Crytek claims that the performance impact of this form of processing is in the region of one per cent. It's a claim that has merit as the notion of 2D plus depth has already been tested out practically via the TriOviz system, which uses the same technique that Crytek is describing, albeit in a paper glasses implementation.
TriOviz announced at E3 that its technology is coming to the new 3DTVs, which should hopefully clear up our main concerns with the existing version: the green/pink paper glasses kill the colour and the 3D effect sometimes falls apart completely. However, in terms of the processing, performance analysis of Batman: Arkham Asylum in its Game of the Year guise bears out the low cost of the solution. First up, here's some Xbox 360 2D versus 3D comparisons from the game.
As you can see, there is virtually no impact at all on overall performance. Frame-rates are virtually constant at 30FPS, and even the level of screen-tear is much the same (the top tear readings are for the 2D version, the bottom for TriOviz).
However, the Unreal Engine 3 technology that powers Arkham Asylum favours Xbox 360 and the game is very enclosed for the most part, with lots of occluders keeping scene complexity down. Virtually all Unreal titles are capped at 30FPS, so is TriOviz simply making use of the additional processing time available when frame-rate is being artificially limited?
The best way to test is to recreate the same tests on the PS3 version of the game, which already drops more frames and tears more often in its 2D guise. In theory in those same big open scenes, we should be able to see a reduced 2D performance and thus get more of an idea of the cost of TriOviz.
Clearly there is more tearing more often in the scene set outside Arkham, but certainly from a performance angle, TriOviz does what it sets out to do - offering some level of 3D support with a barely noticeable hit on frame-rate. While there is obviously some additional overhead to the process, it's not going to affect the overall performance level to anywhere near the degree we've seen on many of the PS3 3D titles we've looked at thus far. It's not real 3D as two bespoke viewpoints (one per eye) are not being generated, leading to "gaps" in the image, but for many games it may well be good enough.
Overall then, it's fair to say that the Xbox 360 is in fairly good shape for providing a good, though not optimal stereo 3D experience. While there remain very real question marks over proper HDMI 1.4 compliance, the fact that there is such a large melting pot of different 3D rendering techniques, and a very real standard in the form of the side-by-side framebuffer, suggests that compatibility with 3DTVs won't be problem.
However, there is a possibility that the 360 will be unable to match the true 720p per eye introduced by HDMI 1.4 and supported on PS3. The move to a 1080p framebuffer to accommodate full resolution seems like a step too far, and questions remain about whether the 1080p side-by-side format is actually fully supported in the HDMI 1.4 spec. The headline specs say it isn't (it's 1080i only), but the supported display timings say it is, and the available 3D displays do work with it.
However, assuming that it remains unutilised on 360, PS3's exact standards compliant 720p HDMI 1.4 implementation gives the Sony platform the advantage in terms of resolution, but how much of an advantage remains to be seen - the performance limitations of current gen architecture have seen that virtually all of the PS3 3D titles released to date (Super Stardust HD being the exception) have performance issues of some description, be they resolution or frame-rate based in nature.
So the overall question remains whether the advantages you receive from stereoscopic 3D is enough to outweigh drawbacks in other areas.
The E3 demo of Gran Turismo 5 is probably the most compelling argument in favour of 3D we've seen. The implementation of the stereo cameras has a direct, positive impact on the quality of the gameplay: it's immersive, pretty and effective (replays are wonderful) but crucially it helps you to play the game better. It's easier to take corners and judge apexes, and discerning distance and depth is simply more natural. It's as clear a demonstration as any of what 3D brings to the table.
From our perspective, so long as the choice of display mode is always there, and that the conventional 2D mode is not compromised by the drawbacks of implementing stereoscopy, 3D can be hugely worthwhile but like any tool it's how well it's implemented that is the primary concern.