PNG to STL FAQ
How heightmap conversion actually works, what it can and cannot do, and the print settings that make the difference.
What does a PNG to STL converter actually do to my image?
It reads your image as a heightmap. Every pixel's brightness becomes a height: bright pixels rise toward your relief height setting, dark pixels stay near the base, and everything in between scales proportionally. Colour is discarded for the geometry — a red shirt and a green wall of the same brightness come out at exactly the same height. That is why high-contrast images convert best, and why a flat grey photo produces a flat grey slab.
Can this turn a photo into a real 3D object?
No, and no tool can. A photo is a single projection — it contains no depth information about the back or sides of an object, so any tool claiming to reconstruct a full 3D shape is guessing. This tool deliberately does not guess: it produces a raised relief on a flat base, which is exactly what name plates, coins, art panels, keychains and lithophanes need. If you want a freestanding figurine of a real object, you need photogrammetry (many photos from different angles) or AI reconstruction tools — a different category of software with a different output.
Why is my logo coming out inverted or recessed?
Most logos are dark on a light background, which means the background becomes the tallest part and your logo sinks into it. Turn on Invert and the relationship flips, so the dark logo rises and the light background drops away. The same applies to any dark-subject-on-light-image case. It is the single most common fix for logo conversions.
What does the Invert control do?
It reverses the brightness-to-height mapping: black becomes tall and white becomes short, or the other way round. Use it when your subject is darker than its background and you want the subject raised, or when you want a negative relief such as a mould or stamp. It flips the geometry only — it does not affect colours in your 3MF export.
What does Smoothing fix, and how much should I use?
Raw heightmap conversion turns every sharp brightness jump into a cliff, and single-pixel noise becomes a spike that your slicer has to support. Smoothing applies a blur pass to the height field before the mesh is built, removing spikes and stair-stepping. Use low smoothing for logos and text so the edges stay crisp, and a moderate amount for portraits, pets and landscapes where organic curves look far better. Too much smoothing melts fine detail like eyelashes or thin lettering.
Are the exported meshes watertight and printable?
Yes. The mesh is generated with a marching-squares surface-net algorithm that produces a closed, manifold solid — every edge is shared by exactly two faces, with no holes, no inverted normals and no non-manifold edges. You can import it straight into Cura, PrusaSlicer, Bambu Studio, OrcaSlicer or Chitubox without running a mesh repair step. The outer edge is also rounded by a true signed-distance bevel, so prints do not start with a razor-sharp fragile cliff.
What is the difference between a relief and a lithophane?
They are the same heightmap idea used for two different viewing conditions. A relief is read by ordinary room light — you see it by its shadows, like a coin or a medal, and it wants a solid base plate. A lithophane is read by light coming from behind: thin regions glow and thick regions go dark, so the image only appears when backlit. Because light does not fade linearly through plastic but falls off roughly exponentially, a lithophane needs a specific narrow thickness range (typically 0.6–0.8 mm at the brightest point up to 3.0–3.2 mm at the darkest) rather than the wider, flatter range a relief wants. A good rule: relief for shelf display and signs, lithophane for lamps, night lights and photo frames.
What print settings should I use?
For a relief on FDM: 0.2 mm layer height, 15–20% infill, print face-up, no supports needed. For a lithophane: drop the layer height to 0.08–0.12 mm, print vertically on edge so layer lines do not cross the image, use 100% infill or a high wall count so no internal grid shows through, use white or natural PLA, slow down to roughly 30–40 mm/s with the cooling fan at full, and place the light source 20–50 mm behind the print behind a diffuser. Coloured, silk, wood, metal-filled and ABS filaments block light and ruin the effect.
Will the background become part of the model?
It depends. If your PNG already has a transparent background, the transparent areas are treated as empty and the subject is extracted automatically. For opaque images, background detection is on by default: the tool samples the colour around the image border, grows that region inward, and treats anything it cannot reach as the subject. If your image has a busy or graduated background that confuses it, switch detection off to keep the full rectangle, or use a transparent PNG instead.
Which image formats are supported?
PNG and JPG, plus other browser-decodable bitmaps such as WebP. PNG is recommended whenever a transparent background matters. HEIC files from an iPhone need exporting to JPEG first, because browsers cannot decode HEIC. For the cleanest geometry, use an image at least 1000 px on the long edge with strong tonal contrast.
How long does generation take?
Parsing, meshing, preview and export all run locally in your browser, so a typical image finishes in under about 3 seconds with no server round trip and no upload queue. Large images are automatically downsampled to a sensible mesh density to keep generation fast and the resulting file size reasonable.
What's the difference between STL and 3MF?
STL is the universal format: every slicer, CNC package and 3D editor reads it, but it stores only triangle geometry — no colour, no units, no material. 3MF is a modern XML-based format that can carry vertex colours, so it is the right choice for colour-capable printers and for re-editing in Blender or ZBrush. Export 3MF when you want the surface colours of your original image preserved; export STL when you just need a reliable, universally readable solid.
Can I export a coloured model?
Yes. Turn on Colour mapping and the 3MF carries vertex colours taken from your source image, which is what multi-material and colour-capable printers need. STL cannot store colour, so the coloured path is 3MF-only — this is one of the few heightmap converters that offers it.
Is my image uploaded anywhere?
No. All parsing, meshing and export happen locally in your browser, so your image never leaves your machine. There is no upload queue, no retention window and no server-side processing — the tool works even with the network disconnected after the page loads.
Do I need to sign up or pay?
No account and no login. Everything runs in your browser, exports carry no watermark, and there is no resolution cap or export limit.