I converted the same photo twice and got slightly different results — why?

By design. AI models are non-deterministic — every run makes small interpretive choices about which lines and curves best represent the subject. Like asking two draftspeople to redraw the same photo, the result is close, never identical. If a particular conversion looks right, save the file; running it again will give you a similar but not pixel-identical drawing.

My upload failed, was rejected, or came back blank — what happened?

A few things can cause this. Our AI declines to convert some content — graphic violence, certain public figures, copyrighted artwork. It also struggles with very low-contrast images and extremely busy scenes where there's no clear subject. If your conversion fails, try a clearer photo, better lighting, or a tighter crop on the subject.

How can I remove the background or isolate the subject in my photo?

Turn on the Remove Background toggle before converting. The AI focuses on the main subject and strips out the surrounding scene — useful for clean outlines, product shots and decals where you only want the subject in the final vector.

What does the Detailed toggle do?

Detailed mode tells the AI to extract more detail from your photo — useful for intricate or busy images where Standard simplification would drop too much. Standard mode is the right choice for most CAD use cases — it produces the appropriate level of simplification for manufacturing and design work.

Can I edit the file after I download it?

Yes. The output is real editable vector geometry — polylines and curves you can modify, dimension, recolor, extrude or parameterize in any vector or CAD program. Open it in Illustrator, Inkscape, AutoCAD, Fusion 360, Rhino or whatever you use, and treat it like any other drawing.

What size will my drawing be when I open the SVG or DXF?

Your drawing opens at real-world scale — about 300 mm on its longest edge — plus a 5 mm margin all around to keep the linework clear of the viewport edge. You can scale up or down freely in any vector or CAD program without losing quality.

What paper size can I print the PDF at?

Any size. The PDF is pure vector — it scales without quality loss. Print at A4, A3, A2, or larger on a plotter. The drawing is sized to about 300 mm longest edge by default, so it fits comfortably on A4 with room to spare; for larger output, scale up in your print dialog.

Why is my output black-and-white?

PhotoCAD produces line drawings — the geometry of the subject as a vector outline, not a color reproduction. If you want color fills, open the SVG in any vector editor and apply them after import.

Does it work for photos with multiple subjects?

Yes. The AI captures all the subjects in frame. For busy compositions, try Detailed mode to retain more of the geometry. Single clear subjects against a plain background convert cleanest — but multi-subject photos absolutely work.

How long does a conversion take?

Typically around 110 seconds — just under two minutes. Simpler subjects come back faster; detailed or busy images take a little longer.

How it works

From photo to manufacturing-ready vector

Upload a photo and PhotoCAD's AI reads the subject and returns it as a clean, editable vector drawing — DXF, SVG and PDF, ready for CAD, cutting and design. One automatic step, usually under two minutes.

Try it on your photo Jump to technical specifications →

How a conversion works

One automatic process. Upload, AI converts, download — usually under two minutes.

Step 1

Upload your photo

Any photo, any subject — JPEG, PNG, HEIC, WebP. iPhone photos work natively. Your image is normalized in your browser before it leaves your device. Optional toggles: Remove Background isolates the subject; Detailed extracts more detail from busy or intricate images.

Step 2

AI converts it to a CAD drawing

Our AI semantically understands what your photo depicts and produces a clean, CAD-ready vector drawing of it. Unlike traditional tracing, it isn't confused by shadows, tone or texture — it captures the real geometry of the subject.

Step 3

Preview and download

Review the result in your browser before you commit a credit. Then download DXF, SVG and PDF — same geometry in three formats. DXF for CNC and CAD, SVG for laser, vinyl and design software, PDF for sharing and printing.

Semantic understanding, not pixel tracing

Auto-tracers — Vectorizer.AI, Inkscape Trace Bitmap, Adobe Image Trace — look at pixels and draw paths along contrast boundaries. They work on logos and clean line art. They explicitly fail on photographs.

PhotoCAD does something fundamentally different. Our AI understands what a photo depicts. It knows a labrador's outline is one continuous line, not a hundred jagged edges chasing fur boundaries. It knows a Victorian facade has straight columns and arched windows, not noise from brick texture.

That semantic understanding is the difference. Shadows, tone, texture and noise are interpreted and ignored. What you get back is the geometry of the subject — an accurate, well-judged CAD drawing of what's actually in the photo.

Understands the subject
The AI recognizes what the photo shows — a chair, a dog, a building — and draws the geometry that matters.
Ignores noise
Shadows, gradients, brick texture, fur, JPEG artifacts — interpreted away, not traced.
Clean editable paths
Real vector geometry. Open in any CAD or design program and modify, dimension or extrude like any other drawing.
Cuttable by default
Smooth continuous curves with minimal nodes — loads directly into laser, plasma and CNC controllers without cleanup.

Reference

File specifications

Every conversion produces the same three files from the same vector geometry. Here's what to expect when you open them.

Shared across all formats

Units: Millimeters. Real-world scale.
Scale: Longest edge of the drawing scaled to 300 mm.
Page margin: 5 mm on all sides — keeps the linework clear of the viewport edge.
Stroke: 0.25 mm black, round caps and joins. Stroke is visual only — CAM ignores it.
Fill: None. Open vector outlines.
Layer naming: Single named layer per drawing — easy to identify in CAD.

DXF (.dxf)

Version: AutoCAD R2000 — the modern baseline every laser, CNC and CAD program reads cleanly, including LightBurn's recommended version.
Curves: Encoded as arc-segment polylines — universally supported across CAM controllers and CAD editors.
Y axis: CAD convention (Y-up) so the drawing opens the right way round.
First open: Page bounds set so Zoom Extents centers the drawing on first open.

SVG (.svg)

Dimensions: Real-world millimeters. 1 SVG unit = 1 mm.
Curves: Native Bezier paths — editors keep them fully editable, no flattening.
Layers: Inkscape-compatible layer grouping for clean import.

PDF (.pdf)

Page: Single page. Sized to the drawing plus a 5 mm margin. Orientation auto-picked.
Content: Pure vector line art — scales to any print size without loss.

Tested compatibility

Files have been verified against the following software and hardware. Every modern CAM, CAD, design and laser controller that reads DXF or SVG should accept PhotoCAD output without modification.

CAD & Design

AutoCAD, Fusion 360, Rhino, SketchUp, Vectorworks, Revit, SolidWorks, Adobe Illustrator, CorelDRAW, Inkscape, Affinity Designer.

Laser & Vinyl

LightBurn, Glowforge App, xTool Creative Space, Cricut Design Space, Silhouette Studio, LaserGRBL, Roland CutStudio, Sure Cuts A Lot.

CNC & Plasma

VCarve Pro, Aspire, Carbide Create, Easel, Mach 4, LinuxCNC, SheetCam, ProNest, CamBam, PlasmaCAM.

Need a tool that's not listed? Email us and we'll verify it.

Frequently asked questions

Try it on your own photo

Upload an image, see the AI-generated preview before you commit, then download DXF, SVG and PDF.

Convert a photo Browse the gallery