3D optimization and how to use it in 3 simple steps

What is 3D optimization, and why is it needed?

3D optimization – when applied correctly to a 3D model – can make it possible to use a 3D model in many different applications.

Since 3D models are a complex medium, and since all applications have different requirements, this usually involves the adjustment of various aspects of the input model for best compatibility and performance - including, but not limited to, polygon count, texture count, texture resolution, or formats used for 3D model and texture files.

What aspect does a typical 3D optimization workflow consist of?

3D optimization consists of various aspects which are typically all important, especially when aiming at real-time 3D and XR applications:

Learn more about recommended setting for typical publishing targets in the Khronos 3D Asset Creation Guidelines.

Useful Definitions

The following definitions might be useful to understand when diving into 3D optimization:

• 3D mesh optimization: Process of making an input mesh more suited for a target application – typically involves the reduction of polygon count and re-defining the topology.  
• Interior culling & small feature culling: Process that removes invisible inner parts or very small parts (both kinds typically encountered in CAD data, for example), which aren’t really so impactful or even visible in the final 3D visualization and can hence be omitted for the sake of having a compact asset that renders much faster.  
• Scene optimization & draw call reduction: Process of restructuring the 3D scene logic, so that it is easier to further process, say, in real-time and XR applications. When reducing draw calls, this typically directly speeds up rendering, especially when the input has hundreds of separate parts.  
• Mesh & texture compression: Method of applying compression to mesh data (e.g., Google Draco) or texture data (e.g., UASTC or ETC1S via KTX), so that the data consumes less memory during transmission or even at rendering time (the latter is usually the case for texture compression methods).    
• Material baking: Process of creating textures that represent original material data in a texture image – often a PBR texture set, which makes the asset ready for Physically-Based Rendering. Baking can represent all kinds of attributes in respective textures, such as colors, roughness, “metalness”, or surface mesostructure (normal maps). For this to work, the model to bake on requires non-overlapping UV coordinates.

In the following, you can see an interactive example of 3D optimization on a 3D model, using a wireframe overlay.
The model on the left has 9 textures / 7 draw calls, while the model on the right has a single texture atlas / single draw call. Also, polygons were reduced from 900K (left) to 100K right:

Optimizing a 3D model in 3 simple steps

If you have 3D models that you'd like to optimize, dont' worry - it's actually easier than it sounds! Here's how you can do it using RapidPipeline, in 3 simple steps:

1. Upload 3D model. Upload your 3D model at RapidPipeline. If you don’t have an account yet, you can sign up for free (no credit card required) .  

2. Launch optimization. Dial in your own target settings (polygon count, texture resolutions, baking settings, …) in a custom preset (option 1) OR choose a factory preset (option 2).

   Example for using a factory preset: Choose the “Single-Item Mobile” preset to prepare your asset in a way that works well for most assets and mobile phones, for Web-based 3D and mobile AR.

   Select the models you want to optimize by checking them in the list, then press the “Optimize Selected Models” button to start the optimization.

3. Download results. Download your resulting files for use in your own application (typically, creators will use .usdz and .glb files for mobile targets, but in the preset settings other formats can be selected too for use in game engines, etc., including OBJ and FBX, for example).