Test: Crysis

[Heartborne]

So, I tested Crysis out on my system just to see how it would go.

I used the in-game "optimal settings" button to configure the settings according to my hardware and it chose "high" settings. All right. It set the resolution to 1024x768, which I can live with.

I didn't benchmark it, but I played the game through the first level and I found the following:
the game ran smoothly.
the only stuttering occured when crossing checkpoints and the game auto-saved.
No screen tearing or bad renders.

All in all, I'd say Crysis' reputation as a system killer precedes it - my mid-range gaming machine handles it quite well and it looks good. Of course I didn't crank up the AA, but if you've seen this game... it doesn't really need it.

[jlreich]

1024x768 is nothing. That's the standard resolution of an old 17" CRT! Try running it at least 1280x1024 and you will find things start slowing down dramatically. Then try it at the common widescreen resolutions of 1440x900 and 1680x1050. ;)

Also wait until you get into the jungle in the daylight.

1024x768 or the widescreen equivalent would look absolutely horrible on my 22" widescreen. I might as well be playing on an Nintendo 64. :p

[George Hallam]

Then try it at the common widescreen resolutions of 1440x900 and 1680x1050.

my system crys at that :D

i did (using XP)

all settings high
2 x AA (nvidia override)
16 x AF (nvidia override)
1680 x 1050

And i get 25 FPS which is just playable (drops to <18 happened though :() , but after a while you get used to it :p

But in all honesty i thought crysis was a bit of a cr@p game

Ok the single player (only about 9 hours) was great, but thats it. The online was really bad and once you have played the game once and finally got over seeing what breaks when you shoot it it got boring.

COD 4 = 600 Hours
BF 2142 = 300 Hours
Crysis = 12 Hours :eek:


hopefully when far cry 2 is released it has a far better multilayer side to it

[Heartborne]

again, I don't see a need for those high settings. That's why everyone complains that Crysis is such a resource hog - the lower settings are still more demanding and more visually impressive than any other available game on much higher settings! That's the whole point of the crytek engine!

The game does not run well on higher settings because it neither needs to be played that high nor has any hardware out there that supports it yet! In other words, gamers are treating Crysis like it's any other game when in reality it is built to look amazing even on low settings and the higher settings only exist for future scalability.

In conclusion: gamers are greedy.

[jlreich]

OK, whatever! :p

You bet I am greedy. I want to play every game I have at my native resolution of 1680x1050 at max settings! And why shouldn't I want the best visual experience? What's wrong with that? :confused:

Can I? No, I am not rich enough.

With most of my other games I can indeed do this. But not with Crysis. That's why it is called a system killer. With Crysis I have to back off the resolution and eye candy. If I don't I get a slide show.

If you are happy playing it at 1024x768 that's fine. But sorry that's way to low for my 22" widescreen. 1024x768 was a common resolution what? Five years ago? Come on! Five years ago is ancient history in computer terms.

Sorry it doesn't look near as good at lower resolutions as it does at my native resolution.

[Heartborne]

It's apples and oranges to me. Maybe I'm not "hardcore" enough, but I'm going to go ahead and stick to my guns on this one. Crysis doesn't NEED to be played in high resolutions or on high settings to look good, and the ceiling is blown off the settings as a gimmick which seems to be working exceptionally well.

In any case, when the next generation of video cards rolls out you will be able to play crysis in much higher resolutions, but it's not going to happen until then.

I, for one, am happy to be able to play a game and have it look awesome at a lower resolution and on "high" settings instead of "ultra high" or whatever. But that's just me.

Anyway, Crysis already looks amazing and the very idea that it can look even better is pretty exciting. The game was obviously built with the idea of scaling up in the future, so it will probably be en vogue for a few years yet. Bravo for Crytek!

I think a lot of techies think too much about numbers and not enough about what's right in front of them. Who cares if it's 1024x768 rather than 1680x1050? Do you notice the difference while playing the game? Hell no, you don't! That's because Crysis looks better at 1024x768 than most games do at 1680x1050. Besides, a game can only look so good as it's developed to look. You can crank a game up to some insane resolution, but it hits a ceiling at some point or another and most of the time... overkill!

