Revision 2020 - new/fresh compo idea?

category: residue [glöplog]
Also let's conveniently forget that what you so easily dismiss as "animation players" is in 99% of cases way more complicated, well planned and optimized code than your copperbars and poly fillers, shall we.

So much this. Let's have a relatable example, shall we? Just because
Do we really need writeups now to appreciate anims? ;)

Smoke & Mirrors was criticised (by Photon, among others) for having a lot of pre-rendered 3D content. Here's how it works:

The 3D engine is written is Amiga assembly and integrated into the demo code, so the 3D scenes can be scripted in Rocket along with the rest of the demo. It features texture mapping, glossiness mapping, 16-bit Z-buffering and some world-space post-processing (for the rust and de-rust effects). The main 3D scenes in the demo contain tens of thousands of triangles and are far from being able to run realtime on a real Amiga.

The 3D scenes are composed of realtime and non-realtime parts (typically background and 3D models, respectively). When scripting the demo in Rocket, everything is rendered in realtime. Since this work is done in an emulator, it achieves a workable framerate (around 5fps IIRC).

To build the demo, it is run in a special recording mode, where every frame of the demo is rendered with and without the non-realtime elements. The two renders are compared in 4x4 pixel blocks, and all the blocks that differ are compressed using a jpeg-inspired compression scheme and stored in a data file. The parameters for the compression (four of them) are also controlled in Rocket, so that the size vs. compression artifacts tradeoff can be adjusted on a scene-by-scene basis. Remember, this was for a size-limited competition: the demo archive had to be at most 20MB.

When running the demo proper, only the realtime elements are rendered, and the non-realtime elements are decompressed on top.

It would have been so much easier to just throw in the usual, plain texture mapper on a few hundred polygons with slightly buggy z-sorting. But that would have been extremely boring, both for me and for the audience.

So here's a challenge for you, Photon: release a demo (for a constrained platform, of course) containing a substantial amount of pre-rendered content that blends (reasonably) seamlessly with the realtime content. Then come back here and say that it was easy. :)
added on the 2020-02-18 22:06:13 by Blueberry Blueberry
Let's not also forget that Photon considers every effect he doesn't understand (or effect that isn't shown for minutes on end) to be an animation.
added on the 2020-02-19 08:30:23 by britelite britelite
Last but not least, Photon's next post (if he doesn't find the wisdom to crawl back under his rock instead) more likely than not will bluntly ignore the obviously valid points you just raised and instead pointlessly reiterate his own
added on the 2020-02-19 10:09:56 by havoc havoc
So here's a challenge for you, Photon: release a demo

Game set match.
added on the 2020-02-19 10:24:43 by Gargaj Gargaj
added on the 2020-02-19 13:41:22 by Sir Sir
or make it yourself easy and release a demo riddled with animations!