At the twilight of our civilization, endless patent and copyright fights over obvious things. A keyboard at the bottom rather than the side of a phone? Obviously this is a precious work of genius that must be protected at all costs. Soon someone will make a new generation of 8K HDTV's and they will patent "the use of a remote control with 8K HDTV's" When will someone see through this horseshit and revoke these stupid patents.

Doom 3 was not a failure from a technology stand-point. It did just what it set out to do. You try writing an engine with full dynamic lighting that runs on a "Geforce 2". That's right, Doom 3 has a render pipeline that is DX7 compatible. If anything Doom 3's only technical foible is that it relied on OpenGL which happened to become a "big mess" around the time Doom 3 was in development. He was lucky that OpenGL even offered shaders at that point and had to write them in ARB because GLSL was so broken it's not even funny. Megatexture is showing up from other engine developers now too. The concept of data hierarchy wont go away simply because most mappers have gotten by better with tiled texturing. Eventually that limitation will be broken down. The only failing there is that John tried too much to eat his own dog-food. Rather than use a combination of standard tiled textures and Megatexture he tried to force everything to be compiled from the Megatexture. The hardware wasn't ready. Doom 3 BFG's source code is a treasure trove of improved rendering concepts afforded by better OpenGL standards. If we had the OpenGL then that we have now, it would have held-off UE3 no doubt. Some informal tests on a few hundred dynamic lights show you can have a scene with the original Doom 3 render at 15fps while BFG will render at 300fps. Easily competitive now and it's not even a deferred renderer.

Since the site is down, you can also checkout the moddb presence: http://www.moddb.com/mods/the-dark-mod

The community has already improved shadow rendering performance by (at least) 1.5x. http://forums.inside3d.com/viewtopic.php?f=9&t=3491&start=45

Sikkmod already has an option to brighten the Ambient light so that shadowed areas have better visibility. No need to wait, get it now. Better Ambient light rendering will surely happen with source modifications.

Please make sure to read some technical documents about BRINK before posting: http://charles.hollemeersch.net/node/58 Otherwise known as Virtual Texturing see Bethesda Press: http://www.bethsoft.com/eng/games/games_brink.html

Thanks for the update! If you get a chance, stop by The Dark Mod forums to discuss your findings. http://forums.thedarkmod.com/

Megatextures still have potential. They just need to be more conservative instead of trying to make everything out of them. Mixing traditional textures and Megatextures gave the game BRINK it's distinctive look and the technique could be extended further. John Carmack was trying too hard to: http://en.wikipedia.org/wiki/Eating_your_own_dog_food A more realistic approach would've been optimal.

Stencil shadows haven't been removed. Just the depth-fail method (which may mean a performance penalty unless Carmack worked some new magic...) Even without the shadows, the per-pixel dynamic lighting is still valuable and new shadow methods can be implemented anyway (especially with so much of the engine tracking light positional effects).

Yes:

Hexen: Edge of Chaos:
http://www.moddb.com/mods/hexen-edge-of-chaos
Classic Doom 3:
http://www.moddb.com/mods/classic-doom-3
The Ruiner:
http://www.moddb.com/mods/ruiner
In Hell:
http://doom3.filefront.com/file/;83607
Shambler's Castle:
http://www.moddb.com/mods/shamblers-castle
Serengrove:
http://www.moddb.com/games/doom-iii/addons/sp-serengrove-tudor-town-of-doom

The Dark Mod, a Doom 3 total conversion which turns the game into something similar to the "Thief" series games, will now be able to optimize and fix render, AI, and physics. This is a day for celebration! www.thedarkmod.com http://www.moddb.com/mods/the-dark-mod

Yes, lower IPC was a risky move but it was a gamble that they had to take with this strategy. The whole idea still makes sense: 1) Specialize on threaded server workloads on the CPU 2) Increase the pipeline stages for clocking head-room at the die shrink nodes (Even though they backed-off a bit, Intel will do this too) 3) Get Application developers to move to GP-GPU 4) Next-gen Fusion 5) Now you have x86 only doing minimal or perfunctory work while the on-die GPU handles the muscle. I don't know why people are applauding Intel for marginal jumps in FP capability and a hand-full of decode tricks when we are on the cusp of a CPU with MOUNTAINS of Floating Point and parallelism. You think that PowerPC had an advantage in media-rich applications with AVX? Think of how crazy good Fusion will be if the plan comes together. That CPU will be a media-rich messiah! Sorry if I don't share your excitement over opening Word documents quicker because of super cool x86 decode tricks. I want a CPU with a next-gen attitude.

Perhaps they are practicing image replacement for Google maps images to censor forbidden areas? They can use those markings for calibration.