[jlreich]

Who cares if it's 1024x768 rather than 1680x1050? Do you notice the difference while playing the game? Hell no, you don't! That's because Crysis looks better at 1024x768 than most games do at 1680x1050.!
Hell yes I do notice a difference. First any LCD looks best at its native resolution. Period. Anything less takes away from how good it looks. There is nothing the Crytek engine can do about that.

Second, I could care less about the numbers. When I am playing a game what matters is it looks good without getting too much lag. I am happy with 25-30 FPS in Crysis. I prefer to have above 40 for smoothness, but I don't expect to get that with this game and my current hardware. But you get much below 25 and it gets pretty bad and not as fun to play.

When I got Crysis I still using a 19" standard aspect LCD with a native resolution of 1280x1024. I could play at that native resolution on high settings with 8x AA and get about as good FPS and I am now with running at 1440x900 with 4x AA on my 22" wide. 18-25 FPS. That's not that great. To keep it in the 25-30 range I have to drop the res lower and it does make a difference in the quality of the picture. I don't care what you say it does.

On my 19" LCD I could play Oblivion totally maxed out and get around 50 FPS outside with a minimum of 40. When I bought the 22" I had to back off things a little to keep those FPS. I knew this was going to happen when I ordered the new 22". But with Oblivion backing off a little but keeping native resolution didn't take all that much away from the looks of the game while still keeping FPS up to par.

When you have a game that is really above your hardware, and honestly it is above everyones hardware, you try to find a balance between eye candy and FPS. With Crysis this is hard to do. It's usually one way too far or the other.

[Heartborne]

I guess we'll have to agree to disagree. I'd rather keep it above 60 fps, even if that means Crysis has to look only "****ing amazing" rather than "better than anything anyone has ever looked at ever." :)

[Heartborne]

Okay, so I went ahead and cranked Crysis up to 1280x960 with all settings on high and 8x AA.

There were a few minor stutters, and after an hour or so of play some textures in the distance were flickering. Also, some of the cut scenes suffered from the kung fu movie syndrome where the voices were not in sync with the characters' movement.

Otherwise, however, the game played just fine. Again, I didn't benchmark, but I am sure I had over 40 fps most of the time, but that it dipped below once in a while.

What I do find particularly remarkable is that this game is hard! The enemies are almost superhuman in their ability to avoid being shot. I can't believe how hard it is to kill these guys! It's like bullets sail through them more often than not... maybe the game's physics are lousy, or maybe I just suck at fps... but even when using the weapon's sights I only got clear headshots at point blank range and only once in a while.

[jlreich]

Don't forget you have a 9800GTX. Until just this week it was the best single GPU card out there. Considering that those results are not very good. ;) The game is demanding. Even on the best hardware.

[Heartborne]

Very true, JL. I was a little disappointed to discover the the GTX 280 costs more than a 9800 GX2 but doesn't match it in performance.

Of course, the dual-GPU solutions that are forthcoming will blow away anything we've seen thus far... and the HD 4000 series has yet to launch.

Yes, the game is demanding, but still playable at high settings with modern hardware.

[Heartborne]

Don't forget you have a 9800GTX. Until just this week it was the best single GPU card out there. Considering that those results are not very good. ;) The game is demanding. Even on the best hardware.

You forgot about the 8800 Ultra. That pretty much owns the 9800 GTX.

[Heartborne]

Would any of those of you who have played this game please give me some advice as to how I manage to not die constantly?

I find Crysis to be very frustrating. The near-invincible Koreans shoot me to pieces before I have a chance to react pretty often. And this is in EASY mode. After about an hour of trying to get into the NK base in the third level I got fed up. I must have died about 30 times.

I tried using strength mode to hold my weapon steady, but then I got killed faster. In defense mode I can't kill anyone because my weapon won't hold steady at all. Laying prostrate makes me a sitting duck and standing up makes me a big, shiny target.

Please help me with my lack of fps skills!

[Heartborne]

no one's following this thread, and that makes me sad... but I will have you know that I have finally got the hang of Crysis. I can see why a lot of folks don't like it... I enjoy the parts where I actually get to shoot people, but it's mostly just cloak, run, hide, cloak, run, hide, cloak, run, hide, cloak, run, hide... then maybe if you're really lucky once in a very great while there will be few enough enemies that you can ACTUALLY engage them.

