How to 3D Model: Mesh, Vertex and Modeling Types Explained

3D modeling is the process of building a digital object out of points, lines and surfaces in 3D space. Anyone who has opened a modeling program and stared at an empty grid has asked the same question: how do you actually turn that grid into a finished object. This guide walks through the basic parts of a 3D model, the main modeling types, and the steps that take a project from a blank scene to a finished model. For a plain definition first, see what 3D modeling is.

3D modeling supports games, product design, film, engineering and architecture, and each field leans on different techniques. 3D modeling services exist because building a clean, accurate model takes practice, and not every team wants to build that skill in house. The rest of this guide covers the terms, types and steps you need whether you plan to model yourself or brief someone else to do it.

3D modeling basics: the building blocks and principles

Every 3D model, in every modeling program, is made from the same small set of parts.

• Vertex. A single point in 3D space. It is the smallest unit of a model.
• Edge. A straight line that connects two vertices.
• Face. A flat surface bounded by edges. Three edges make a triangular face, four make a quad, five or more make an n-gon.
• Mesh. A collection of vertices, edges and faces connected together. A model can be made of one mesh or several.

A model with fewer faces is a low poly model: lighter to move around the viewport and quicker to render, but less detailed. A high poly model holds far more detail but is slower to edit and heavier to render.

A few principles carry across every modeling type and every program. Keep as few points as the shape needs. Route edges so they follow where a surface bends or a hard edge sits, not at random. Add detail only where the camera or a close up view will actually see it. These are the same principles of 3D modeling you will meet again and again, whatever technique you end up using.

Most modeling programs let you switch the viewport to a wireframe view, which shows only the edges of the mesh and hides the shaded surface. Checking a model in wireframe is the fastest way to spot topology problems: edges that pinch together, faces that are not needed, or areas where the mesh is denser than the shape actually requires.

3D modeling or 3D modelling: is there a difference

No. “3D modeling” and “3D modelling” are the same thing, spelled two ways. American English uses one L, “modeling”. British English uses two Ls, “modelling”. This site uses the American spelling throughout, but if you search or read either version, you are looking at the same field, the same techniques and the same software.

Types of 3D modeling

There is no single correct 3D modeling type. Games, product design and CAD engineering each favor a different one, often more than one on the same project.

Polygon modeling

Polygon modeling builds a shape from connected vertices, edges and faces, the same building blocks covered above. A modeler typically starts from a simple primitive, like a cube or a sphere, and pushes, pulls and cuts it until the target shape appears, then refines the edge flow one section at a time. It is the most common technique for game assets, characters and props because the result is a light, flexible mesh that renders fast. See the full breakdown of a polygon mesh for more on how the faces and edges are organized.

NURBS modeling

NURBS stands for non-uniform rational basis spline. Instead of vertices and flat faces, a NURBS surface is defined by a mathematical curve, so the surface it describes is perfectly smooth rather than built from small flat panels. That makes NURBS the standard choice for product design and automotive surfaces, and for CAD work in general, where a curved panel or a car body needs to be geometrically exact rather than just visually smooth.

The tradeoff is workflow. NURBS patches are always four sided, which limits how surfaces can be joined, and a NURBS model cannot be UV unwrapped or textured directly the way a polygon mesh can. For games, animation or anything that needs a texture, a NURBS surface is usually converted into a polygon mesh once the shape is locked in. NURBS is rarely used where render time matters most, since the curve math is heavier to calculate than a flat polygon.

Tip: Use NURBS to lock in a precise curved surface early, then convert to a polygon mesh before texturing or animating it.

Digital sculpting

Digital sculpting works like traditional clay sculpting inside a 3D program. Instead of building up edges one at a time, a sculptor pushes, pinches and smooths a dense mesh directly, starting from the broad shape and working down to fine surface detail. It suits organic forms such as characters, creatures and skin folds, where the shape does not follow straight edges or flat panels. Compare this to hard surface modeling, which covers mechanical, manufactured objects with straight edges and flat panels, the opposite end of the same spectrum.

Solid modeling

Solid modeling represents the full volume of an object, not just its outer surface, which makes it common in mechanical CAD. A part is usually built from a 2D sketch that is then extruded, revolved or cut, with every step recorded in a history so a dimension can be changed later and the whole part updates. This solid, parametric approach favors accuracy over speed, since every model has to be geometrically correct inside and out.

Procedural modeling

Procedural modeling generates a shape from a set of rules and parameters instead of manual point by point work. A modeler sets up the rules once, then adjusts the parameters to produce variations. It is used for terrain, foliage, city blocks and other cases where hand modeling every instance would take far too long, and where changing one number should regenerate the whole result.

