Yeah engines don't really mean much they all do the same thing, render normal maps, colour maps, specular. It really comes down to the art that is created in it and how easy it is to itterate that art. Both these engines run the render engine as the world editor but I would prefer to work with the Cry engine as I think the sandbox editor is easier to use. These are pc engines so I'm wondering if there is a bottleneck exporting to ps3 and xbox360, it's not as simple as just exporting at the end of the day there are gamma issues to deal with on both platforms and the pc has a different gamma.
Are you guys for real? The actual rendering engine plays a big part in making the game look good. You could take beautiful art but if you put it in an engine with a shit-tastic rendering engine it's gonna look like ass. Although no rendering engine in the world is gonna make shit-tastic art look good, it's a balance of both the engine and the art quality.
The Cry Engine looks great, but I would venture to guess that when that game ships it will only be playable on maybe 10% of machines. Also, there is no way it is going to look that good on a 360 if they can even get it to run on one.
Unreal is great for doing closed in environments but it sucks for the kind of thing they are doing with Crysis and the big, open, lush environments.
its all about the lighting model. and i think UE3 lighting model is kinda crap (personal opinon of coarse) it makes every game made in it look like a UE3 game, kinda like doom3. the cry engine has a more realistic lighting model and seems to mimic lighting conditions very naturally. another thing you have to take into consideration is that the cry engine was based and built around dx10 tech and unreal was built on dx9.. i dont know too much about the specifics of either but i do know dx10 lets you pull off more shit in way fewer passes. i am sure epic is constantly upgrading its tech and with a new lighting model and perhaps some dx10 optimization who knows.. we are really close to it being an apple and oranges type thing i think.. in 5 years all engines will look great..
and besides that, its not all about looks, engines do way more crap like physics, crowd simulations, ai.. all of this is being integrated into newer engines, asthetics is only part of the battle.
I guess it always depends what the engine and level editor is built for.
I'm sure the Crytek engine is better for large outdoors, and if you have seen the videos of their level editor, with all the mesh brushes etc. well, it's awesome for the kind of game they're making and it's easier and faster to get better results than with other engines.
Indoors or indoor like levels (small blocked city ala GoW) or UT like MP maps are i think a completely different beast and you're better off with UE3 style CSG tools and a grid, imo.
No you got it all wrong!!! TEH BEST ENGINE EVER IS: SAUERBRATEN, JAWOHL!
Realtime, multiplayer mapediting FTW!
And it is a damn nice oldschool FPS, too.
My only concern is if cryengine is flexible enought to allow games with a different mood than Crysis, it seems that it was built to render perfectly a island, but would it be applied to other styles?
Vitor, there are some tech demo videos of CryEngine2 rendering different environments. I don't see why it wouldn't be able to, it has all of the groundwork in there for pretty much any location or lighting setup you need, it seems.
However I think a lot of the technology it's using is fairly overrated, and mostly the only thing that looks interesting are their shadow maps, and that's pretty much standard DX10 stuff, I think. Certainly all the high-res screenshots you see floating around on the web, and their demo videos, are all running on ridiculously fast PCs and videocards you probably can't even buy yet.
I know that when it was demoed at a German game convention late last year, it was running on very high-end machines, and was generally quite slow and jerky, averaging 15fps. I assume they've streamlined it somewhat since then, but if you consider that the Unreal Engine currently runs everything very smoothly on existing hardware, and without a really huge difference in visuals, I'd probably go for UE3 at the moment.
yes mop, probably that is very true. I only asked as when i tryed the Farcry editor it seems that pretty much of it was just focoused on a tropical island style, and even that the mod community for it was indeed small i didn't saw anyone coming with a more city-like look. But for sure Farcry was a completly different thing.
About hardware, i got some feedback from inside that i'd run it fine, not on the best looking sittings but it would run kinda nice on my current system.
[ QUOTE ]
AZ: while I think the lighting is better in the Crytek engine, you can still pull off a very nice outdoor scene using the U3 engine. Example:
[/ QUOTE ]
While that is a very nice looking scene, and while it is outdoors, it isn't really what I would call a wide open environment. I was speaking more about large, wide-open, mostly flat terrain. Those buildings are being used to occlude what may be behind them or possibly to give the illusion that there is something behind them when they may very well simply be facades that the player never gets close to.
