ASCII

An ASCII STL file begins with the line:

solid name

where name is an optional string (though if name is omitted there must still be a space after solid, for compatibility with some software). The remainder of the line is ignored and is sometimes used to store metadata (e.g., filename, author, modification date, etc).^[14] The file continues with any number of triangles, each represented as follows:^[15]

facet normal ni nj nk
    outer loop
        vertex v1x v1y v1z
        vertex v2x v2y v2z
        vertex v3x v3y v3z
    endloop
endfacet

where each n or v is a floating-point number in sign-mantissa-e-sign-exponent format, e.g., 2.648000e-002. The file concludes with:

endsolid name

An example ASCII STL of a sphericon

The structure of the format suggests that other possibilities exist (e.g., facets with more than one loop, or loops with more than three vertices). In practice, however, all facets are simple triangles.

Whitespace (spaces, tabs, newlines) may be used anywhere in the file except within numbers or words. The spaces between facet and normal and between outer and loop are required.^[9]

Binary

Because ASCII STL files can be very large, a binary version of STL exists. A binary STL file has an 80-character header which is generally ignored, but should never begin with the ASCII representation of the string solid, as that may lead some software to confuse it with an ASCII STL file. Following the header is a 4-byte little-endian unsigned integer indicating the number of triangular facets in the file. Following that is data describing each triangle in turn. The file simply ends after the last triangle.

Each triangle is described by twelve 32-bit floating-point numbers: three for the normal and then three for the X/Y/Z coordinate of each vertex – just as with the ASCII version of STL. After these follows a 2-byte ("short") unsigned integer that is the "attribute byte count" – in the standard format, this should be zero because most software does not understand anything else.^[9]

Floating-point numbers are represented as IEEE floating-point numbers and are assumed to be little-endian, although this is not stated in documentation.

UINT8[80]    – Header                 -     80 bytes
UINT32       – Number of triangles    -      4 bytes

foreach triangle                      - 50 bytes:
    REAL32[3] – Normal vector             - 12 bytes
    REAL32[3] – Vertex 1                  - 12 bytes
    REAL32[3] – Vertex 2                  - 12 bytes
    REAL32[3] – Vertex 3                  - 12 bytes
    UINT16    – Attribute byte count      -  2 bytes
end

There are at least two non-standard variations on the binary STL format for adding color information:

• The VisCAM and SolidView software packages use the two "attribute byte count" bytes at the end of every triangle to store a 15-bit RGB color:
  • bits 0–4 are the intensity level for blue (0–31),
  • bits 5–9 are the intensity level for green (0–31),
  • bits 10–14 are the intensity level for red (0–31),
  • bit 15 is 1 if the color is valid, or 0 if the color is not valid (as with normal STL files).
• The Materialise Magics software uses the 80-byte header at the top of the file to represent the overall color of the entire part. If color is used, then somewhere in the header should be the ASCII string COLOR= followed by four bytes representing red, green, blue and alpha channel (transparency) in the range 0–255. This is the color of the entire object, unless overridden at each facet. Magics also recognizes a material description; a more detailed surface characteristic. Just after COLOR=RGBA specification should be another ASCII string ,MATERIAL= followed by three colors (3×4 bytes): first is a color of diffuse reflection, second is a color of specular highlight, and third is an ambient light. Material settings are preferred over color. The per-facet color is represented in the two "attribute byte count" bytes as follows:
  • bits 0–4 are the intensity level for red (0–31),
  • bits 5–9 are the intensity level for green (0–31),
  • bits 10–14 are the intensity level for blue (0–31),
  • bit 15 is 0 if this facet has its own unique color, or 1 if the per-object color is to be used.

The red/green/blue ordering within those two bytes is reversed in these two approaches – so while these formats could easily have been compatible, the reversal of the order of the colors means that they are not – and worse still, a generic STL file reader cannot automatically distinguish between them. There is also no way to have facets be selectively transparent because there is no per-facet alpha value – although in the context of current rapid prototyping machinery, this is not important.

The coordinate units are not specified in the STL standard. One idea is to add an ASCII string UNITS=xx somewhere in the header, where xx are two bytes representing the units used, when writing this file. Possible options are:

Facet normal

In both ASCII and binary versions of STL, the facet normal should be a unit vector pointing outwards from the solid object.^[16] In most software this may be set to (0,0,0), and the software will automatically calculate a normal based on the order of the triangle vertices using the "right-hand rule", i.e. the vertices are listed in counter-clock-wise order from outside. Some STL loaders (e.g. the STL plugin for Art of Illusion) check that the normal in the file agrees with the normal they calculate using the right-hand rule and warn the user when it does not. Other software may ignore the facet normal entirely and use only the right-hand rule. Although it is rare to specify a normal that cannot be calculated using the right-hand rule, in order to be entirely portable, a file should both provide the facet normal and order the vertices appropriately. A notable exception is SolidWorks, which uses the normal for shading effects.