thaw@pixar.UUCP (Tom Williams) (07/21/89)
eugene@eos.UUCP (Eugene Miya) writes: >Sorry, I've not found the terrain rendering of things like Yosemite Valley >all that impressive. Its getting there, but you have to learn to take >a closer look. There is more detail and information in SOME of the original >images. The surfacing mapping loses some critical details, some shadows >get portrayed wrong (due to original image capture, etc). Of the better >renderings are LA the Movie, the aforementioned Calgary piece, but they >all still lack a few things here and there. Incorrect shadows in terrain rendering can be caused by many problems. A lot of terrain rendering suffers from pre-rendering steps which are ignored or handled badly. For example, elevation-color image reqistration is very important. So is contour interpolation, and inverse "camera" projections. Most of these are because people don't need and/or have the time to do it right. The only problem inherent with current techniques is the limitation we have by using 2 1/2 dimensional data. Assuming you have a good acquisition, you should never lose data contained in the orignal image. Some data can be filtered out and if you are forced to deal with data which doesn't support your output resolution then you will strech the source data (unless you have a nifty algorithm for adding fractally correct elevation & color). Of course, a big "problem" is the dynamic range of the data required for a good animation. Terrain rendering consumes vast quantities of data and computrons. Alot of this can be avoided by carefully planning you animation and only using resolutions of data required to preserve the Nyquist (sp?) limit for output. The again assumes (for most projects incorrectly) that you can access data at the resolutions you desire AND handle source images acquired at different times of day, with different camera geometries, and with different color responses. Remember results vary greatly depending on desired effect. LA the movie used relatively low resolution data and a rendering scheme with artifacts which caused a lot of shimmer. But it was IMHO a breakthrough animation, I loved it (and Mars the movie too). If they had tried to use the highest resolution data available and state of the art rendering techniques they'd still be rendering today. But the picture of Yellowstone (not Yosemite) that was done for National Geographic (Feb89) attempted to handle most of the problems detailed above as well as the budget allowed, but used a frightening amount of disk space and cpu-trons. -thaw- ---- Thomas Williams ..!{ucbvax|sun}!pixar!thaw ----