On Mon, Oct 27, 2008 at 11:19 PM, Enrico Kochon <[EMAIL PROTECTED]> wrote:
> Does the bug in the screenshot result from the drawing order-stuff? Yes, that's exactly the kind of problem that there currently is. > Well, I were a little bit to quick by saying how easy the solution was. > What I mentioned would involve a lot of changes at all places. The > framework seems to be 2D-only, thus it cant cope with a 3D-rendering > pipeline, where the Z-Buffer-Thingy is a natural part. There are some pieces borrowed from the 3D world, like the collision stuff or the bounding boxes you see in the screenshot. But on the outside, it is 2D and I think we want to keep that. OTOH, I don't think we're doing anything different then what has been done for Ultima 7 in 1992 for example, so the problem is more in my ability to come to a solution that works 100%. (I'm lousy when it comes to math). The interesting thing to note is that everything worked for alpha 2. However, the rendering algorithm was different there, as the underlying data structures were different. Basically, it looked at grid and rendered each row of the grid separately. Now we get the complete scene from an octree and try to render it in one go. >From my playing around on the weekend, I think that one of the problems is the way objects "intersect". If I have an object 40px wide at position 20, it will intersect with an object that starts at position 60, if you implement common intersection-algorithms. However, in this case, we don't care about the intersection, because both objects are next to each other and don't really overlap. But by changing the intersection algo to test for <= instead of <, there occur problems with other cases. It's kind of a situation where I am too blind or too narrow-minded to see the solution. Which is kind of bad if there is no one else to look at stuff and possibly see the obvious flaw at a glance. Therefore I'd be really glad if somebody did look :-) I'll keep on trying and I'm sure I'll figure it out eventually, but it isn't very motivating at all ... I'd rather move on to new stuff :-). Kai _______________________________________________ Adonthell-devel mailing list Adonthell-devel@nongnu.org http://lists.nongnu.org/mailman/listinfo/adonthell-devel