University of Alberta2026AI & MLResearch
Perception-Oriented Volume Rendering
How transfer functions, sampling density and shading change what a reader can actually see in subsurface seismic volumes.
Direct volume rendering has a lot of knobs and very little guidance on which ones matter for interpretation. This project treats that as an empirical question: build a VTK pipeline over a real 3D seismic velocity model, then generate 27 controlled render conditions varying transfer function, sampling density and shading, and compare them side by side.
The output is an atlas for structured comparison plus an interactive prototype that switches between presets live, so a domain reader can find the configuration that reveals the structure they care about rather than accepting a default.
Highlights
- 27-condition controlled sweep over transfer function, sampling and shading.
- Comparison atlas for side-by-side perceptual evaluation.
- Interactive VTK prototype for live preset switching.
- Built on the USGS Cascadia v1.7 seismic velocity model.
Gallery
