According to GamesIndustry.biz, Project Natal will launch in a cut-down format, omitting the originally planned onboard processor, which will see its workload hived off to the Xbox 360's CPU.
The story cites developer sources who claim the move from hardware to software serves at least two different purposes. Firstly, in order to keep hardware costs manageable Natal needs to launch cheaply to ensure take-up, and Microsoft will not want another loss-leading product. Secondly, and more intriguingly, the story suggests the move from hardware to software presents Microsoft with the ability to upgrade Natal's capabilities via software, and also allows for greater flexibility developer-side.
First up, let's tackle the credibility of the story itself. As it happens, the latest CES press materials from Microsoft itself seem to obliquely confirm it. Natal's make-up is described as consisting of "RGB camera, depth sensor and multiarray microphone running proprietary software" - so no mention of the internal processing chip at all. This ties in with Digital Foundry's experience of the Natal demos at last year's gamescom where it appeared that a PC was used to process Natal input, while a 360 devkit actually ran the games.
Just what impact this has on performance in the final product remains to be seen. Even in Microsoft's own single-player demos, analysis of video taken during our one-on-one with the system suggests latency is an issue, with anything up to 200ms in overall lag on a mini-game where Natal is processing just one complete human skeleton. This video illustrates the issue clearly.
More recently, leaked video showing purportedly showing Half-Life 2 played via Natal suggested even more of a lagged experience compared to what was seen at gamescom, and there have been several rumblings from developers that the latency, in concert with the lack of any sort of conventional controller, is going to cause issues.
Traveller's Tales director Jon Burton is one of the few to have gone on the record, telling Develop that "[Natal is] exceedingly clever, but the lag on the input and lack of physical buttons is really going to restrict the kind of games that can be done with it."
While GamesIndustry.biz's source talks about a 100ms latency for the controller, presumably this does not factor in the lag inherent in console gameplay itself. Lag tests suggest that the absolute fastest that a 30FPS game can update is 100ms, and the gamescom test suggests that the latency described by GamesIndustry.biz's source is added on top of that. That's for a single player. Factor in the support for up to four players, and you do have to wonder what sort of performance impact this will entail for multiplayer games.
Natal director Kudo Tsunoda is on the record as suggesting that patches for existing titles to enable support for the motion sensor aren't likely to happen. If the Natal software layer does indeed run on the Xenon CPU, this makes sense. Natal will want its own chunk of CPU time and while it is unlikely that all existing games max out the triple core chip, they will definitely not be developed or optimised with Natal's overhead in mind. It's unlikely to be cost-efficient to support Natal on an existing title if major re-engineering is required.
However, this has both its bad and good points. In terms of competition, Microsoft is going to find that a lot of multiplatform titles available on Xbox 360 will get motion control upgrades on the rival PS3 platform (Resident Evil 5, for starters). More support from more games, particularly if you already own them, makes a motion control purchase more appealing.
However, on the plus side, this means that the games we do see on Natal will hopefully be tailored especially for the control system as opposed to being unimaginative afterthoughts - so fingers crossed we won't see the equivalent of the ubiquitous plank-walking that was added to several PS3 games simply in order to tick off Sixaxis motion sensor support. You would think that committing a serious level of CPU time to the control system will be matched by the commitment level in terms of how it is utilised.
It's unlikely that this scaling back of Natal will have any impact on the quality of the final product, so long as the software layer matches or preferably exceeds the performance of what we played at gamescom. This news makes sense in making Natal an affordable upgrade, and an inexpensive addition to existing SKUs: the lower cost is much more attractive than the notion of motion control upgrades to the current library of 360 games which were never designed for use with this hugely innovative system.
Looking forward to the next generation of console, assuming the controller-free interface takes off, integration at the design stage should all but eliminate any latency problems.
In the meanwhile, not much has changed from our initial conclusions having tested Natal at gamescom. Whether the processor is in the camera or working via software within the 360 is close to an irrelevance in the greater scheme of things: the system is so new and so divorced from current control methods that games are going to have to be tailored to its strengths, which are considerable, as well as the weaknesses.
Adding waggle to existing titles met with limited success on Wii - it was the games that used motion control intuitively and innovatively that worked, and so it will be with Natal.