Sam Donow’s scientific contributions

What is this page?


This page lists works of an author who doesn't have a ResearchGate profile or hasn't added the works to their profile yet. It is automatically generated from public (personal) data to further our legitimate goal of comprehensive and accurate scientific recordkeeping. If you are this author and want this page removed, please let us know.

Publications (1)


Figure 2. The latlong indexing scheme shown for encoding vectors with a maximum error of 10 @BULLET to make the regions clear (derived from Smith et al.'s [2012] figure 4). Only a few indices are labelled in the diagram.  
Figure 4. Optimal rounding direction for the y-component during float→snorm conversion under the precise oct encoding algorithm at 12 bits per pixel. Black = floor, white = ceiling.
Figure 5. Distribution of representable unit vectors at 16-bit precision, with 24-bit snorm8×3 added for comparison. 24-and 32-bit representations yield distributions too dense to visualize effectively at this scale. Darker areas represent regions of lower error.
A Survey of Efficient Representations for Independent Unit Vectors
  • Article
  • Full-text available

April 2014

·

1,792 Reads

·

51 Citations

Zina H. Cigolle

·

Sam Donow

·

·

[...]

·

Download

Citations (1)


... Our implementation includes several buffers: the surface parameter buffer, which requires 9 floats (36 bytes per surface), the encoded parameter buffer, utilizing 39 half floats or 78 bytes per surface and an additional 4 bytes for each direction, where a neural sample is requested, are stored in the directional request buffer. The unit direction vector is mapped to an octahedron, stored as 32 bits as in previous works [Cigolle et al. 2014;Meyer et al. 2010]. In our experiments, the NIRC operates on up to 3 non-specular vertices per light path, with a total of 25 neural samples. ...

Reference:

Neural Two-Level Monte Carlo Real-Time Rendering
A Survey of Efficient Representations for Independent Unit Vectors