3D Modeling: Software, Techniques and File Basics

3D modeling is the process of building a mathematical description of an object’s surface in three dimensions, most often as a polygon mesh. A Wikipedia survey of the field notes that polygonal meshes are “by far the most common representation.”

What 3D modeling produces and how models are represented

A Wikipedia entry on the field lists four popular ways to represent a model, and the choice among them shapes every later step: which software opens the file, how the surface can be edited, and whether the result suits a game engine, a machine shop or a 3D printer. A finished 3d model is therefore never just a picture of an object. It is a data structure with rules about points, surfaces, scale and appearance.

Two of the four representations are easy to tell apart. Parametric modeling is a feature-based structure that relies on parent-child relationships between features, which is why mechanical CAD systems can change one dimension and let the rest of a part rebuild itself. Polygonal modeling connects points in space, called vertices, into surfaces, and it is the approach that dominates entertainment work. The same source states that the vast majority of 3D models today are built as textured polygonal models because they are flexible and because computers can render them so quickly.

The craft side of the work is closer to the plastic arts than to drafting. The manual modeling process of preparing geometric data for 3D computer graphics is similar to sculpting, according to the same Wikipedia article, and that comparison explains why artists talk about pushing, pulling and carving a form rather than calculating it. Engineers, by contrast, describe parts by constraints and dimensions, and treat the geometry as the outcome of a specification.

Three questions help sort any project before software is chosen:

What is the model for? A film asset needs a clean surface and textures. A bracket for a machine needs exact dimensions and tolerances.
Who consumes the file? A slicer for a 3D printer, a real-time game engine, a CNC toolpath generator and a web viewer each accept different formats.
How will it be edited later? A parametric history can be revised by changing a value, whereas a sculpted mesh is edited by hand.

The rest of this hub follows those questions: representation first, then software, techniques, file formats and the practical steps that carry a model into a printer or a headset.

Polygon meshes, tessellation and why triangles dominate

A polygon mesh, in the Wikipedia description, is a collection of vertices, edges and faces that define the shape of a polyhedral object’s surface. If the mesh’s edges are rendered instead of the faces, the model becomes a wireframe model, which is the familiar see-through look of early computer graphics and of the viewport in modern editors.

Faces can be triangles, quadrilaterals (quads) or larger polygons called n-gons. Artists tend to prefer quads because they flow along the form of a limb or a hull and subdivide predictably. Hardware prefers triangles. The 3D modeling article explains that meshes of triangles are popular because they have proven to be easy to rasterize: the surface described by each triangle is planar, so the projection is always convex. Most renderers can also convert polygons to triangles on the fly, which, per the Polygon mesh article, makes it unnecessary to store a mesh in a triangulated form.

The step that turns an abstract description into triangles has a name. The process of transforming representations of objects, such as the middle point coordinate of a sphere and a point on its circumference, into a polygon representation of a sphere is called tessellation. The same idea gives the STL format its alternate name, “Standard Tessellation Language,” which the Library of Congress records in its description of the format.

Tessellation involves a trade. A coarse mesh is light and quick to display but shows flat facets on curved surfaces. A dense mesh looks smooth and costs memory, file size and print or render time. Subdivision surfaces, which the Wikipedia article ranks as the second most common representation after polygon meshes, let an artist model a low-resolution cage and have the software smooth it, so the working file stays small while the output looks round.

Not every rendering technique uses polygons. The Wikipedia article points out that polygon representations are not used in all rendering techniques, and in those cases the tessellation step is not included in the transition from abstract representation to rendered scene. Ray-traced renderers can intersect a ray with a mathematically exact sphere, for example, without ever producing a triangle. For production pipelines in games, film and 3D printing, though, the mesh is the common currency.

Software for 3D modeling: Blender, CAD and the free options

Blender began as an in-house tool at the Dutch animation studio NeoGeo and was officially launched on 2 January 1994, with version 1.00 following in January 1995, according to Wikipedia’s history of the program. Its primary author was Ton Roosendaal, the company co-owner and developer. That long lineage matters because it explains why blender bundles modeling, sculpting, texturing, animation and rendering in one application.

