This library has two primary goals:
- Implement state of the art and production ready animation compression algorithms
- Serve as a benchmark to compare various techniques against one another
It is very common for most game engines or animation libraries to implement their own animation compression but the problem is so narrow and focused in scope, that it makes sense for the industry to converge on a single implementation.
Academic papers on the subject often will not compare techniques using the same error metric or raw size function which can skew the results or make them very hard to evaluate. They might also compare against techniques which are not state of the art or even used in production. This library aims to implement as many techniques as possible, side by side, in order to compare them fairly.
Much thought was put into designing the library for it to be as flexible and powerful as possible. To this end, the following decisions were made:
- The library consists of 100% C++ header files and is thus easy to integrate in any game engine
- An intermediary clip format is supported in order to facilitate debugging and bug reporting
- All allocations use a game provided allocator
- All asserts use a game provided macro
The library aims to support the most common platforms for the most common use cases. There is very little platform specific code as such it should work nearly everywhere.
The math library is not yet fully optimized for every platform. The overwhelming majority of the math heavy code executes when compressing, not decompressing. Decompression is typically very simple and light in order to be fast. Very little math is involved beyond interpolating values.
- Windows
- Windows
- Android
- Uniformly sampled
- Linear key reduction (TODO)
- Spline key reduction (TODO)
- Wavelets (TODO)
- Install Visual Studio 2015
- Install CMake 3.9 or higher
- Install Python 3.3 (version 3.3 is required for the FBX SDK by some scripts)
- Generate the IDE solution with:
python make.py
The solution is generated under./build
Note that if you do not have CMake in yourPATH
, you should define theACL_CMAKE_HOME
environment variable to something likeC:\Program Files\CMake
. - Build the IDE solution with:
python make.py -build
In order to test the algorithm implementations with real world data, we took the Carnegie-Mellon University motion capture database. It contains over 2500 animation clips. This is the data all graphs and charts shown will be based on. Data available upon request, it is far too large for GitHub.
Copyright (c) 2017 Nicholas Frechette & Animation Compression Library contributors