2007.01.18 11:09 "Re: [Tiff] Elevation Data", by Andrey Kiselev
On Wed, Jan 17, 2007 at 04:18:14PM -0600, Bob Friesenhahn wrote:
The TIFF specification is as old as dirt. Instead of attempting to intuit the use of old tags (over which we have no control at all) it seems better to define new tags and get them ratified as part of a new specification so that when the new tags are present, they may be relied upon.
The discussion up to now has made it very clear that there is little that can be done to salvage the existing tags.
I am agree with that statement. The major problem of this discussion is a mixing of terms STORAGE and REPRESENTATION. TIFF is flexible enough to store a big range of various datasets. That is why it is used so widely in scientific applications. But representation of that data is something completely different from storage. It is not enough to store a pair of values in addition to numeric matrix to properly visualize it. In first of all, that matrix can be visualized a number of ways: grayscale or color encoded raster image, countour map, 3D surface, profile map, vector map, and other possibilities exist. Usual raster image is just a one of possible ways to represent data and not always the best. Actually we can't visualize data without user interaction. We need user anyway, so storing the range along with the data just useless. Also we need much more info in addition to range: calibration, NODATA value (the code which marks "empty" areas), etc. And hystogram is mandatory anyway.
So I am prefer to leave things as they are now (which works).
Andrey V. Kiselev