For discussion: non-realtime animations

:rolleyes: Maybe it takes a few years for the little light-bulb to blink on, but eventually it does. (And, finally having a computer with a decent video-card undoubtedly helps!)

I am now looking at a non-realtime project … another kiosk-animation for DVD … for which I would like to use GameBlender. My objectives for doing so are: - To achieve hardware rendering as much as possible, in order to generate completed frames (or useful composite-layers thereof, as the case may be…) in the minimum amount of time by using the graphic-card’s GPU. - To be able to describe the effects of such things as “a metal-working tool striking the piece of metal to be worked” in an algorithmic way (I am a fluent Python programmer), again to produce realistic animations more quickly.

As I said, this project is “non-realtime.” In other words, I intend to use GameBlender to create the frames and then use Rasterizer.makeScreenshot to cut a frame, as this article demonstrates. It does not matter to me if the computer requires a few milliseconds to acceptably render the frame, as long as I am not faced with minutes per-frame as I am right now. It also does not matter if additional compositing and post-processing steps are needed … if the computer can get me 95% of the way quickly, “I shall be a very happy boy.”

What I’d like to know, and to discuss, right now is … - What are the realistic limits of this technique? - “If you had to do it all over again,” what do you today sorely wish that you had known yesterday? - What are the best articles and pointers that I should read?

i wish i knew about full-render bake when trying to make (poor) textures for my meshes that result in flat looky boxy stuff in GE.

and if real-time is not the goal, some decent mesh models with some actualy polygons might actually look good.