PrusaSlicer) to test a mesh. Always work any real machine under fablab supervision.
Goal: build a reusable preflight checklist and catch planted errors with it Inputs: a simple 2D drawing (real or sketched), pen and paper, optionally free CAD and a slicer Time: ~30 minutes
- 1Write your preflight as an ordered checklist: geometry type, units, XY origin, Z origin, layers and colours (with holes-before-profile), machine envelope, then material, tool and a dry-run frame. Keep it to one card.
- 2Take a simple laser drawing (or draw five parts). Assign every line to a cut, score or engrave layer deliberately, and confirm all cut curves are closed and none are doubled.
- 3Now sabotage a copy on purpose: change the units to inches, drop the origin off the bed, and leave one stray construction line on the cut layer. Hand it to a friend - or your future self - and run the preflight to catch all three.
- 4Draw a rectangle at the true size of a real machine bed you can name (say 600 by 400 mm) and nest your parts inside it with a margin for clamps. Note whether they fit on one sheet or need two.
- 5If you have access, open an STL in a free slicer, run its mesh-repair or analysis tool, and see whether it reports the model as watertight - the 3D equivalent of the same discipline.
You’ll walk away with
A one-card preflight checklist plus a short note proving you caught all three planted errors and correctly nested your parts inside a real bed. This card is the thing you will actually run before every real job.
Three altitudes on the same idea
Read the band that fits you — or all three.
A fabrication file is a contract, and the machine reads it literally. When you issue geometry to a shop, the units, origin, layer conventions and envelope are the difference between a component that arrives right and one that arrives wrong at full scale. Understanding what the machine needs lets you hand over data a fabricator can run - and lets you catch a costly error before a sheet of material dies for it.
One stray line or a wrong unit can ruin a bespoke panel. Whether you outsource a laser-cut screen or run a desktop cutter yourself, the preflight - closed vectors, correct cut-versus-engrave layers, millimetre units, parts nested inside the real bed - is what protects the finish and the budget. It is the unglamorous habit that makes made-to-measure reliably come back right.
Reading fabrication data is the skill that saves your material budget. In every fablab, the people who waste the least stock are the ones who preflight: match the geometry to the machine, check units and origin, audit layers, place parts inside the bed, and trace a dry-run frame. Build that ten-second checklist into muscle memory now and you will look like you have been making for years.
“If the file opens and looks correct on screen, the machine will make it correctly.”
Do it yourself
No machine - run the checks in your head.
- 1Which geometry type does a laser cutter need, which does a 3D printer need, and which does CNC CAM prefer?
- 2What does it usually mean if a laser engraves a line you meant to cut?
- 3Why must internal holes be cut before the outer profile?
- 4A 100 mm part comes out 2540 mm long - what almost certainly happened?
- 5List the steps of your preflight, in order, from memory.
The one line to carry out
Peer-reviewed journals & authoritative standards
- 01Engineering tolerance — Wikipedia, 2026.
- 02Nesting (process) — Wikipedia, 2026.
- 03Design for manufacturability — Wikipedia, 2026.
- 04The Fab Foundation — the global Fab Lab network — Fab Foundation, 2026.
That completes the foundations: you can now see any process as a file-to-factory pipeline, place it in a family, and read the data a machine needs. From here the course gets specific - Module 1 opens the CAD-to-CAM bridge in depth, turning your models into real toolpaths and G-code.
The author
Amogh N P
Architect, interior designer, and creative polymath. Studio Matrx began in his notebooks — his vision of design made honest, useful, and open to everyone. Its Academy is written and taught in his memory, and free, forever.
More about Amogh →