Picture to STL Converter
This free picture to STL converter turns any drawing, logo, screenshot or photo (PNG, JPG or WEBP, up to 10 MB) into a printable binary STL in your browser, with no signup and nothing uploaded. Line art is extruded into a solid; photos and paintings become a brightness-to-depth relief, in millimetres on a flat base.
Last updated: 2026-10-22
What counts as a picture here
The word covers a lot of ground: a child's drawing photographed on a table, a company logo exported from a design tool, a hand-drawn sketch scanned at 600 dpi, a meme screenshot, a product photo, a watercolour scan. They all convert, but they need different modes, because they are not the same kind of data — line art is a boundary problem, and a photograph is a brightness problem.
The rule of thumb is simple. If the picture is made of shapes with edges — logos, lettering, icons, black-and-white drawings — extrude it, so everything on one side of a brightness threshold becomes solid geometry. If the picture is made of continuous tone — photos, paintings, renders — use heightmap mode, so brightness becomes depth.
Pictures also arrive with unwanted context. Screenshots carry status bars, browser chrome and grey UI panels. Phone photos of a drawing carry a wooden table and a shadow. Scanned sketches carry paper grain. All of it converts too, since the tool is not clever about intent, which is why cropping is usually the highest-value step.
Scans and photos of flat artwork arrive with two fixable defects. A phone photo of a drawing is almost never square to the page, so the subject is skewed; a scanned sketch is square but may carry dust, staples and paper grain. Correct perspective before converting, or crop tightly enough that a small tilt stops mattering. Conversion cannot straighten what it receives — height comes from brightness, so a skewed rectangle becomes a skewed slab with the tilt baked into the print.
How to convert a picture to STL
Drop the image — up to 10 MB, in any format the browser decodes. Pick extrude to turn dark pixels into solid columns, or heightmap to turn brightness into a relief on a base plate. The preview updates as you change settings, so you can compare a handful of thresholds before committing.
There is no automatic background removal, and that is deliberate: guessing a subject boundary produces worse results than a plain threshold. If the background has to go, remove it before uploading — export the artwork with transparency, or cut it out in an image editor. A transparent PNG is the cleanest input, because transparency is composited onto white and reads as empty.
Output is binary STL in millimetres, centred with its bottom on Z=0. Extrusion produces one column per solid cell, with side walls only where the neighbouring cell is empty, so the model is a closed solid rather than a shell. Heightmap output is a relief fused to a flat base plate of adjustable thickness.
Line art, drawings and screenshots
Pencil and ink drawings are lines, not regions, and that is the tricky part: extruding a line gives you a thin ridge that stands 5 mm proud of the plate and snaps off in handling. Converting a drawing into a printable relief usually works better with heightmap mode, where graphite darkness becomes shallow depth and the whole page stays supported.
Screenshots are the opposite case — crisp, flat and high contrast, ideal for extrude. Crop away everything that is not the subject first, and be aware that UI text at screen resolution becomes legible but fragile geometry; lettering needs roughly 4 mm of cap height to print reliably in a 100 mm model.
Drawings that touch only at corners are a real limitation. Because extrusion works per cell, two shapes that meet diagonally share a single edge rather than a face, and some slicers flag that as non-manifold. Thickening the artwork, or nudging a line by a pixel, resolves it before conversion.
Screenshot text and pixel-art edges are already aligned to the picture grid, which makes them the friendliest extrude inputs, but watch the scale: a 12-pixel-tall caption in a 1080p screenshot is roughly 3 mm across a 100 mm model, under the reliable limit for a 0.4 mm nozzle. Render the text larger at the source or upscale the artwork before uploading instead of relying on the converter to rescue it.
Picture to STL problems and fixes
If the result is the inverse of what you expected — a solid plate with a hole where your subject should be — flip the invert switch. Extrude treats dark pixels as solid by default, so a light subject on a light background disappears and a dark background fills the build plate.
If edges are ragged, the source has antialiasing or compression noise and the threshold is landing in the soft zone. One smoothing pass cleans up noise before the cut; raising resolution keeps edges sharper; exporting without a drop shadow removes a halo that is otherwise baked into the geometry.
If thin elements disappear in the slicer, they were smaller than a printable wall for your nozzle — extrusion cells are about 0.4 mm at 100 mm width, and a 0.4 mm nozzle cannot lay down anything thinner than about 0.4 mm reliably. Scale the model up, thicken the artwork, or accept the loss.
Picture to STL FAQ
What kinds of pictures can I convert?
Any image the browser can decode: PNG, JPG or WEBP, whether it is a drawing, a logo, a screenshot, a painting, a scanned sketch or a photograph. Line art and logos extrude well; photos and paintings work better in heightmap mode. The only real requirement is that the file is under 10 MB.
Is my picture uploaded?
No. Everything happens inside your browser: the image is decoded, sampled and turned into geometry on your own device, and the STL is created locally. No account, no queue, no watermark, and nothing to clean up afterwards.
Can it remove the background automatically?
No, and it does not try. Automatic subject detection fails unpredictably on drawings and logos, so the tool relies on a brightness threshold you control instead. If you need the background gone, export with transparency or cut the subject out in an editor first — then the transparent area reads as empty.
Should I use extrude or heightmap?
Extrude for anything with shapes and edges — logos, lettering, icons, silhouettes. Heightmap for continuous tone — photographs, paintings, pencil shading. A quick test: if you could describe the image as a few black shapes on a plain background, extrude it; if it needs greys to make sense, use heightmap.
How do I turn a drawing into a printable model?
Photograph or scan it flat-on with even light, crop to the drawing, then convert. For a delicate pencil drawing, heightmap mode keeps the lines supported by the base instead of turning each stroke into a fragile ridge. For bold marker or ink shapes, extrude converts them into solid plates that handle well.
Why do parts of my picture vanish?
Two common reasons. Either those areas fell on the wrong side of the threshold and were treated as background, or they are thinner than one grid cell — about 0.4 mm at a 100 mm width — so they cannot be represented. Adjust the threshold, thicken the artwork, or scale the model up.
Is the output ready for a slicer?
Yes. It is binary STL in millimetres, centred on the origin with a flat bottom at Z=0, which Cura, PrusaSlicer, OrcaSlicer and Bambu Studio all accept. Check the scale on import and confirm that thin details are at least two nozzle widths wide before you print.
Is this picture to STL converter free?
Yes — free, with no account, no watermark and no limit on how many pictures you convert. The only constraints are 10 MB per image and the 256-cell grid. The paid AI mode is a separate option at 20 credits per generation, and logos, drawings and reliefs do not need it.