. Author manuscript; available in PMC: 2009 Sep 14.

Published in final edited form as: J Parallel Distrib Comput. 2009 Jan 1;69(1):39–53. doi: 10.1016/j.jpdc.2008.07.006

Table 1.

Comparison of compressed isosurface sizes in bytes using edge index coding with a general surface compression algorithm

Model	Value	Data type	Orig. size	Gen. alg.	Edge index
Hipip	0.0	float	1,465,862	129,068	88,804
Foot	600.	u_short	7,536,227	587,344	407,658
Foot	1500.	u_short	8,645,243	761,413	386,506
Engine	180.	u_char	3,601,040	224,893	121,504
Engine	89.	u_char	15,888,473	1,012,491	618,492

The general compression algorithm uses 8 bits for each coordinate per vertex. For unsigned short and char data type, we encode them directly using the predictive coder. Float values are normalized and quantized using 32 bits for the first point and 14 bits for difference.