A comparative analysis of some considerations for browser performance in SVG.
Outline of remarks

David Dailey
Slippery Rock University

See the original paper itself

Introductory remarks

  1. interest in using SVG for animation
    examples -- all JavaScript & no SMIL: balloon ; all SMIL & no JavaScript ovals
  2. SMIL seemed slower (or more jagged) than JavaScript animation -- in some cases
    examples -- rotation of clipped image ; Gaussian blur
  3. JavaScript and SMIL animation or filters sometimes seemed to conflict.
     examples -- egg cloning on washboard
  4. There seemed to be browser differences
    a. SMIL and filters may or may not work in certain browsers, and in context of certain attributes
    examples -- wiggleworm in Opera or IE/ASV ( animate attributeName="stroke-dasharray" apparently not in IE )
    b. IE/ASV tended to be faster for many operations
    example: starflakes
  5. Need to quantify these observations

The programs

  1. Run time vs. render time and how to measure.
  2. Animation chamber (SVG Open 2006)
  3. SVG DOM chamber

Main results

  1. Table 1 : Time to build basic graphic elements, by browser
  2. Figure 2 : Adding to the DOM concurrently vs successively, by browser
  3. Table 3 : Time to render path objects
  4. Figure 3: Time to render path objects (with browser effect partialled out)
  5. Table 5 : Time demands of transparency, by browser
  6. Figure 5: same thing broken down by type of object
  7. Table 6 : Time demands of transform=translate, by browser
  8. Table 7: Time demands of clipPath, by browser
  9. Table 8: Time demands of various filters
  10. Table 9 : Methodology for timing animation

Other results?

See paper

Conclusions

Questions?? (mail to ddaileyAtsruDotedu)