It does look very nice and works well, though. In fact, I would say that it is a great example of the kind of thing Unreal is great for. What game is that from?
Personally I still think Doom 3 is the best looking Next-Gen game released thus far . I don't know if it is the best engine but im willing to bet ID's new engine will blow them all away I just hope you guys don't become 3D realms ..I mean Carmack did say right after D3 was released that they would be showing off the new game shortly and well 2 years later ...lol
[ QUOTE ]
While that is a very nice looking scene, and while it is outdoors, it isn't really what I would call a wide open environment. I was speaking more about large, wide-open, mostly flat terrain. Those buildings are being used to occlude what may be behind them or possibly to give the illusion that there is something behind them when they may very well simply be facades that the player never gets close to.
It does look very nice and works well, though. In fact, I would say that it is a great example of the kind of thing Unreal is great for. What game is that from?
[/ QUOTE ]
Huxley, a FPSMMO on the U3 engine. I see what you're saying about the open lands though. I'm sure that programmers could come up with workarounds and illusions to make the U3 engine render beautiful landscapes, but just like Farcry, there seems to be no performance hit from looking at a building 10 feet in front of you and then turning around to see a whole row of islands.
As mentioned earlier, it all depends on what you want the engine to do. You can write an engine that does one thing superbly well, yet fails in other areas.
I know a place developing a FPS at the moment using a great engine with superb lighting model, but the streaming they are getting is poor, since the engine was designed around portal based interiors.
I have watched just about every one of the videos for the cry engine and the UE3 and the cry engine looks like it has more versatility than the UE3. The main thing that it can do that would save months in dev time is the way its editor works. The what you see is what you play, the auto sectoring, the lighting, the ability to make new/variant materials in editor from existing assests... OMG god would that save time. So many things that needed stream lining that are in that engine puts it way over what UE3 has.
Yes the cry engine can do citys and inside areas, its not just a island making engine. Its just that it can make islands very fast with less work than citys.
hmmm ... the UE3 editor does a pretty good job of "what you see is what you play" ... post processing, particle effects ... not really sure you've seen the same UE3 that I have :-/
I really like the lighting engine Cry have got going, nice soft shadows too though I wonder how much they can retain past the demo stage.
Their foliage stuff is really nice too but theres some pretty big open levels in UT and theres a comparable what you see is what you get / ability to use old shaders to make new ones and a similar set of features between both engines that people are mistakenly thinking is the domain of one side of the other.
I think the real difference between the engines, for those making games atm, is that Cry runs on vista and UE3 runs on 360,PS3,XP so choosing cry means you limit your target audience by a few million as people are going to be slow to pick up vista, as slow as they were with XP.
I would say from the videos of Cryengine 2 that I've seen, the Cryengine seems much less in-your-face about just dropping down into a level to explore the layout you just spent several hours creating.
In UE3, after everything is placed, you get to build lighting. You even have to rebuild the map to update the positions or cutouts you've made to BSP objects. It just seems like that stuff should happen on the fly, not after some sort of "build" process.
both engines have their own strengths. If crytek can market their tools as much as Epic does, then hopefully we'll see a rise in games using their engine.
[ QUOTE ]
Although no rendering engine in the world is gonna make shit-tastic art look good, it's a balance of both the engine and the art quality.
[/ QUOTE ]
Quoted for truth. It is a balance. That said, imo If you're smart about what you do and how you use your resources, and depending on the type of game, it *is* possible to make a very good looking game with somewhat average tech, but it's *much* harder to make a very good looking game from very good tech but shitty art. Bethesda pulled off a graphically critically acclaimed game in Oblivion with Gamebryo, largely considered to be an average 3D engine (although its noteworthy that Oblivions performance is less than stellar).
