Just Cause 2 Gameplay Performance and Image Quality

[RavenXXX2]

Very popular game at the moment " I love it " so here is some interesting performance comparisons between the top end GFX cards, IQ is compared as well as cuda water effects which is available to Nvidia cards only.

http://www.hardocp.com/article/2010/05/04/just_cause_2_gameplay_performance_image_quality/1

For gamers looking for the very best performance, the AMD Radeon HD 5870 is a fantastic video card. It will give you higher performance than NVIDIA’s competing GeForce GTX 480, even if it is missing a couple of graphics options. It is very interesting that when all settings are equal, the AMD Radeon HD 5850 and 5870 are faster than the competition in this game. However, they do lack a couple of key image quality features that this game supports.

From a graphics options perspective, the NVIDIA video cards automatically have a leg up, thanks to the Bokeh filter and GPU Water Simulation options. Gamers using AMD's ATI Radeon HD 5000 series video cards will not see these options in the graphics menu, so they will not have a chance to try them out.

Unfortunately, both of those options also come with a measurable performance hit. We believe the image quality increase to be worth the performance cost, so we chose to lower the resolution to 1920x1200 on both the GeForce GTX 480 and GTX 470 video cards so that we could keep them on. At 2560x1600, the game was just not playable even on the GTX 480, even with AA disabled. Using the GTX 480 at 1920x1200, we found 8X AA to give us a satisfactory level of performance and image quality, while we were unable to use AA with the GTX 470 without suffering undesirable performance problems.

In Just Cause 2, GPU Water Simulation and Bokeh filtering make a compelling argument for NVIDIA video cards. While there is no reason that both of these graphical features could not have been developed with compatibility for AMD GPUs, that is simply not how it happened with this game. This is a "TWIMTBP" title, and the developers chose CUDA rather than OpenCL or DirectCompute, which is of course a shame and a marketing sham that screws the end user.

In this comparison, pay close attention to the line created by the intersection of the boat’s hull and the water. On the top half, GPU Water Simulation is enabled, and the surface of the water is uneven with peaks and valleys. On the bottom half, the surface of the water is again smooth, because GPU Water Simulation is disabled.

On the left hand side of the above picture, GPU Water Simulation is enabled, and you can see that the surface of the water looks choppy. Also, the water directly surrounding the player is tilted, as if a wave is incoming from the side. On the right side, GPU Water Simulation is disabled, and the water looks as smooth and level as glass.

 

[colzboppo]

I'd gladly pay a £100-£200 premium for some extra gimmicky and hardly noticable graphics effects.

 

[kylew]

What I'm curious about is that, while the "CUDA Water" is much better looking, using CUDA is most definitely not required, considering Crysis water looks just as good, if not better, than the CUDA water.

I think the funniest thing about Just Cause 2 is that they didn't use PhysX.

Performance numbers are interesting though, HardOCP are slowly becoming a favourite of mine because of the distinct lack of BS that's involved with them.

Though I still don't like how they use slightly different settings for games over different cards.

 

[drunkenmaster]

For me the only real difference in the gpu water simulation is curvy waves and you can ONLY tell the difference on shorelines and boats, and if you're looking for those you're a bit weird tbh, while playing of course.

I'm not actually convinced the gpu simulated water looks better, different, but, neither makes me go, "thats so fake/thats so real that the game is improved for me" . Considering the minimum framerates on the 480gtx with either effect on, and a lot higher min framerate with them off, just like in Metro I simply wouldn't enable them even if I had a 480gtx. Therein lies the problem with these added effects, they aren't mindblowingly high IQ compared to without, and the performance hit hasn't ever been worth it in a single game to date.

Though considering AMD cards are faster than the 480gtx without it, I'd expect them to be faster with it aswell, if only it was done in OpenCL and globally compatible, but I still wouldn't use it(unless those minimums weren't effected as badly on AMD hardware).

The game looks 99.9% as good without those options with higher performance, its a pretty simple choice that I'd make regardless of owning AMD, Nvidia, Intel or if Coca Cola start making GPU's.


Really is a fun game considering the nonsense storyline though, normally very bored with these types of games far too quickly. Its also got AWESOME performance considering the level of detail and the sheer size. I also don't have a clue how loading is so insanely quick, quick travel from one side of the map to the other takes no time at all, its incredibly efficient and well made. Though I do wonder how long loading is on a 360 as opposed to installed on a fast SSD using PC, its still awesome loading compared to basically any other similarly large game.

 

[RavenXXX2]

Though I still don't like how they use slightly different settings for games over different cards.

Yeah they do an apples to apples comparison as well.

 

[drunkenmaster]

What I'm curious about is that, while the "CUDA Water" is much better looking, using CUDA is most definitely not required, considering Crysis water looks just as good, if not better, than the CUDA water.

