I highly suspect the fill rate gains you get from using DirectComposition/Core Animation for scrolling/zooming *static* vector graphics in 2D completely outweigh any gains from any fancy GPU vector rendering algorithm you could come up with in practice.
-
-
Replying to @pcwalton
agreed but it's a weird comparison to make, given you can just do both? just sad about where we've focused our resources?
1 reply 0 retweets 1 like -
Replying to @Gankra_
Oh, I'm not sad at all! I'm actually very excited about the WebRender/Pathfinder roadmap for the reason you state: we should do both. Really I'm just pointing out a blind spot of most vector graphics research. If a technique has no OS compositor story it's a severe problem.
1 reply 0 retweets 3 likes -
Replying to @pcwalton
in what way is it a problem? unrealistic workloads/benchmarks that are "solved" by basic compositor use? algorithms that don't "work" with a compositor? something else?
1 reply 0 retweets 1 like -
Replying to @Gankra_
So the typical vector graphics paper will show panning and zooming on static content are much faster with their algorithm over a CPU rasterizer that rerasterizes every frame. The problem is that this isn't a real world comparison.
1 reply 0 retweets 2 likes -
In the real world, for static content you're competing against a system that caches tiles at various mip levels and interpolates somehow to zoom. It's almost impossible to beat that with a system that rerasterizes every frame, except on very specific content.
1 reply 0 retweets 1 like -
So you either have to show that you can generate new tiles faster and that this matters on static content, or that your quality improvements can beat trilinear filtering. Both are hard, especially if your technique has high setup time.
2 replies 0 retweets 1 like
I think instead of focusing on panning and zooming of static content, researchers would be better off focusing on known-problematic cases like complex PDFs where initial generation of the tiles is horribly slow today or on dynamic content, where tile caching isn't helpful.
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.