When I was At EA we always managed to make good looking games, but we were always *way* behind the technology curve in terms of 3D engines. In a way, the poor tech dictacted the kind of games we could make. EA's action games like the Bond and MOH stuff ended up being these formulaic affairs that lead the player down a strictly pre-determined path, haunted house ride style. And I think that was partly due to the Hollywood folks coming into the co. but also I think that those are the only kind of games you can really make with a mediocre engine because of the smoke and mirror shit you can pull off. You just cant do more open world, ambitious stuff with a poor engine.
There is no 'best engine'. It's been said, but it depends on the kind of game you're making. Brushes, bsp, streaming, heightmapped terrain, there are all sorts of things that you might need an engine to be best at specific to your game. There really isn't a one size fits all yet.
Cryengine is looking amazing though. In game editing has to be the way forward. Baked in lighting really needs to just DIE a horrible death now as far as I'm concerned. If you can rip out the whole shadow map bake procedure and have that shit be truly dynamic, the saving in time and effort and ease of iteration is collosall. Plus of course you get dynamic night & day cycles for free. I wish I could show you guys the stuff we're working with but alas......
I think the real difference between the engines, for those making games atm, is that Cry runs on vista and UE3 runs on 360,PS3,XP so choosing cry means you limit your target audience by a few million as people are going to be slow to pick up vista, as slow as they were with XP.
[/ QUOTE ]
I am pretty sure that Crytech said they also have a XP compatible DX9 renderer for Crysis. They just don't pimp it as much since it looks not quite as good.
You are right about the consoles though, but as far as the actual game goes though, I would consider that a good thing.
Ouch, baked lighting's not going to die until someone figures out how to do ambient occlusion in real time at a reasonable fps cost on the environment. Or do you guys like the lack of shadow in shadow. Everyone jumped forward to flickering real time depth maps and seems to not care about visual quality, real time doesn't mean it looks better it's just a stamp you put on your game now to try to sell it. And yes it would save a lot of time at a huge loss of quality in my opinion. Crysis claims to have real time ao so we'll have to wait and see if it's usable. Bake times aren't really the bottleneck for me any more, it's the second uv set for the lightmap and keeping that up to date when the level designer changes geo and clobbers your unwrap.
To clarify, I didn't predict it happening anytime soon I just think it's a very smart goal.
If your meshes are high enough resolution and built in the right way, you can do your AO bakes on the vertices.
But furthermore, I hate no shadows within shadows as much as the next guy, and in fact If you look at past threads about Cryengine, I've pointed that stuff out. Cryengine doesn't appear to have any as far as I can see.
The no shadows in a shadow is a strange thing - people in the shadow of a building don't cast shadows, but without them they just look a bit wrong. They do occlude the pavement a little bit when they walk, but it's such a subtle thing. However, that image seems to have no shadow underneath the vehicle, which SHOULD be present, even if the vehicle itself is in shadow. All the vehicles in IV have this extra ambient shadow.
One thing that becomes MAJOR in engine lighting and will end up defining how complex you can get with things like shadows within shadows is if you have fixed time-of-day or not in your game. Having a daylight cycle makes things a ton more complicated with lighting and skies. I know Rick and Daz (probably) are familiar with this working on their current games. The Cryengine games have always been built around variable time-of-day, and I'm not sure how well UE3 is set up for this. My guess is that it's built more for fixed time of day, but handles variable time of day alright.
But, really, comparing engines gets hugely complex because there are so many factors at play. Many of these have already been mentioned. But they include:
- what style of gameplay are you going for
- platform compatibility
- tools quality
- your minimum spec (for PC)
- how much money you have to spend on the engine
- physics
- AI
- Character Animation
- Vehicle systems
- seamless world streaming
- network code (and server capabilities within)
- licensor support
- code clarity, commenting and documentation
- your team and their familiarity with certain tech
- how many programmers you have available to work on extending the engine
Sounds cool Rick. Personally I dunno why it would be so hard to add a little cheap trick even soft blob shadow under objects and people that are within shadowed areas.
That's true doc. If you're developing a game that requires dynamic night & day cycles, it forces you to question the accepted norm of baking the crap into everything. You just can't do it, or not in the same way at least.
