Possible future speed-up for the Blender engine and Blender?

I don’t know how the Blender engine compares for performance with other game engines. Anyway, I don’t know if you heard about it already but it’s a new algorithm to compress 3D scenes so they run faster, supposedly much faster:

http://www.usc.edu/dept/engineering/news/2004_stories/2004_06_15_desbrun.html

I don’t know if it’s implementable yet but it might be something to keep an eye on.

Isn’t it just an another decimator algorithm? As far as I understand it’s a designer’s tool not tool for engine.

I think it’s designed for real-time object tessellation so that the Blender engine, for example, wouldn’t have to draw so many polys and therefore run faster. I’m not sure if it would work on a design level (like in the Blender model view). I guess if it was able to draw selected objects (like a mountain) more quickly while you model another object, then possibly.

In linked article there isn’t mentioned word “real-time”. What it describes is converting meshes produced by 3d scanners (and other sources) to form usable for 3d presentation with guaranteed accuracy.

this article isn’t indeed abt speed of real-time 3d applications and how they can be made run faster, but about fast AND accurate 3d-mesh optimization techniques. so, these are two different things. .

One of the main uses is compressing 3D file sizes. But it looks like it might speed games up too because the technique allows real-time rough/smooth poly tessellation. So if you had a game with complex mountain ranges in the background, you could make them rough until you got close to them. I think some games already do this sort of thing don’t they?

The process also allows direct manipulation of the results for special purposes by the user – making it a very convenient tool for digital artists in animation studios. The user can select particular areas of a 3-D representation to make them either less or more detailed, or to emphasize them.

That’s a quote from the link. They seem to think that 3D artists will benefit from it. That is being able to select objects to be more or less detailed. It is sort of like the subdiv button but obviously, that just smooths rough models. Some models are inherently complex and high poly and so cannot be adequately reduced for modelling without losing detail so in that case such an algorithm would be quite handy. Of course, you could just make the complex objects invisible but sometimes that’s not the best way. In fact, I think that’s what they do in video games - when you see buildings appearing on the horizon in the likes of GTA3.