It is close though CGI. Between Gears 2 and KZ2. Side by side both games do not look out of place with the other. As you have seen before.

You may think one over the other. But really there isnt much difference at all. And considering Gears 2 from reviewers said AI was the best available and so on, it's no slouch in those areas either. To me these 2 games are above the rest. And whatever people think they are marginal for what looks best one way or the other. Both have no CGI.

Selnor please, GEARS does not touch killzone 2 visually, stop saying it does, and stop showing extreemly low quality video to prove your point, why dont you get on the KILLZONE 2 BETA? The come back and talk, because KILLZONE 2 $hits on UNCHARTED, which in fact looks better than GEARS OF WAR. *Gears 2 has amazing action sequences though, ive seen it through the PS EYE, my friend played the campaign*

why didnt they use 100 percent, what a bunch of half assers. and how many years is killzone 2 in production now? oh yea 4 years or ever since ps3 debuted.

Too much perfomance headroom, the PS3's Cell has 8 processors. Moving legacy code over to the SPEs or writing new far more modernly designed tech from scratch takes time.

God of War 2 wasn't build in a day and neither was Rome.

@ others

I agree Killzone 2 is way beyond Gears 2's league, this can easily be judged without having completed that game.

Also Killzone 2's environments are bigger, also Killzone 2 supports 32 online players while Gears 2 is 10 online players max (2 more than Gears 1).

Wait, what perspective did I change?

lol, i just dont think a game should take like 4 years to develop, you and i know thats like almost half of  a consoles life, what next another 3-4 years for kilzone 3? i understand that pretty games take time but gears of war 2 look pretty and only took like 2 years to makes ever since gears 1 came out 2 years ago, yes gears have less players on multi play but resistance2 have like how many and didnt take 4 years to develop, all i'm saying is that i hope the next killzone dont take as long.

The original God of War took 3 years to develop. It takes time to create amazing games. BTW, God of War 2 only took 2 while taking the game well further technically as well as assets wise.

If aiming for the same quality, a Killzone 3 taking Killzone 2 assets (re-using much) and Killzone 2's game engine a new game could be made much cheaper and take less time to develop. More likely though is that they are going to enhance the game engine further (considering the huge technical headroom) and in addition create new even better looking and sounding assets, this based on lessons learnt from their first PS3 game. A bit like God of War 2 vs God of War 1 on the PS2.

I'm still waiting for a reply here. Not impossible that I may have changed my perspective for some good reason, but at least I'd like to know what you're referring to.

@ NJ5

You made it sound like having headroom was a bad thing, the SPEs just waiting for each other and thus being unable to perform something else in parallel.

Now your view seems more balanced. A lot of the black areas in that old diagram are now filled, you seem to acknowledge being able to bring this down again through optimisation is a good thing, even more headroom for future games to take advantage of.

That would go well with Killzone 2 dev comments, who state they can still add 40% more parallel systems before having to worry about other processes slowing down (thus requiring code optimisations, like 360 launch games using up 85% and once reaching near hundred needing heavy code optimisations like we saw with later games such as Gears 2).

@MikeB: I never said having headroom was a bad thing from the hardware capabilities perspective. It is however something which software developers try to avoid, since every unused cycle is a cycle wasted. The fact that optimization can give even more headroom still remains of course, and obviously optimization will be done if more spare cycles are needed.

When I talked about SPEs waiting for each other I just mentioned it as an indication that multi-core engines have a hard time fully tapping the CPUs due to the inherent difficulties and limitations of making multi-core engines. There will almost always be some wasted CPU cycles in complex multi-threaded applications with interdependent parts.