This is the implementation of the GPU-parallel marching cubes algorithm for block-compressed data sets described in "Interactive Visualization of Terascale Data in the Browser: Fact or Fiction?" by Will Usher and Valerio Pascucci appearing at LDAV 2020. Please see the paper and the supplemental video for more details.
This is now running to workarounds and updates implemented to get around temporary missing APIs and disabled features or limitations. The application runs in Chrome Canary 92.0.4512.0 (and later versions as long as the WebGPU API is not changed), though some data decompression corruption has been seen on macOS for some data sets. On Windows it runs well, so try it out online!
- Skull (256^3)
- Magnetic Reconnection (512^3)
- Chameleon (1024x1024x1080)
The data sets shown are available on the Open SciVis Data Sets page.
The Chameleon data set
The Plasma data set
