Convert grayscale brightness into a 2.5D surface with a flat base. Choose width and relief depth explicitly; this tool does not import GIS elevation data.
Brighter pixels rise above a flat base. This is a 2.5D relief, not a full 3D reconstruction.
A real conversion made with this tool. Your model will appear here after generation.
Example size: 120.0 × 90.0 × 7.0 mm
Choose an image → Set size → Generate → Download STL
STL coordinates are in mm. No colors or textures. Review walls and orientation in your slicer before printing.
Original inputs, actual generated meshes and downloadable STL files. These are digital previews, not photographs of printed objects.
Black approaches the 2 mm base and white approaches 7 mm total height. These are relative image values, not GIS elevation.
Width (mm): 120 · Resolution: 128 · Base thickness (mm): 2 · Relief depth (mm): 5
120.00 × 90.00 × 7.00 mm · 49,148 triangles · 1 closed shell
A smooth grayscale peak becomes a bounded grid. More triangles can sample the surface more finely but cannot invent elevations.
Width (mm): 120 · Resolution: 128 · Base thickness (mm): 2 · Relief depth (mm): 5
120.00 × 80.00 × 6.96 mm · 43,516 triangles · 1 closed shell
A grayscale heightmap deliberately encodes surface height as brightness. This page uses the Relief engine: a black value approaches the flat base, while a white value approaches the base plus the chosen relief depth. Width and depth are entered in millimeters. Geographic coordinates, DEM metadata, elevation units and real-world terrain scale are not imported; a landscape-looking image alone cannot supply those facts.
The ramp example makes this relationship measurable. With a 2 mm base and 5 mm relief depth, the surface approaches a 2 to 7 mm range. The radial example instead produces a peak whose sampled source may not reach perfect white at every grid resolution. The parameter experiment changes only the ramp depth, so you can distinguish a height-range change from a different image, aspect ratio or footprint.
Raster resolution determines how finely the source is sampled into the mesh. In the original engine, a rectangular image uses a proportional bounded grid, so the triangle count also depends on aspect ratio. Higher resolution adds data and STL size; it is not a vertical accuracy guarantee. Start from a lossless grayscale source when the values themselves matter. Contrast, smoothing and inversion intentionally alter those values, so use them deliberately rather than treating them as cosmetic preview filters.
| Control | Starting point | What to inspect |
|---|---|---|
| Width / base | 120 mm / 2 mm | Sets XY size and the minimum solid support independently. The source aspect ratio supplies the other planar dimension. |
| Relief depth | 5 mm | Adds a relative height range above the base. The same ramp is also generated at 2 mm depth below. |
| Contrast / smoothing | 1 / 0 | Neutral contrast and no smoothing preserve the intended mapping more directly. Increasing either changes sampled height relationships. |
| Resolution | 256 initially; examples 128 | Increase only if the source has detail worth sampling and your device has the memory. Doubling grid density can increase surface triangles substantially. |
Changing relief depth from 5 to 2 mm reduces the height range above the unchanged 2 mm base. Both files use the same input and recorded settings; the single change is shown below.
| Recorded STL result | Baseline | One parameter changed |
|---|---|---|
| Bounds | 120.00 × 90.00 × 7.00 mm | 120.00 × 90.00 × 4.00 mm |
| Triangles / shells | 49,148 / 1 | 49,148 / 1 |
| Volume | 48600.0 mm³ | 32400.0 mm³ |
| Download for inspection | Baseline STL | Comparison STL |
Closed geometry and dimensions were checked after download. These measurements describe these files; they are not a printability guarantee.
STL stores triangles without units, colors or textures. This tool writes coordinates intended as millimeters. Closed-mesh validation checks finite coordinates, oriented shared edges and positive volume; it does not certify your printer, material or every thin feature.
JavaScript and Web Workers are needed for conversion. WebGL enables the interactive preview; if preview initialization fails, an already validated STL can still be downloaded.
The V1 baseline fixture relief-gradient produced actual layered G-code in both slicers. Across the four modes, eight baseline STL files were checked in Bambu Studio and PrusaSlicer. Their recorded files and hashes are retained.
These are baseline slicing checks, not tests of every gallery parameter variation or physical print. Open your own downloaded model, check dimensions and thin features, and choose settings for your printer.
Recorded baseline results
No. It reads image brightness, not GIS coordinates, DEM metadata or real-world elevation units. Width and depth are chosen explicitly in mm.
No. The image is sampled to a bounded grid. More resolution adds triangles; source detail finer than the grid may be lost.
The sampled image may not contain a value of exactly one at the peak. Resampling and source brightness determine the actual highest point. Inspect the measured STL Z bounds instead of assuming every image reaches the configured upper limit.
This browser workflow normalizes decoded image pixels for a bounded height surface. It is not a promise of lossless 16-bit DEM elevation import or geographic accuracy. Use a specialized pipeline when those requirements matter.
Files stay on your device. Desktop: up to 20 MB / 8 million pixels. Touch phones: 10 MB / 4 million pixels. SVG: 1 MB with bounded complexity. PNG, JPG, WebP and prepared SVG only. Use original or permitted artwork; model conversion does not grant image or trademark rights.