So yes, the fact that there is constantly pressure on you is cool and makes the game feel very action packed... but the fact that you spend more time running and hiding than fighting does make the game get kind of boring.

[saphalline]

Would any of those of you who have played this game please give me some advice as to how I manage to not die constantly?They're called cheats (http://www.gamefaqs.com/computer/doswin/code/931665.html)! ;)

Single-player FPS games usually suck in terms of difficulty. The enemies have super-human abilities granted to them by the game engine! I mean, think about it - you've got a Q6600 CPU, right? That's about 100 million transistors of pure logic, spread over 4 separate cores, operating at 2.4GHz (stock) that are probably 70% dedicated to AI trying to kill you!! :eek: Of course you're going to die a lot! Most gamers do. All that power drives the AI to have superior aim, group tactics, the ability to track in underbrush, an inside look at collision detection so that they know when to fire at you and when not to, etc! It's an unfair fight for all but the most talented, hands down!

That's why I like to use a few cheats every now and then. It doesn't have to be over the top, either. (I used to do that years ago but I've since gotten better at FPS's! :p) You don't need God mode AND infinite ammo AND infinite suit power AND the ability to jump over a mountain! Just one or two cheats used sparingly can get you past the parts in the game that are giving you trouble. Although, if I buy a shooter just for the multiplayer aspect and I want to go through the single-player campaign just to enjoy the story, then sometimes I'll cheat like a mad man so I can "fast forward" through the game. Otherwise, cheating doesn't have to mean "Superman vs the geriatric ward". Every now and then, there's just no other way for a player of your skill level to get past a choke-point without 23 grenades! ;) And that's fine.

Multiplayer online is another story. It's one thing to cheat against the unfair AI in a single-player environment. It's another thing to cheat against other humans who have some sort of expectation of a level playing field! Online cheating = NO!

[George Hallam]

[QUOTE=Heartborne;397069but I will have you know that I have finally got the hang of Crysis. I can see why a lot of folks don't like it... I enjoy the parts where I actually get to shoot people, but it's mostly just cloak, run, hide, cloak, run, hide, cloak, run, hide, cloak, run, hide... then maybe if you're really lucky once in a very great while there will be few enough enemies that you can ACTUALLY engage them.
[/QUOTE]


haha that is so true i spend my whole time in a bush cloaked with a sniper and shooting people from miles away then just walking into the base and killing the few remaining people. I played it in Veteran mode (super hard) and if u meet 3 people with machine guns at the same time u just die two shots and you die :eek: yet it take like 15 to kill them :(..

But the game in my eyes was to slow paced you literally cant run into there base and let hell loose because you just die :rolleyes: So its just the slow paced, hide behind rocks and wait until your enemy's caution level have gone back down to green then go and sniper another and wait for the whole process to start again :p

[Heartborne]

I didn't even know about the caution level. I figured that little meter meant SOMETHING.

The in-game tutorial blows.

Anyway, I lowered the settings to 4xAA so that it wouldn't stutter so much. I was getting about 20 fps in the benchmarks with 8xAA. The game plays much better now and it looks exactly the same. :)

[jlreich]

If the game is good I will play it through without cheats. Then for further fun I will use cheats to just go on a blood fest. If I am getting rather board with the game I will use cheats to make it worth playing. Doom3 was one of those games that was horrible until I used cheats. Then it was actually fun for awhile. What a waste of $30 that game was.

[Heartborne]

I LOVED Doom 3! I started a thread about that. It's a very fun game, imo. It's just not a sophisticated fps. After all, it's "Doom 3," not "an entirely different game that isn't Doom." It's derivative of classic fps gameplay, and why shouldn't it be?

I do take issue though with the fact that the maps aren't as good as those found in the 16-bit Doom games. That was a bit disappointing, but since the "holy ****!" factor is through the roof, complex maps would have just made the game too intense.

[jlreich]

The atmosphere was really good. But it was soooooo boring. One demon after another. Nothing more. A very poor plot line barely holding it together by a thread. I didn't even get out of mars city before I lost interest. If it wasn't for cheats I would have never bothered to play it again.

