What "image to STL" actually means
It is worth being straight about this, because a lot of tools are not. A photograph contains no depth information. Nothing can read a picture of a vase and give you a vase. What a converter can do — and what this one does — is treat the brightness of each pixel as a height, and build a surface from it. The result is a relief: a flat-backed panel with the image raised on its face.
That covers most of what people actually want an image converted for: name plaques, coasters, signs, keychains, wall art, pet portraits and lithophane lamps. It does not cover turning a product photo into a printable copy of the product. If that is what you need, the honest answer is photogrammetry — dozens of photos from different angles — or modelling it yourself.
Choose the right mode
| Relief | Lithophane | |
|---|---|---|
| Viewed | From the front, under normal light | With a light directly behind it |
| Read by | Shadow and shape | How much light gets through |
| Typical depth | 1.5 mm base, 2.5 mm of relief | 0.8 mm to 3 mm total |
| Bright pixels | Raised | Thin |
| Best for | Plaques, coasters, signs, logos | Photo frames, night lights, lamp panels |
| Print flat? | Yes, face up | No — vertically, on its edge |
The mode selector at the top of the tool switches between them, and the preview changes with it: relief mode shades the surface as it would look on a shelf, lithophane mode simulates it lit from behind.
Why the preview matters more than it sounds
A relief that looks crisp on screen can print as mush, because the printer cannot resolve detail finer than roughly two nozzle widths, and because the height is quantised to whole layers. A lithophane is worse still: it looks like a blank white tile until it is lit, so a grey 3D render tells you almost nothing about whether it worked.
Both previews here are generated from the actual export geometry, after the thickness has been rounded to your layer height. What you see is what the slicer will get — which is the difference between one print and three.
Getting a good result
- Crop tight. Every millimetre spent on background is a millimetre not spent on the face.
- Raise the contrast before uploading. The tool has a brightness trim, but real editing beats it.
- Do not chase resolution. Below about 0.25 mm per sample you are adding triangles, not detail.
- Watch the grey-level count. Fewer than about twelve steps and smooth gradients will band.
- Print slowly. 30–40 mm/s keeps the fine surface detail from being smeared.
Once you have the file, check it in the STL viewer — it will confirm the mesh is watertight and give you exact dimensions — and price the print with the cost calculator. If you specifically want a backlit photo panel, the lithophane maker is the same engine with the settings and guidance tuned for it.
Frequently asked questions
Can you really convert a PNG or JPG into an STL?
Yes, with one honest caveat: a photo is a flat picture, so it cannot become a true 3D object with sides and a back. What it can become is a heightmap — a surface whose height at every point is driven by the brightness of the pixel there. That is exactly what a relief plaque or a lithophane is, and it is what every "image to STL" tool actually produces.
What is the difference between relief mode and lithophane mode?
A relief is viewed from the front, like a coin or a plaque: bright parts of the image stand proud of a solid base, and you read it by the shadows. A lithophane is viewed with a light behind it: the whole panel is thin and translucent, dark parts are thicker and block more light. They need completely different thickness maths, which is why this tool treats them as two modes rather than one slider.
Which type of image works best?
High contrast, a clear subject, and a simple background. Portraits and pets work well. Busy scenes, text and fine patterns do not, because the printer cannot resolve detail finer than about two nozzle widths. Convert to greyscale in your head before you upload: colour is discarded, so two colours of the same brightness become the same height.
Do I need to remove the background first?
It helps a lot. A cluttered background competes with your subject for the limited height range available. Cropping tight and lightening or darkening the background before uploading gives noticeably better results than any setting in this tool can.
How big will the STL file be?
It depends almost entirely on the detail setting, which controls how many samples the surface is built from. A 100 mm panel at 0.3 mm detail is roughly 170,000 triangles and around 8 MB. Halving the detail value quadruples both. The tool shows the triangle count live and warns you before it gets unwieldy.
Is the STL watertight?
Yes. The mesh is emitted closed — a relief surface, four side walls and a back, with every boundary edge shared by exactly two triangles, so slicers will not complain about holes. You can confirm it yourself by dropping the file into our STL viewer, which runs a manifold check.
Is my image uploaded?
No. The picture is read into a canvas and the mesh is generated on your device. There is no server involved, which is why there is no file size limit and no queue — and it matters, since these are usually personal photos.