Image to STL Converter

Turn a picture into an STL you can open in a slicer and 3D print. Upload the image, convert, and download the STL. No account, no watermark, and the upload is deleted after processing.

This conversion surprises people more than most, so it is worth being precise about what comes out the other side.

What converting an image to STL actually does

A photograph is a grid of pixels, each with a brightness. An STL is a three-dimensional surface made of triangles. There is no depth information anywhere in a photograph, so the conversion cannot extract one — it has to invent a rule and apply it.

The rule is brightness. Each pixel's lightness is read as a height: light areas are pushed up, dark areas stay low. Do that across the whole image and you get a landscape whose contours follow the tones of the picture.

The result is a relief — a raised surface on a flat backing, like the face on a coin or a carved plaque. It is not a free-standing model. A photograph of someone's face produces a face-shaped relief, not a bust you can walk around, because the image only ever described one side of them and nothing in the file says what the back of their head looks like.

Understanding that one point prevents nearly every disappointment with this tool.

Relief or lithophane?

Two quite different objects come out of the same conversion, and knowing which you want changes how you prepare the image.

A relief is printed reasonably thick and viewed directly in normal light. Light parts of the picture stand proud, dark parts sit back, and shadows across the raised surface do the work of showing the image.

A lithophane is printed thin in white or light filament and lit from behind. Thin sections let light through and read as bright; thick sections block it and read as dark. That is the inverse of a relief, so a lithophane needs the image inverted — a negative — before converting, or the finished piece reads back to front.

There is no inversion control on this page, so invert the picture in any image editor first if a lithophane is what you are after.

Which images work well

Well: strong contrast, a clear subject, an uncluttered background. Portraits lit from one side, logos, silhouettes, line drawings, engravings and high-contrast landscapes all convert into something recognisable.

Poorly: busy backgrounds, flat low-contrast images, and pictures with deep shadows. Shadow is the awkward one, because the conversion cannot tell a dark shadow from a genuinely recessed surface — a shadow under a chin becomes a trench. Heavily textured backgrounds turn into noise across the whole plate.

Text is usually a disappointment too. Letters become soft mounds rather than crisp shapes. If you want sharp lettering, start from vector artwork and use SVG to STL, which extrudes outlines exactly rather than interpreting tones.

Prepare the image before converting

Ten minutes in an image editor makes more difference here than anything else, and the steps are simple.

Convert to greyscale. Only brightness is read, so colour contributes nothing and can mislead you about contrast while you work — a strong red and a strong green can be nearly identical in brightness.

Raise the contrast. Separating the tones is what separates the heights. A flat image produces a flat model.

Simplify the background. Fill it with a solid tone, or crop it out. Anything left in the background gets height of its own and competes with the subject.

Blur very slightly. Counter-intuitive, but sensor noise and compression speckle become thousands of tiny bumps and a much heavier file. A blur of half a pixel removes them while leaving real detail alone.

Crop tight. Empty margin becomes flat plate, which costs print time and material for nothing.

How to convert an image to STL

1. Upload your image, or drag it onto the upload area.
2. Convert.
3. Download the STL.

There is nothing to configure. Most people arrive looking for a depth setting, and it is worth saying plainly that there isn't one — depth, size and orientation are all set afterwards, which is where they are easiest to judge anyway.

Setting size, depth and orientation afterwards

STL files carry no units. The format stores bare numbers and each program decides what they mean; nearly all slicers assume millimetres.

Open the STL in Cura, PrusaSlicer, Bambu Studio or Orca and check the dimensions before anything else. Scale X and Y together to set the physical size. To change only how pronounced the relief is, turn off uniform scaling and adjust Z alone — that is the depth control this page does not have, and it is better placed there because you can see the result against the plate.

Orientation matters more than people expect. A relief usually prints well lying flat. A lithophane prints far better standing vertically, because the detail then runs across layer lines rather than along them, and layer height controls the resolution of the image itself.

Why the result can look noisy or banded

Compression artefacts. JPEG stores an approximation, and around sharp edges it leaves faint blocky halos. Those halos have brightness, so they get height, and they print as a rippled surface. Starting from a lossless file avoids it entirely — see PNG to STL.

Only 256 levels. An ordinary 8-bit greyscale image has 256 possible brightnesses, so a smooth gradient becomes 256 discrete steps. On a large smooth area that shows as faint terracing. Slight blurring softens it; so does keeping the relief shallow.

Too much detail. Every pixel can become geometry, so a large photograph makes a very heavy mesh. Reducing the image to around a thousand pixels on its long edge before converting usually loses nothing visible and makes the file manageable.

Images and STL side by side

Raster images — JPEG, PNG, BMP, TIFF — store a fixed grid of pixels. Enlarging one cannot add detail that was never recorded, which is why a small source image gives a soft, blocky relief no matter how large you print it.

STL (stereolithography) was created by 3D Systems in 1987. It stores a surface as a list of triangles and nothing else: no colour, no materials, no units. That simplicity is why almost every 3D printing program reads it, and why the file cannot tell your slicer how big the object is meant to be.

FAQs