1. CS475 3D Game Development Level Of Detail Nodes (LOD)
Roger Webster, Ph.D. D&E Communications Computer Science Wing Caputo Hall Dept. of Computer Science Millersville University Millersville, PA
Some info & images from: University of Wisconsin - CS679
Copyright © 2016 Roger Webster, Ph.D.

2. Level of Detail Nodes (LOD)
When an object is close to the viewer, we want a high resolution model
Maybe thousands of polygons, high-resolution textures
When an object is a long way away, it maps to relatively few screen pixels, so we want a low resolution model
Maybe only a six polygons, or tens of polygons
Level of Detail (LOD) techniques draw models at different resolutions depending on their relevance to the viewer
Covers both the generation of the models and how they are displayed
A large research topic

3. Images from: University of Wisconsin - CS679
Sample LOD Models
Images from: University of Wisconsin - CS679

4. Sample LOD Models

5. Standard LOD
The current standard for LOD in games is to use a finite set of models
Models may be hand generated or come from automatic decimation or simplification of the base mesh (highest resolution)
Models are switched based on the distance from the object to the viewer, or projected screen size
Better metric is visual importance, but it is harder to implement
For example, objects that the viewer is likely to be looking at have higher resolution, objects at periphery have lower resolution

6. LOD Models Medium resolution model High resolution model
Low resolution model

7. LOD Sub-problems LOD Generation: Creating the different resolutions
Best method is to do it by hand
Otherwise, automatic simplification algorithms, of which there are many
Selection: Choosing which model to display
Simple per-object methods, like distance
Complex global methods for optimizing cost/benefit
Switching: Moving from one resolution to the next without artifacts
Just pop between them
Various blending methods

8. LOD Switching
The biggest issue with switching is popping, the sudden change in visual appearance as the models are swapped
Particularly poor if the object is right at the switching threshold – may flicker badly
There are several options:
Leave the popping but try to avoid flicker
Show a blended combination of two models, based on the distance between the switching points
Image blend – e.g. render both models for a transition period
Geometric blend – define a way to geometrically morph from one model to the next, and blend the models
Have more models that are so similar that the popping is not evident

9. Selecting LOD Models
For any given frame, we have to decide which model to display for each object
Assume that we have a discrete set of models for each object, at varying resolutions
Option 1: Choose each model for each object independently based on some error tolerance
Typically assume error depends on distance from viewer or similar
The current standard practice: fast, simple
Potential problem: You still might have too many polygons, in total
Option 2: Choose a number of polygons you can display, and find a set of models that lead to minimum error
Ensures near constant frame rate, but harder to implement

10. LOD for Animated Characters
Character->GetDetail()
Character->SetDetail(d)
Cal3D provides automated LOD generating and rendering

11. Switching Thresholds
Distance: Simply compute the distance from the viewer to a representation point on (in) the object
Projected Area for spherical bounds
Store the “radius” of the object
Compute distance in viewing direction
Scale by near clip plane distance (why?)
Estimated projected area is:
Projected area for bounding cubes have a different method

12. LOD Mesh Simplification
Take a high resolution base mesh and approximate it while retaining its visual appearance
Desirable properties:
Fast (although not real-time)
Generates “good” approximations in some sense
Handles a wide variety of input meshes
Allows for geometric blending from one mesh to another
Two essential questions:
What operations are used to reduce the mesh complexity?
How is “good” measured and used to guide the approximation?

13. Mesh Operations
Operations seek to eliminate triangles while maintaining appearance
Key questions:
What does the operation do?
Can it connect previously unconnected parts of the mesh?
Does it change the mesh topology?
How much does it assume about the mesh structure?
Must the mesh be manifold (locally isomorphic to a disc)?
Can it handle, or does it produce, degenerate meshes?
Can it be construed as a morph?
Can it be animated smoothly?

14. Vertex Clustering Partition space into cells
grids [Rossignac-Borrel], spheres [Low-Tan], octrees, ...
Merge all vertices within the same cell
triangles with multiple corners in one cell will degenerate
Slide from Michael Garland,

15. Vertex Decimation Starting with original model, iteratively
rank vertices according to their importance
select unimportant vertex, remove it, retriangulate hole
A fairly common technique
Schroeder et al, Soucy et al, Klein et al, Ciampalini et al
Slide from Michael Garland,

16. Iterative Contraction
Contraction can operate on any set of vertices
edges (or vertex pairs) are most common, faces also used
Starting with the original model, iteratively
rank all edges with some cost metric
contract minimum cost edge
update edge costs
Slide from Michael Garland,

17. Edge Contraction Single edge contraction (v1,v2)  v’ is performed by
moving v1 and v2 to position v’
replacing all occurrences of v2 with v1
removing v2 and all degenerate triangles v’ v2 v1
Slide from Michael Garland,

18. Vertex Pair Contraction
Can also easily contract any pair of vertices
fundamental operation is exactly the same
joins previously unconnected areas
can be used to achieve topological simplification
Slide from Michael Garland,

19. LOD Nodes