I think the funniest thing about Just Cause 2 is that they didn't use PhysX.

Performance numbers are interesting though, HardOCP are slowly becoming a favourite of mine because of the distinct lack of BS that's involved with them.

Though I still don't like how they use slightly different settings for games over different cards.

I don't mind the different settings, but I HATE it when they don't include the apples to apples comparison, sometimes they only do the "what we feel is max playable settings" and they won't show the flat comparison of say all the cards on one graph at one setting. They also rarely use a basic bar graph with everything from a 480gtx down to say a 4770 or something, showing the min/average to show the difference at various resolutions, a mix of both is preferable but they often leave out apples to apples, and never do large scale comparisons anymore.

 

[arc@css]

Looking at the screen shots, I wouldn't say there looks to be 'that' much difference.

 

[kylew]

I don't mind the different settings, but I HATE it when they don't include the apples to apples comparison, sometimes they only do the "what we feel is max playable settings" and they won't show the flat comparison of say all the cards on one graph at one setting. They also rarely use a basic bar graph with everything from a 480gtx down to say a 4770 or something, showing the min/average to show the difference at various resolutions, a mix of both is preferable but they often leave out apples to apples, and never do large scale comparisons anymore.

Indeed, they've only started doing the like for like comparisons. In which case, makes it a non issue for me (I hadn't realised before my other post that they'd made like for like comparisons).

 

[Rroff]

To be fair the difference in the water quality is really only apparent - and its quite quite noticeable - when your playing or watching a video... that said after the initial novelty had worn off I'm not sure whether I'd really care either way.

Also to be fair implementing something like this with CUDA would be relatively a doddle, even without nVidia's support which makes it even easier... using open CL or DC would currently put a lot more workload on the developer.

 

[arc@css]

Yes, I expect when playing the game, it would look better. Quite a lot of screen shots tend to look a little rough imo. Very different once you get in to the game.

 

[Lightnix]

Whilst I think directcompute is the probably the correct 'tool for the job' in this context (i.e. mainstream graphics compute effects for interoperability with DirectX), the effects didn't really go missed here. The sea is by far the most boring part of the game anyway. I think if they thought these effects were any way integral to the game play experience they would have implemented them in a vendor neutral language, much like how they used the Havok physics engine rather than using PhysX.

Besides, who needs to travel by sea when you can just go by 'AIIIIIR SHELDON'?

 

[Rroff]

Directcompute is the "correct" tool as far as any for the job here but unfortunatly its lagging behind right now.

 

[abraxax]

I'd gladly pay a £100-£200 premium for some extra gimmicky and hardly noticable graphics effects.

me too

 

[bee]

both those nvidia only options are pretty disappointing really, the biggest exclusive feature this game has by far is 3D, its the best game in 3D and looks incredible

 

[mmj_uk]

I'd gladly pay a £100-£200 premium for some extra gimmicky and hardly noticable graphics effects.

Most people did when 5870 first came out.

 

[Rroff]

ooof lol

 

[Broken Hope]

Even if they used Directcompute they would likely still have to have multiple paths optimised for each graphics architecture, since generic directcompute code doesn't always run that fast on ATI hardware.

 

[rounddodger]

I'd gladly pay a £100-£200 premium for some extra gimmicky and hardly noticable graphics effects.

What you on about? Hardly noticeable? Looks awesome tbh.

 

[Phoenixdancer]

What you on about? Hardly noticeable? Looks awesome tbh.

yea but worth the extra £100 odd for a gpu slower then atis?

 

[drunkenmaster]

To be fair the difference in the water quality is really only apparent - and its quite quite noticeable - when your playing or watching a video... that said after the initial novelty had worn off I'm not sure whether I'd really care either way.

Also to be fair implementing something like this with CUDA would be relatively a doddle, even without nVidia's support which makes it even easier... using open CL or DC would currently put a lot more workload on the developer.

THis is the main thing for me, sometimes better, is arguable, as you said, it looks different, but in game, when playing the game I'm not sure that the Nvidia water is "better" its different, the curvy edges against coast/boats would be barely noticeable, the actual look of the water.. well. For me the both look more than fine for in game water and neither stands out and says "my god thats soo realistic" they are just different, and neither changes the game in any way.

Its not completely flat blue solid water that doesn't move like a decade ago, both are more than good enough to be taken for water. I would say there were plenty more things they could spend time on rather than a second way to do water, but they've got most of the game area pretty well done anyway. Its a truly great overall gameworld thats incredibly impressive for size and efficiency.

I'd of much prefered more people and time spent on the story and variations in mission goals, than a second set of water thats no better or worse, but thats life.

I don't think the water takes anything away (but performance) but I don't think it adds to the feel of the game in even the slightest way either.