Also, the perception of fancy graphics in the eye of joe public is an interesting thing and you might not necessarily need everything in an engine that you think you do. People marvel at Oblivions exteriors and interiors, but if you actually look at it, other than a couple of cheap hacks for tree canopies and characters, there simply is *no* shadowing system at all in Oblivion.
Yes the public doesn't seem to care about the shadow in shadow problem. Heck, gears of war didn't even have shadows in the environment and it's rated top visuals ever. I should also clarify that when I say shadow in shadow you don't really want a secondary directional shadow you want occlusion. Like Rick said people don't cast shadow in shadow they occlude the ground below their feet. I remember your exact post about the floating car which is why I was taken back when you said real time lights own all. I don't agree with the high poly vertex baking of ao, it simply doesn't work. I've investigated it so many times with meshes so high res you can't see the spacing between the verts. The baricentric blending vertex colours offer does not come anywhere close to the bilinear blending light maps offer. There is really no reason you can't bake black and white ao maps to the whole world and still have a dynamic time of day, it would actually look brilliant.
Well, really I was making the point that the general public doesn't even notice that Oblivion has literally no world shadows AT ALL. As for 'shadows within shadows', of course we're talking about occlusion here, not directional. I assumed that went without saying.
Never did I say anything about 'realtime lighting owning all' though. Clearly in the looks and performance dept. baked lightmaps 'own'. I was saying that a worthy goal in lighting for the future would be to have everything dynamic.
I would have to disagree that vertex AO baking 'simply doesn't work'. It works to a degree. It worked just about fine for every building in Oblivion. But it isn't ever gonna look as good as AO textures true (assuming you can get decent resolution maps which is questionable in a large open world game). But anyway, I'm fine with baking AO as lightmaps. But shadowmaps are starting to feel dated and limiting to me, but again, it all depends on what type of game you're developing and I think we're looking at it from different perspectives. Baking shadow maps is fine for many, many kinds of games, but as doc says, start messing around seriously with daylight cycles and huge open world stuff, and it's a new set of problems.
Unreal for anyone with a hope of running a game on most comps, and cryengine for people with machines that have parts that were made within the last year
Daz I'm working on an open world game right now. Not sure how the scale compares to GTA4 as we're yet to see any actual footage of the size of the new liberty city. There are all types of nifty tricks you can do to store the ao maps at high resolutions. Ao doesn't need to be as high res as directional shadows as you won't see stair stepping in ao due to its soft gradient quality. Ao doesn't need to have colour either as it is typically multiplied on to the ambient light. In theory you could store one ao map in each channel of a .dxt1 texture which is more than enough room to bake ao at a good resolution onto any streaming world. I'm not arguing the point that working with real time lights is awesome, it will make my job a snap when I can just tweak the lighting in real time, I was just commenting on the fact a lot of engines and artists are ignoring good lighting practices and techniques because they want the instant real time feedback. And I think this thread has become significantly more interesting now that we are discussing lighting in render engines specifically.
Both are awsome, I dont think people would have been so excited about UE3 if theyre tech demo was just bushes and water. The art in gears is definatley easier to impress with. But ofcourse UE3 will be on more platforms, used in more games and therefore accessable to more people. So like people said, "best" is a subjective term because a engine performs a number of functions, and what functions are important to each individual must determine their choice.
[ QUOTE ]
In theory you could store one ao map in each channel of a .dxt1 texture
[/ QUOTE ] interesting.
Asmuel, I'm not so sure you can rely on GDC tech demos for a realistic indication of an engines true abilities quite honestly. Whilst the Art is unquestionably amazing, clearly the technology in Gears Of War falls significantly short of that '04 UE3 video. Either the transition from pc to 360 lost some horsepower, or more likely the reality of actual game code running in the background and having to fit everything into memory meant some bells & whistles had to be cut. (and that's just an observation rather than a negative comment about UE3. It's obviously a very capable engine). But that's pretty much why I'll believe half that stuff in Cryengine only when I see it in a *shipped* product (I did see it at E3 but it ran like a vwcamper van and on god only knows what hardware).
