Linear Algebra for FE Developers
summary
Describes using linear algebra and homogeneous coordinates to treat image and sensor operations as single matrix transforms so translations, scales, rotations and reflections compose linearly and stay invertible when expected. Discusses performance issues with very large matrices and demonstrates a hair simulation implemented as a damped mass–spring system solved with the Euler method over 50,000 points to produce plausible animation.
AI-written summary · about
chapters (13)
- 0:00Introduction
- 1:05Linear algebra for FE
- 4:00Matrix functions in vision
- 7:55Coordinate systems
- 11:29Homogeneous coordinates
- 12:13Translation and scale
- 15:042D object rotation
- 16:30Linear transformation properties
- 18:17Reflection and affine
- 20:47Isometric transforms
- 23:38Numerical methods in VFX
- 27:09Hair simulation mass-spring
- 34:15Summary
related talks
01h 01mGuide to Linear Algebra With the Eigen C++ Library
28mCommunication Avoiding Algorithms for Linear Algebra and Beyond
32mFrom Geometry to Algebra and Back Again: 4000 Years of Papers
26mFinite Difference Methods Fengshui: Alignment through a Mathematics of Arrays
23mPattern-Based Optimization (...)
48mC++ Algorithmic Complexity, Data Locality, Parallelism, Compiler Optimizations, & Some Concurrency
06mLightning Talk: Let's Fix Sparse Linear Algebra with C++. It'll Be Fun and Easy!
36mA LUT (lot) of image filters on Android