The licensing story is unusually well documented. Roosendaal founded Not a Number Technologies in June 1998 to develop Blender further, and the company went bankrupt in 2002. In July 2002 he started the “Free Blender” campaign, which aimed to open-source Blender for a one-time payment of €100,000. On September 7, 2002, it was announced that enough funds had been collected and that the source code would be released. Today the Blender Foundation’s own About page describes a community project, primarily funded by donations, whose mission is to put “the world’s best 3D CG technology in the hands of artists as free/open source software,” and which operates under the GNU General Public License (GPL).

Blender also illustrates how a polygon modeler handles rendering. Since version 2.80, released in 2019, it has shipped with three render engines: EEVEE, Workbench and Cycles. Choosing between a real-time engine and a path-tracing one is a workflow decision rather than a modeling one, but it affects how much material detail a model needs.

Mesh and sculpting tools versus CAD

The other major software family is computer-aided design. cad software works from dimensioned sketches, constraints and a feature history, so a hole can be moved by editing a number. This is the parametric representation described earlier, and it is the standard for mechanical parts, enclosures and anything that must fit another part. Mesh modelers such as Blender do not preserve that kind of design intent by default, which is why a bolt pattern drawn by eye in a polygon tool is harder to revise than one defined by constraints in CAD.

Budget is the usual second question. A guide to free cad software covers the no-cost parametric options, and Blender covers the free polygon side. The editorial position of this site is that neither category is automatically the better choice. A product earns a mention here on its merits, and no affiliate links, sponsored placements or paid reviews are involved.

Matching tool to task

Characters, creatures and props for games or film: polygon and sculpting tools, with UV unwrapping and texturing.
Mechanical parts, jigs and enclosures: parametric CAD, with exports to STEP or STL for downstream use.
Architecture and product visualization: either, depending on whether the geometry came from a CAD package or was built by eye.
Web and AR delivery: any modeler that can export glTF, usually after a polygon reduction pass.

Core techniques: box modeling, sculpting, parametric design and scanning

The 3D modeling article identifies four popular representations, and each one maps to a working technique. Polygon work often starts from a primitive such as a cube, sphere or cylinder, and the artist extrudes faces, inserts edge loops and moves vertices until the form appears. This is usually called box modeling, and it builds clean topology that deforms well when animated.

Sculpting treats the surface like clay. Because the manual process is similar to the plastic arts, digital sculpting suits organic shapes such as faces, rock formations and creature anatomy. The result is a very dense mesh, so sculpted models usually go through retopology, a pass that rebuilds the surface with a lighter, cleaner mesh, before they reach an animation or real-time pipeline.

Parametric design reverses the priorities. A sketch is constrained, extruded or revolved into a solid, and every feature depends on those before it. Changing a dimension early in the history regenerates the later features automatically. The cost is rigidity: a feature tree can fail to rebuild if a referenced face disappears, and organic shapes are awkward to describe this way.

Capturing real objects

Scanning supplies models without manual construction. A 3d scanner measures a surface directly, while photogrammetry reconstructs it from photographs. Wikipedia describes stereophotogrammetry as a special case that estimates the three-dimensional coordinates of points on an object using measurements made in two or more photographic images taken from different positions. Common points are identified on each image, and the software computes where they sit in space.

The practical sequence is covered in a walkthrough on how to 3d scan an object. The outcome of any scan is a dense point cloud or triangle mesh that usually needs cleaning: holes are filled, noise is removed, and the mesh is decimated before it is useful. The raw scan is a starting asset rather than a finished model.

Photogrammetry has a second use beyond geometry. Wikipedia notes that it can extract accurate color ranges and values representing quantities such as albedo, specular reflection, metallicity or ambient occlusion from photographs of materials, for the purposes of physically based rendering. That is why game studios photograph real surfaces to build material libraries as well as to capture shapes.

3D model file formats: STL, OBJ, STEP, glTF and 3MF

The Library of Congress documents STL as an openly documented format for describing the surface of an object as a triangular mesh, first specified in 1988 by 3D Systems, Inc. in the StereoLithography Interface Specification. It was built for stereolithography, a form of 3D printing used in the late 1980s for rapid prototyping, and it became a de facto standard in the 1990s. The choice of container matters as much as the choice of modeler, and the detailed comparison lives in the guide to 3d model file formats.

STL: simple, universal, and limited

STL comes as ASCII and binary subtypes, and in practice binary files are more common because they are more compact. The binary layout is fixed: a header, a count of triangles, then a record for each triangle, made of a normal vector, three vertices and a short attribute field. Each triangle record has a fixed size, so a model of one million triangles occupies tens of megabytes before any compression.