[Heartborne]

To each his own, I suppose.

I benchmarked Crysis at an average of about 30 fps, so I lowered it to 2x AA and benchmarked again, where I got around 40 fps most of the time. A friend of mine was saying that with AA you don't notice a large difference between one setting and another; it's more of a yes/no question rather than a how much question. He usually knows what he's talking about, so I took his word for it. He said that AA just renders the image at a higher resolution, then shrinks it down... so there's not point in going more than 2x because the difference is so subtle.

Personally, I see no difference. But then I'm using a CRT monitor. With LCD monitors it may be a completely different story.

[jlreich]

Anti-Aliasing (http://en.wikipedia.org/wiki/Anti_alias) is used to smooth out square edges of pixels. By adding some gray to the edges of the a curved object it makes it appear to the brain that it is smoother.

AA is nice in gaming because it cleans up all the jagged edges. Instead of the graphics looking like they were rendered in the original Doom engine, they now look nice and clean. The more AA the better. :)

CRT's are better monitors in this way. They have a much better resolution capability, more accurate color, and response time is not even a concern since it is measured in ns instead of ms. But most people opt for the lighter, less bulky, and less power consuming LCD's these days. And they are getting better and better at the things they lack in every day.

The last CRT I used was a 19". The thing was a monster! I think I have a hernia from lifting it. :p

[Heartborne]

Well, my friend being quite the "know it all" will be shocked when I tell him that he is totally wrong about what AA really is.

Of course, since it doesn't really make a difference to me I will go ahead and sacrifice a bit of smoothness for the better gameplay.

[saphalline]

A friend of mine was saying that with AA you don't notice a large difference between one setting and another; it's more of a yes/no question rather than a how much question. [...] He said that AA just renders the image at a higher resolution, then shrinks it down... so there's not point in going more than 2x because the difference is so subtle.Actually, that is how AA can work. But not anymore.

That type of AA is called "super-sampling" and was the first! It's a rather brute force method that sucks up the resources, though, so it was dumped in favor of more efficient methods long ago.

"Multi-sampling" was one such more efficient method that tried to combat certain special cases in aliasing, such as a brick building viewed from afar, by rasterizing the image more than once at sub-pixel offsets. This avoids the intense performance hit of SS (which is usually a single 4x image) because each sample is the same size as the original & final image.

After that, things get tricky. For all practical intents and purposes (ie, what showed up in games), fragment-based AA was next. Pioneered by ATI, this was a method of AA built at the hardware level (and managed at the driver level) wherein the raster units themselves were redesigned. This had to be done at some point since higher levels of parallelization were required regardless. Fragment-based methods from both ATI/AMD and NVidia are still used today. This method works by rasterizing the final image in clusters of pixels (similar to tile-based techniques but only at the raster stage). The hardware is able to test each cluster of pixels (granularity from 128 x 128 all the way down to 4x4! :eek:) to determine if AA is even required. If not, no AA is used and performance for that cluster improves. If so, any AA method can be chosen for that cluster. This was more of an efficiency thing but definitely worth noting in any AA discussion!

Later AA methods were derived from grids, and are considered sub-sets of MSAA as far as I know. "Rotated grid" (RGMS) and "ordered grid" (OGMS) and "sparse grid" (SGMS) take the multisampling idea to further extents with various offset patterns and numbers of samples.

Upgrades to the hardware-level fragment-based AA methods were developed into "custom filter" (CFAA) techniques by ATI & NVidia about 2 years ago. New terms like "wide tent" and "narrow tent" were introduced at that time, representing different forms of edge detection. These techniques have rather bad results with lines, however. (Lines confuse the issue of edges.) But this explains why newer vid cards handle AA better than older ones.

Modern AA techniques are almost entirely hybrids. On a good 3D site, you'll see AA tests listed as "SGMS + CFAA, 6 taps" or whatever. They tell you exactly what's going on! It's all rather complex these days.


Technically speaking, your friend was correct at some point in time! ;) Let him know that AA hasn't been idle all these years, however! Like all technology, it marches ever onward...