Hi, > What about the approach taken in those "rounded corner" scripts ? > I know that some versions use images and it seems that others have an > algo approach.
IIRC it uses the same approach as Walter Zorns library. That is a great idea, but it needs to generate many DOM-Nodes. In case you draw similar to an airbrush-effect with many single pixles, multiple times at the same point with varying alpha-values, etc., you get into trouble. That is fine with a real canvas-implementations but with Walter Zorns library it will quickly eat up all your memory (I have tried). > The first way could be a quick'n easy hack, by using a large image that > can be resized smaller. The scaling algorithms used in Browsers are of - erm - varying quality. Actually they are usually optimized for speed (i.e. often they simply leave out lines and rows) instead of quality. You get very poor looking results with that. > I don't know much about the pure algo way by handling 1x1 pixel divs, > but if IE's memory can handle it, then it might be powerful. I don't know much about the various ActiveX-controlls, but I think I remember that there were some that could take data-URLs as Parameters. Modifying that Parameter should also work with IE. There is also another aproach: IE interprets javascript-URLs as sources for Images. It exspects the JS-code to return monochrome XBM Images as a string. Using filters you can give that image a specific color, make the white pixles transparent and add an alpha-value for the "black" (now colored) pixles. If you overlay multiple images, you should be able to create "True-color"-Images that way. The Point is that you still only have a fixed set of images instead of very many divs. Setting the same pixle with varying alpha-values multiple times doesn't add more and more divs, but only changes the images without eating up the memory. That behaves much more like Canvas in the memory-aspect. Christof _______________________________________________ jQuery mailing list discuss@jquery.com http://jquery.com/discuss/