The limits are as important as the layout. STL files contain no scale information, and the units are arbitrary, so a part exported in millimeters can arrive in a slicer as inches. They describe only surface geometry, without any representation of color, texture or other common CAD model attributes. A vendor workaround exists: Materialise Magics uses the file header to represent the overall color of the entire part, but it is not a universal convention.

OBJ, STEP, glTF and 3MF

OBJ is a simple data format that represents geometry alone: the position of each vertex, the UV position of each texture coordinate vertex, vertex normals, and the faces. Faces are written as lists of indices counted from the first element, and texture coordinates are normalized to the unit range. Because it is plain text and widely supported, it remains a common interchange choice for textured meshes, though it carries no animation or scene hierarchy.

STEP, an ISO standard family, serves the other side of the field. Work on STEP replaced older exchange formats such as IGES and VDA-FS, and the standard has been extended in major releases since. It exchanges exact CAD geometry and product data rather than a tessellated approximation, so it is the safer choice when a part moves between CAD systems.

The Khronos Group markets glTF as “the JPEG of 3D.” The core of a glTF asset is a JSON file describing a scene, which can be stored in a single binary file with the .glb extension. In 2022, glTF 2.0 was released as the ISO/IEC 12113:2022 International standard, according to Khronos. It suits web viewers, AR and real-time engines.

The consortium format targets printing directly. It is an XML-based data format designed specifically for additive manufacturing, and it can carry color and material information that STL cannot. The consortium behind it was first formed with members including Autodesk, Dassault Systèmes, FIT AG, HP, Microsoft, Shapeways and SLM.

Format	Stores	Typical use	Main limit
STL	Triangle surface only	3D printing slicers	No units, color or texture
OBJ	Vertices, UVs, normals, faces	Textured mesh exchange	No animation or scene data
STEP	Exact CAD geometry, product data	CAD-to-CAD exchange	Heavy for mesh-only work
glTF / GLB	Scene, meshes, materials	Web, AR, real-time	Not a CAD format
3MF	Mesh, color, materials	Additive manufacturing	Narrower tool support than STL

Table compiled from the Library of Congress STL format description and the Khronos glTF page.

From model to printer or headset

A model destined for a printer must pass tests that a rendered image never faces. The surface has to be closed, with no holes or inverted faces, and walls must be thick enough to print. The STL format’s lack of scale information, with units that are arbitrary, makes it essential to confirm the unit setting at export and again in the slicer. The detailed rules are in the guide to 3d modeling for 3d printing.

The wider process of turning a file into a physical part is covered under 3d printing. That history connects directly back to modeling: STL was created for stereolithography in the late 1980s, and the same triangle-mesh logic still feeds slicers today.

Real-time delivery has different demands. A model for a headset must render fast enough to avoid discomfort, so polygon counts, texture sizes and material complexity are budgeted tightly. Hub readers weighing hardware can start from the vr headset overview. glTF is the natural container for that work because it was designed for runtime delivery, and the Khronos glTF Validator checks assets against the glTF 2.0 specification and its extensions.

A reliable pipeline keeps a master file in the richest format available: parametric CAD or a layered Blender project. Lightweight exports, such as STL for a slicer or GLB for the web, are generated from that master as needed rather than edited directly. Doing so preserves design intent and lets the same source serve a printer, a viewer and a renderer.

Frequently asked questions

What is the most common way to represent a 3D model?

The polygon mesh. Wikipedia’s 3D modeling article states that polygonal meshes are by far the most common representation, and that the vast majority of 3D models today are textured polygonal models.

Which file format is best for 3D printing?

STL is the long-standing default: 3D Systems first documented it in 1988, and binary files store each triangle in a fixed-size record. 3MF, an XML-based format created by a consortium formed in 2015, adds color and material data that STL lacks.

Is Blender really free?

Yes. The Blender Foundation’s About page states that the project runs under the GNU General Public License and is primarily funded by donations. The source code was released after a September 2002 community campaign raised its target of €100,000.

Does an STL file store units?

No. STL files contain no scale information, and the units are arbitrary, according to the format’s Wikipedia entry. A part exported in millimeters can therefore be misread in a slicer, so the unit setting should be checked at export.

What is glTF used for?

glTF is a delivery format for web, AR and real-time scenes. Khronos calls it “the JPEG of 3D,” and glTF 2.0 became the ISO/IEC 12113:2022 International standard in 2022.