Thursday 19 February 2009

Basic Shear Warp - First Pass

I have done a first pass on the warp part of the shear warp algorithm, to skew the image back to its proper perspective. It works for the most part, but is still off at times especially at extreme angles.



For now I am happy enough with it while I work on run-length encoding and multithreading. Once the algorithm is running much faster (potentially at mostly interactive frame rates) I'll be able to see what's going on better and fix the warp then.

There are also obvious quality issues that I'll look at then too (mainly involving using a bilinear filter during the warp rather than reading the nearest pixel).

The speed of the algorithm is promising, currently a render such as the one above takes 2.3 seconds compared to 17.3 seconds for a raycasting version (single threaded).

I have begun to lay down the groundwork for RLE. I was planning on having it finished for this week's tutor meeting, which is tomorrow, but time has run out. The new plan is to have it done by the weekend. My plan for the meeting is just the usual checkup on progress.

No comments: