- From: Thomas DeWeese <Thomas.DeWeese@Kodak.com>
- Date: Tue, 23 Nov 2004 15:57:19 -0500
- To: Tim Rowley <tor@cs.brown.edu>
- CC: www-svg@w3.org
Hi Tim, This approach was taken in some earlier drafts. The major problem is that with LOD you also need to provide cropping bounds otherwise when the user is zoomed in and viewing a tiny portion of the document, say in the upper left hand corner. The viewer will have to load/interpret the high resolution data for the entire document on the chance that some piece of the high resolution data has different bounds from the low resolution data (think of the use element or complex paths). Additionally, it provides some significant semantic information that all the children of the multiImage are alternate representations of the same entity. Remember that all test attributes can also be used outside of a switch. Tim Rowley wrote: > If a level-of-detail metric could be added to the conditions used > for <switch>, then images and any geometry could be used for LOD > display. This would allow multiImage/subImage/subImageRef to be > replaced with appropriate use of <g> and <image>. > > One possible LOD metric might be the ratio of userspace > coordinate units to device pixels. > > An example of how this might appear: > > <switch> > <image width="360" height="240" xlink:href="m8.jpg" max-lod="2"/> > <image width="360" height="240" xlink:href="m4.jpg" min-lod="2" max-lod="4"/> > <image width="360" height="240" xlink:href="m2.jpg" min-lod="4" max-lod="8"/> > <rect width="360" height="240" fill="#ff0000" min-lod="8"/> > </switch> > >
Received on Tuesday, 23 November 2004 20:57:21 UTC