This pair shows the same polygon model at two stages: the untextured mesh, with a polygon and vertex count visible in the corner, and the finished product rendering after materials and lighting were added.

How to 3D model: step by step

The exact steps shift by project and technique, but most 3D models are built in this order.

1. Gather reference. Collect photos, drawings or measurements of the object before opening the modeling program. Guessing proportions from memory is where most early mistakes start.
2. Block out the shape. Start from simple primitives and rough in the main proportions before adding any detail. Blender is a common program to start in, and the blocking stage works the same way in any of them.
3. Refine the mesh and check topology. Add edges only where the shape needs them and keep the polygon count as low as the level of detail allows at this stage.
4. UV unwrap the model. Flatten the mesh into 2D space so a texture can be applied without stretching or visible seams. See UV unwrapping for the common projection types.
5. Texture the surface. Apply color, roughness and other surface maps once the UVs are laid out. Texture mapping covers the main map types and how they combine.
6. Prepare the model for its final use. A game asset usually needs a set of LOD versions, from LOD0 down, so it can be shown at a lower detail level from a distance. A product or architectural model is instead exported at full detail for rendering.

Common 3D model file formats

Once a model is finished, the file format you export it in depends on what happens next: whether the model is going to a game engine, a 3D printer, or another CAD program.

• STEP. Describes product and mechanical part data without depending on one modeling program, so it moves cleanly between different CAD software.
• STL. The standard format for 3D printing, since it is easy for a printer or slicer to read. It stores only the surface, not color or texture.
• OBJ. Also common for 3D printing, and widely supported for carrying faces, coordinates and texture maps between programs.
• FBX. Carries geometry, color, texture and animation together, which is why it is common in games and film.

None of these formats is universally better. Pick the one your next tool in the pipeline, whether that is a printer, a game engine or another CAD program, actually expects.

Common 3D modeling mistakes to avoid

Most modeling mistakes come from moving too fast through the early stages. Watch for these:

• Starting a complex project before the basics of the chosen technique are solid.
• Subdividing the mesh for extra detail before the base shape and proportions are correct, which locks in errors that are hard to fix later.
• Ignoring topology: stray vertices and uneven edge flow cause problems later, especially before texturing or animating a model.
• Trying to model an entire object as one continuous piece instead of separate, manageable parts that can be checked and fixed on their own.
• Skipping reference images and guessing dimensions instead of measuring or checking them against a drawing.
• Aiming for one continuous mesh with no visible joins when the real object is actually built from separate parts. Match the model to how the object is really built, not to an idea of what looks tidy.

When to have 3D models made by a studio

Game studios often outsource 3D modeling, usually for game ready assets, LOD sets and texturing on a schedule the internal art team cannot fit in. Product, furniture and building models are other common jobs: a manufacturer needs an accurate product model before a rendering or a print run exists, and a developer needs a building modeled before it is built.

What a studio needs from you: reference drawings, photos or CAD files, a sense of scale, and any brand or material specs. What you get back: a checked, correctly scaled model in the file format your pipeline uses, ready for rendering, printing or import into a game engine. A studio also carries the modeling technique decision for you: whether a part should be built as a polygon mesh, a NURBS surface or a sculpted form is judged case by case, based on what the model needs to do once it is finished, not on which tool is easiest.

See the 3D modeling service page for what is included, and the game studios page for how this works specifically for game assets.

Questions about 3D modeling

What is the difference between 3D modeling and 3D rendering?

3D modeling builds the shape: the mesh, surfaces and geometry of an object. 3D rendering adds materials, lighting and a camera to that shape and produces the final image. Modeling comes first, rendering after.

Which 3D modeling technique should a beginner start with?

Polygon modeling is the most common starting point, since it uses a small, consistent set of tools and the results work in almost any other program or engine afterward.

Can a NURBS model be used in a game engine?

Not directly. Game engines expect polygon meshes, so a NURBS surface is converted into a polygon mesh first, then textured and imported like any other asset.

What file format should I use to send a model for 3D printing?

STL and OBJ are both standard for 3D printing and read reliably by most printers and slicers. STEP is a better choice if the model still needs to move between CAD programs before printing.

Is 3D modeling hard to learn?

The basic tools in any modeling program can be learned quickly. What takes longer is topology, proportion and the judgment to know how much detail a model actually needs, and that comes from practicing on simple objects before moving to complex ones.