I came to laser cutting from graphic design. I knew Illustrator inside out — paths, anchors, bezier handles, the whole thing. When I started working in a company that ran CNC and laser machines, I assumed the transition would be straightforward. I already knew how to draw. How hard could the file preparation be?
It turned out the gap between a file that looks correct and a file that a laser machine can actually use is wider than most people expect — and crossing it costs real time and money if you do not know where the problems are.
The first attempt: export DXF from Illustrator
The obvious starting point was to draw in Illustrator and export directly to DXF. Illustrator has a DXF export option, it has been there for years, and on paper it should work. In practice, the files it produces are full of splines — a curve type that most CNC and laser controllers either cannot read or handle so poorly that the output is unusable. The machine either rejected the file outright or produced cuts that bore little resemblance to the original design.
So I tried exporting to PDF instead and converting from there. Same fundamental problem — the converters I found online either returned broken files with missing paths, or produced DXF files where every curve had been approximated with dozens of straight segments. Visually acceptable on screen. Mechanically problematic on the machine.
Why segments are a production problem, not just a technical one
This is the part that is easy to underestimate until you have actually run production jobs. A laser cutter processing a curve made of segments does not flow smoothly through the shape. At each node — each endpoint of each segment — the machine slows down, processes the next command, and accelerates again. On a circle approximated with 80 segments, that happens 80 times.
On a one-off piece, this is annoying but tolerable. The job takes a bit longer, the edge quality on curves is not perfect, but you get through it. On a production run of 1,000 or 2,000 or 3,000 identical pieces, the math becomes brutal. A difference of 3 to 4 seconds per piece — which is a realistic penalty for an unoptimized file on a moderately complex shape — adds up to nearly 3 hours of extra machine time on a 2,500-piece run. That is time the machine is not running other jobs. That is money.
Beyond time, there is the quality issue. On polygonal shapes — things with actual corners — the node hesitation blends into the geometry and is often acceptable. On curves it is not. The slight stutter at each segment boundary creates a visible faceting effect on the cut edge. On acrylic, on anodized aluminum, on any material where surface finish matters, clients notice. And when clients notice, you have a problem.
The manual fix and its real cost
The standard solution in most shops is to have a technician redraw the file in AutoCAD, replacing the splines and segment approximations with proper arc and line geometry. For a simple shape this might take 10 to 15 minutes. For a complex customer logo — the kind with interlocking letterforms, decorative elements, and curves everywhere — the same job can take four to six hours.
And customer logos in PDF are not a rare edge case. They are a daily reality for most laser cutting operations. A client sends their brand file, expects you to cut it, and has no idea that what lands in your inbox needs hours of technical preparation before it can go anywhere near a machine. That preparation time is invisible in quotes, unpaid in most cases, and quietly absorbs a significant chunk of the technical team's capacity every week.
A friend of mine runs a laser cutting company and described exactly this situation before finding a better workflow. His technical office was spending so much time redrawing client PDFs that it was becoming a bottleneck — jobs were backing up not because of machine capacity but because of file preparation. Every new client with a complex logo meant another few hours of someone's time before production could even start.
What a production-ready DXF actually needs
The requirements are not complicated, but they are specific. A DXF file that works cleanly on a laser cutter needs:
Real arc entities, not segment approximations. Every curve should be represented as an ARC or CIRCLE entity — a single geometric object with a defined center, radius, and angle range. Not a polyline. Not a spline. An arc.
No splines. Splines are mathematically elegant but practically useless for most CNC controllers. If your file contains splines, they need to be converted before the file goes to the machine.
Closed paths where needed. Open paths cause problems on laser cutters — the machine cuts a line instead of a shape, or skips the path entirely. Paths that are meant to be outlines for cutting need to be properly closed.
Correct scale. DXF files can be in any unit system. Make sure the file is in the units your machine expects — usually millimeters — and that the dimensions match the actual intended size of the part.
The workflow that actually works
Starting from a vector PDF — which is what you get when a client sends a file exported from Illustrator, CorelDRAW, Inkscape, or any proper design tool — the workflow that consistently produces clean results is:
1. Convert the PDF to DXF with proper arc fitting. This is the critical step. The converter needs to analyze the Bezier curves in the PDF and rebuild them as arc sequences, not just sample points along the curves and connect them with segments. PDF2Laser does this automatically using biarc fitting — the mathematical technique that finds the closest arc approximation for any given curve.
2. Open the DXF in your CAD software for a quick sanity check. Verify the dimensions, check that all paths are present and correctly positioned, and confirm the scale is right. This takes a few minutes, not hours — you are checking, not redrawing.
3. Send to the machine. With a properly optimized DXF, there is nothing left to fix. The file goes straight to the CAM software and onto the machine.
My friend's company switched to this workflow and the difference was immediate. File preparation time dropped from hours to minutes. The technical office stopped being a bottleneck. And the cut quality on curves improved because the machine was finally running smooth arc interpolations instead of stuttering through hundreds of segments.
When this workflow does not apply
One important caveat: this only works with vector PDFs. If the PDF is a scan — a photograph of a drawing, a document printed and then scanned back in — there is no vector geometry to extract. The file is just a raster image, and no converter can produce clean DXF geometry from pixels alone. In that case, manual redrawing or AI-based tracing tools are the only options, and neither produces perfectly clean geometry without some intervention.
If you are not sure whether your PDF is vector or raster, try zooming in very close in a PDF viewer. If the edges stay sharp at any zoom level, it is vector. If they become blurry or pixelated, it is raster.
Have a PDF ready to convert?
Try PDF2Laser — it's free