Scientific Computing · Cluster
Scientific Visualization
Encoding data into perceivable form.
- 01 Perceptually Uniform Colormaps and the Failure of Rainbow Ramps How a scalar field becomes color through a lookup table, why monotonic lightness and equal perceptual steps are the two constraints that matter, and where banding and false contours come from.
- 02 Sample Positions, Trilinear Interpolation, and the Half-Cell Offset Why a scalar grid carries two coordinate conventions, what trilinear interpolation costs and how it errs, and how resampling a field to a second grid trades resolution against aliasing.
- 03 Marching Cubes and the Case Table That Cannot Be Interpolated How a 256-entry table turns a cube of eight samples into triangles, why edge interpolation places the vertices, and where the classic ambiguity creates holes or spurious connections.
- 04 Scattered Data Interpolation and the Conditioning of Radial Basis Functions Why meshless reconstruction solves a dense linear system, how the kernel choice decides between smoothness and conditioning, and what local stencils recover at a fraction of the cost.
- 05 Streamline Integration in Vector Fields and Where the Error Accumulates How a vector field becomes a curve through numerical integration, why cell-crossing interpolation dominates the error, and what separates a valid line from a decorative one.
- 06 Direct Volume Rendering and the Opacity Correction Nobody Applies How ray casting accumulates colour and opacity along a ray, why a transfer function tuned at one step size breaks at another, and where pre-integrated classification changes the answer.
- 07 Multi-Dimensional Transfer Functions and the Gradient-Magnitude Histogram Why a value-only transfer function cannot separate two materials with the same density, how the joint histogram of value and gradient magnitude classifies boundaries, and what the extra dimensions cost.
- 08 Tensor Fields and the Ambiguity of Eigenvector Glyphs How a symmetric rank-two field is decomposed into three directions, why those directions have no sign, and what happens to the picture at isotropic and planar degeneracies.
- 09 In-Situ Analysis and the Bandwidth That Is Not There Why moving simulation output to a separate visualization process stops working at scale, how coupling the analysis into the solver changes the algorithm, and what the shared-memory boundary costs.
- 10 Sort-Last Compositing and the Depth Order Across Ranks How independently rendered partial images are merged into one correct image, why binary-tree compositing beats direct all-to-all, and where blending order and depth precision break the result.