John Carmack Leaves id Software 154
jones_supa writes "John Carmack has left id Software completely. 'John Carmack, who has become interested in focusing on things other than game development at id, has resigned from the studio,' id's studio director Tim Willits told IGN, and continues: 'John's work on id Tech 5 and the technology for the current development work at id is complete, and his departure will not affect any current projects. We are fortunate to have a brilliant group of programmers at id who worked with John and will carry on id's tradition of making great games with cutting-edge technology. As colleagues of John for many years, we wish him well.' Carmack, a co-founder of id, recently joined Oculus VR as Chief Technology Officer, and at the time remained at id Software in some capacity. Earlier this year, id president Todd Hollenshead departed id as well."
Oculus Rift! (Score:3, Interesting)
Re:The end of an era. (Score:5, Interesting)
Armadillo Aerospace lost the race for the SpaceX prize. It didn't develop any compelling intellectual property that set it much apart from the other commercial offerings in space travel, so it's become an also-ran. There are no plans for it to do much of anything unless another tycoon comes along and injects vast sums of cash. Carmack is done floating it with his own personal wealth.
His new passion is Oculus Rift. He brings great momentum to that project.
His presence at iD and Oculus probably became strained due to Oculus wanting to be platform & engine independent, while iD would obviously want priority compatibility built into Oculus for their engine.
Re:The end of an era. (Score:5, Interesting)
Armadillo was a profitable company (as in, showed a profit at the end of a fiscal year) with several different research contracts for NASA before it imploded due to mismanagement. I'm not sure what your ambiguous comment about "SpaceX Prize" means, do you mean the Lunar X prize? Armadillo never made a bid for the commercial crew program as far as I'm aware (where SpaceX is competing with two other, non-Armadillo affiliated companies).
Re:The end of an era. (Score:4, Interesting)
I own a dev kit and it's far from figured out. Right now it's just a motion sickness machine.
Totally understandable! (Score:5, Interesting)
Buying iD was a massive mistake (Score:3, Interesting)
Who didn't see this coming? Carmack has been the BIGGEST problem at iD for almost too may years to count now, and Bethesda has a near impossible task in recovering anything profitable from its purchase of the company.
-Carmack was the genius behind Doom, a truly staggering achievement, but bested by far a few years later by the 'Build' engine created by Ken Silverman and used for Duke Nukem.
-Carmack and Abrash (the second name is ignored and unknown by too many of you) created the Quake engine, a project that most of you don't even know was designed to handle SOFTWARE rendering of 3D games with great efficiency on Intel's emerging PENTIUM family of processors. iD was taking advantage of the vastly faster FPU on the new CPUs, especially the floating-point divide unit required to do decent quality approximate perspective correct texture mapping.
Unfortunately, even before Quake came out, it was out-of-date. 3D hardware had started to appear for the PC, and this hardware SOLVED the perspective-correct mapping issue with an efficiency (and quality) that no x86 assemble-code could hope to match. It is notable that years later, when the Unreal games were still supporting a software-only render path, the high-quality render they used was NOT licensed from iD.
-Carmack never really recovered from a world where hardware was doing all the heavy lifting, and spent the following years seeking algorithms that he could design and implement on the CPU side. At first, this idiotic obsession didn't harm iD, as various versions of the engines behind Quake 1/2/3 were licensed for incredible numbers of projects by other companies. HOWEVER, most of this licensing success occurred because the clean code base could be heavily modified to add all the features Carmack never bothered to place in iD games. Tame local developers like Rogue, Ritual, and Raven did the work iD couldn't be bothered to do.
Most notably, Carmack expended a vast effort to add 'smart' curved surfaces to Quake 3, a complete waste of time because such software generated mesh data could not be properly interacted with, and could be better done with ordinary models and LOD by the hardware available at the time. At best, Carmack 'forced' competing engines at the time to waste time and money attempting to implement their versions of this 'feature'. Better by far was the introduction of 'shaders'- a method of forcing the hardware to render far more interesting surfaces (animated effects and the like) than the flat 'pictures' that had dominated 3D games for years before.
-Doom 3 was the final failure, and the proof that the age of iD was over. The brand new engine was a licensing disaster, and the Unreal engine took over the marketplace with a level of success that crushed any iD previously enjoyed. Carmack and co over-saw an unthinkably expensive attempt to resurrect the Wolfenstein success that followed Quake 3, but iD's two parts of that project were so bad, they never saw the light of day, and iD released for free the independently produced third part, Wolfenstein- Enemy Territory, the last genuinely good and popular game that iD would offer.
-And then, of course, we end with Rage, one of the biggest software failures of all time. Carmack was obsessed like never before with a 'problem' that didn't even matter- how to turn the entire texture data set for a game level and all assets into a uniform, general, 2D indexable data structure we know as MEGATEXTURE. Carmack was, by this time, so UTTERLY clueless, he had ZERO knowledge of state-of-the-art in AAA 3D gamine engines, and boasted he preferred to play Nintendo kiddie games with his family.
Megatexture has dozens of problems, but here are the two main ones.
1) The maths behind megatexture means that to have access to VERY poor textures, your data set, even at maximum feasible compression, has to be unthinkable large. Rage had putrid 'close' textures' and dreadful 'far' textures. Only in the mid-ground did megatexture textures look half-decent. If you don't 'bake' the light