Preparing an image for the generator

Output quality depends on input quality. How to prepare an image at home — without Inkscape or Photoshop, optionally with AI tools.

What works best

The generator extracts a single closed outline from your image — the shape the cutter will cut out. The cutter itself is the wall around that outline. It works best when the image meets three conditions:

Silhouette — a dark shape on a light background. Detail inside the shape does not transfer to the cutter, only the outer outline.
High contrast — a sharp edge between shape and background. A pale pencil sketch on white paper works worse than a black marker.
One connected shape — a cutter is one closed curve. If the image has separate pieces (detached limbs, two figures side by side), the generator notices and offers to crop to a single motif. Decline, and it uses the largest piece and discards the rest — but with touching figures the “largest piece” can be both of them at once, which is why cropping is worth it.
A reasonable size is enough — around 1000–2000 px on the longer side. Larger images are downscaled during processing anyway; smaller ones may lose detail.

Your own drawing

The simplest path — draw the shape on paper and take a phone photo. A digital drawing works just as well — in a drawing app, on a tablet, or in a simple paint program. Either way the goal is the same: a dark closed line on a light background. A few tips that lift quality significantly:

Use a dark marker or a thicker pen, not a pencil. A black line on white paper is ideal.
Plain white paper without lines or dots. Lines under the drawing can confuse the generator and end up attached to the outline.
Photograph from directly above, parallel to the paper. Avoid the shadow of your hand — light should come from behind or the side, not from the front.
The line must be closed — if a stretch is missing somewhere, the generator will break the outline and the shape will be incomplete.

Photo of a real object

Works well for objects with a clear silhouette — a leaf, a key, a simple toy. Less well for things with internal detail you'll miss (a logo, eyes). Tips:

Put the object on white paper or a single-colour surface clearly distinct from the object.
Even diffused light — a window behind you, not direct sun (it casts sharp shadows that the generator adds to the outline).
Shoot straight down. A tilted angle distorts proportions — the cutter ends up shaped differently from the source.
Crop so the object fills as much of the frame as possible — extra background only slows processing and makes the background harder to detect. You do not have to do it beforehand: the generator has a “Crop the image…” button that crops the uploaded image in place, with the result visible straight away.

An existing image from the internet

If you don't want to draw or shoot, search for “silhouette” + what you need (e.g. “heart silhouette”, “deer silhouette”). Good sources:

Wikimedia Commons (commons.wikimedia.org) has a large “Silhouettes” category with thousands of freely usable images. Openclipart.org is free, no sign-up needed. There are also paid stock-image banks with wider selection.
Children's colouring books and stencil pages are often black lines on white — exactly what the generator needs. Photograph the page, or use freely available PDF colouring books.

Watch copyright. A cutter generated here is yours — under our terms and conditions you may use and sell it without restriction, including the STL file itself, printed cutters, and anything you make with them (ceramics, jewellery, decorations and the like). But mind the source image — our licence covers only the generated shape, not third-party rights in the artwork. If the shape is a recognisable film character or a protected logo, selling products in that shape may still infringe someone else's copyright, even if the cutter itself is “yours”.

AI tools (ChatGPT, Claude, Gemini and others)

An AI tool with image generation can do two useful things — create a clean custom illustration from a description, or take your existing photo and re-paint it in the same style. Most AI tools offer a limited number of free generations; regular use needs a paid subscription. Here is a prompt for creating a new motif from a description — copy it into your AI and fill in what you need:

Generate a black-and-white illustration of [DESCRIPTION — e.g. “a cat seen from the side”] in a colouring-book or clipart style: one clean closed outer outline (thick black), with only thin simple lines inside for a sense of detail. White background, no shadows or gradients. The outer outline must be one connected shape with no separate pieces — that becomes the cutter; the inner lines only show what the shape represents so you can later see where to add colour or decoration.

The generator helps with re-drawing your own photo: for an uploaded image it offers a prepared prompt and an “Open ChatGPT” button — the redraw itself then happens in your ChatGPT, and you upload the resulting image back into the generator. If you use a different AI tool, attach the photo in the chat with this prompt: “Convert this image into a black-and-white illustration in a colouring-book style: one clean closed outer outline (thick black), with only thin simple lines inside for a sense of detail, on a white background, no shadows or gradients. The outer outline must be one connected shape with no separate pieces.” Useful for an animal photo or a child's drawing you want to clean up.

Important: we use AI images only as a clean raster (PNG/JPG) — not as a vector the AI would “draw”. Generating an illustration as an image works well; asking for a vector output usually ends weirdly. If you have multiple attempts, pick the one that best matches “clean closed outer outline, no separate pieces”.

What will not transfer to the cutter

Some things the generator won't do — not because it can't, but because the cutter format itself doesn't allow it:

Multi-piece shapes (detached limbs, two figures side by side) — a cutter is always one piece. If you want each part separately, upload them as separate images and you'll get a set of cutters instead of one combined design.
Holes inside (e.g. the digits 0 or 8, a face with cut-out eyes) — sometimes you want them, sometimes you don't. The generator currently doesn't model holes — the outline comes out solid. We plan to support this later; in the meantime get in touch if you need holes now.
Very thin or pointed protrusions (a whisker, tiny arrows). The cutter profile itself is strong enough; the problem is with the cut-out shape — thin sticking-out parts break off easily when you lift the cutter or handle the cut piece. The generator won't discard them, but the final shape may not last.

If you have a more complex design (a precise motif with internal lines, combined shapes, embossed relief, holes inside), get in touch — we do custom work by hand for those. Embossed relief and holes are both on the roadmap for the generator itself.

Contrast not ideal? The generator will still try. If the first result is off, a “Different outline…” button appears under the preview — it offers several outline variants computed at different brightness thresholds, and you pick the best one. Clicking the first card returns you to the original. A well-prepared high-contrast image is still more reliable than tuning after the fact.