How LOD (Levels of Detail) Saves Time in 3D Animation

Animators don’t always work with the highest level of detail.

It’s similar to watching YouTube videos—sometimes, you prioritize loading speed and lower the resolution to the minimum watchable quality. Other times, you want the full cinematic experience and opt for 4K resolution.

Likewise, animators adjust levels of detail (LOD) to balance visual fidelity with performance efficiency. When refining a scene’s timing, high-detail models aren’t necessary. But in post-production, ensuring the final render meets quality standards is essential.

This article explores the importance of LOD, how it optimizes production without compromising quality, and key techniques like bump mapping, retopology, texture baking, and displacement map baking.

What's LOD?

In animation, the level of detail (LOD) defines the varying degrees of complexity applied to models or scenes, especially when objects are viewed at different distances by a viewer or camera.

When an object is far from the camera, it's unnecessary to render all its fine details in high geometric detail so lower LOD models with fewer polygons are used: a tree seen in the foreground might have detailed leaves and branches, while a distant tree might be a simple shape with textures to conserve processing power.

Why Levels Of Details Matter

LOD allows animators and rendering artists to optimize computational resources: objects far away can be rendered with less detail without noticeable loss in visual quality, reducing the load on the processor. If your animated film is one hour long, you need to render 108,000 frames at 30 frames per second, so these processing optimizations quickly stack up.

Different levels of detail also mean only the necessary amount of polygons or textures are processed, saving up storage memory.

This is particularly crucial for real-time rendering, where you need to maintain high frame rates without bloating the animator's PC, but it's also important for the whole production pipeline to reduce rendering costs.

1. Bump Mapping

Bump mapping is a computer graphics technique that simulates bumps and wrinkles on an object's surface to create the illusion of intricate textures without increasing the model's geometric complexity with additional polygons.

It works by changing the appearance of the surface texture by manipulating the lighting calculations during rendering rather than changing the actual geometry. This is done by using a texture known as a bump map, which is typically a grayscale image where the intensity of the color represents the height of the surface perturbation.

Traditionally, modeling each brick in a brick wall with its unique surface details would require a dense mesh, with each bump and groove captured by additional polygons, which can be computationally expensive.

Instead, you can just create a bump map where the lighter areas indicate raised sections of the brick texture, while darker areas correspond to deeper sections like the mortar lines. When applied to a simple flat plane with minimal polygons, the rendering engine uses this bump map during shading calculations to perturb the surface normals. The interplay of light and shadow on these altered normals gives the viewer the illusion that the flat plane has all the intricate geometry of an actual brick wall.

2. Retopology

Retopology is a process in 3D modeling where the topology, or the structure of the mesh surface, is redefined to achieve better geometry flow and reduce polygon count while preserving the original shape and detail of the model.

Imagine you start with a highly detailed character model created using sculpting software like ZBrush. This model might have millions of polygons capturing every intricate detail. The retopology process would go like this:

1. Simplification - The software creates a new mesh on top of the high-poly model using retopology techniques (automatic like ZBrush's ZRemesher or manual for more granular edits) to define a more manageable polygon structure while capturing the character's essential forms.
2. Flow utilization - The tool considers edge loops around significant anatomical features like the eyes, mouth, and joints to help animation deformation (bending and stretching).
3. Polygon reduction - The new topology should have significantly fewer polygons.

Retopology is important to create clean, efficient mesh structures. It's especially important to create multiple versions of the character model corresponding to different levels of detail:

• LOD0 - The most detailed version is used when the character is close to the camera.
• LOD1 - A less detailed version for mid-range shots.
• LOD2 and beyond - Even simpler versions for distance shots.
• Vertex Weighting - The lower detail versions maintain essential shape and silhouette but use minimal geometry to optimize rendering performance.

3. Texture Baking

Texture baking pre-calculates and transfers various surface details and lighting information into a texture map. These baked textures include shadows, reflections, global illumination, ambient occlusion, or complex material properties that are computationally expensive to calculate in real time. Once baked, this data is applied to simpler versions of the 3D models to optimize performance without sacrificing quality.

By baking textures, the artist ensures visual consistency across different levels of detail: features like shadow placement and surface details can be preserved even with reduced geometric complexity. Lower LODs require less processing power to render, allowing more straightforward shading techniques to be used.

An intricate 3D model of a building that will be part of a large city scene in a real-time application has highly detailed geometry and complex materials with bump maps and reflective surfaces, which can be rendered beautifully in high-end renders but are too resource-demanding for real-time rendering during production:

1. High-Resolution Model Preparation - We start with a high-resolution version of the building model, where lighting and material effects are meticulously applied.
2. Baking Process - Using 3D DCC software, we bake the model's lighting information into a texture map―shadows, highlights, and any ambient occlusion effects into a 2D texture. You'll typically bake several maps, like diffuse, normal, and specular maps.
3. Create LOD Models - We generate several lower-resolution versions of the building model with fewer vertices and simplified geometry.
4. Apply Baked Textures - Artists apply the baked textures to these LOD models. Even though they have a reduced vertex count, the baked textures convey intricate details and lighting effects, keeping the visual quality high.
5. Implementation - In the real-time environment, the LOD system selects which version of the model to render based on the camera distance or screen space size. The pre-baked textures ensure that even the simplest model retains much of the visual complexity of the high-resolution asset.

4. Displacement Map Baking

Displacement map baking transfers the high-resolution details of a 3D model onto a lower-resolution version. This is also achieved by generating a texture, the displacement map, that stores the height differences from the low-resolution model's surface to the high-resolution model's surface. This texture can then be used to recreate the appearance of the high-resolution model during rendering without having to load the full complexity all the time.

1. Create High and Low-Resolution Models - We start with a high-resolution 3D model that has all the detailed features like bumps, creases, and other surface details. Then, we create a simplified version of this model with a reduced polygon count.
2. Bake the Displacement Map - Using software like ZBrush, Blender, or Maya, we calculate the difference between the two models' surfaces. The result is the displacement map, a grayscale texture where the intensity of the color indicates how much to displace the surface of the low-res model to match the high-res model's details.
3. Apply the Map to Low-Resolution Model - Once the map is baked, it's applied to the low-resolution model. The rendering engine will use the map information to simulate the high-detail surface during real-time rendering or animation playback.
4. LOD Implementation - We add the low-res model with the displacement map into a LOD system, which swaps out different levels of detail depending on the camera distance. When the model is far enough away, the low-res model appears as detailed as the high-res one thanks to the displacement map, saving processing power.

Displacement maps allow different LOD versions of a scene to share high-resolution details without the computational overhead, saving memory and simplifying asset management.

Conclusion

A level of detail is an important concept in animation to increase your animation pipeline's efficiency by adjusting the complexity of your 3D models based on parameters like distance from the camera or production environments.

Several techniques like bump mapping, retopology, texture baking, and displacement map baking allow animators to adapt their model's LOD, but there are more, and our article only gives a glimpse of how they work: you'll have to figure out by yourself how to make the best of them in your DCC tool. Consider playing with these techniques in your software and see how they impact the polygons and visual quality of your 3D models!