[ QUOTE ]
Either the transition from pc to 360 lost some horsepower, or more likely the reality of actual game code running in the background and having to fit everything into memory meant some bells & whistles had to be cut. (and that's just an observation rather than a negative comment about UE3. It's obviously a very capable engine). But that's pretty much why I'll believe half that stuff in Cryengine only when I see it in a shipped product.
[/ QUOTE ]
and thats pretty much it right there. both companies are selling their engines commercially so it would be in their best interests to show everything that the engines *can* do than what they will ultimately will do at runtime on a target system.
I think it is still too early to tell with both of these. While UE3 is dx9 based for now and Cry is dx10 with all the latest bells and whistles.
I think it mainly comes down to the type of game you are making, how everything is coded, hardware it is running on if it's a pc based title, and how many souls were sold to get the game to run good.
I also sat in on the Crysis presentation at GDC and saw all the neat little things they were doing and how they did them. I have UE3 right now and it's been fun to play with and see what I can do with it. Guess I will make my decision when I get my hands on Crysis.
Oh yea I agree Daz, thats why I posted both tech demos rather than gears ingame footage on 360 vs the Crysis tech demo. Both demo's are probably better than games will look on either engine, and both were running on a similar platform so its about as fair of a comparison as we can expect until Crysis comes out.
I just checked out the Crytek 2k7 trailer. Wow, that is some beautiful crap.
You know, ever since Half-Life 2 was released, some realistic looking games make me motion sick. I would have to go lay down for about 10 minutes every 30 minutes or so when I played through HL2. I also got motion sickness from playing Doom3, Prey, FEAR, and SIN episodes. Surprisingly, a ton of people at work got motion sickness from Gears of War, but I didn't. It seems to me that the motion blur/depth of field would make me nauseas for sure, along with the realistic foliage, etc.
People keep saying its dx9 vrs dx10, does anyone actually know of a single feature dx10 has over dx9? It seems like dx10 has become a huge buzzword and people dont really understand it. From what i remember the biggest plus of dx10 is that theres finally strict standards graphics cards have to meet to be dx10, ie all cards need to support the same feature set.... I really dont think theres anything amazing you can do in dx10 you cant in dx9, or opengl, or whatever you're using. Its just a way for MS to force vista down your throats.
Looks like theres a couple cool things there, i dont know enough about it technically to really say tho, it would be nice to get some opinions on it from some of the more technical guys on here. To me it really just seems like a bullshit standard they're using to force everyone to buy an OS that nobody wants or needs.
There are additional features in dx10 but it's mostly just a stricter standard for hardware vendors to comply with. No legacy hw support, everything is shaders, etc.
But the general marketing pitch from developers seems to be "Hey our engine uses dx10! We are hot shit and our graphics are better." which I've never really understood. So you're using a new driver specification with random new features, are you doing anything new with them?
Same goes for shader models. Companies claim to have Shader Model 3.0 (tm)(r)(c) support like it's a magical ring you put on your engine for +7 graphics. It's a language specification, it doesn't mean anything to the consumer unless you can say "I needed feature X to pull off realtime AO" and even so I'd be much more impressed with the AO than the random implementation details.
Replies
main is work of artists, and not teh power lol
The Cry Engine looks great, but I would venture to guess that when that game ships it will only be playable on maybe 10% of machines. Also, there is no way it is going to look that good on a 360 if they can even get it to run on one.
Unreal is great for doing closed in environments but it sucks for the kind of thing they are doing with Crysis and the big, open, lush environments.
and besides that, its not all about looks, engines do way more crap like physics, crowd simulations, ai.. all of this is being integrated into newer engines, asthetics is only part of the battle.
main is work of artists, and not teh power lol
[/ QUOTE ]
I think this is the most elegant phrase I have ever seen grace the pages of Polycount. Perhaps I should use this as my sig.
I'm sure the Crytek engine is better for large outdoors, and if you have seen the videos of their level editor, with all the mesh brushes etc. well, it's awesome for the kind of game they're making and it's easier and faster to get better results than with other engines.
Indoors or indoor like levels (small blocked city ala GoW) or UT like MP maps are i think a completely different beast and you're better off with UE3 style CSG tools and a grid, imo.
doom 3 = best engine ever, nuff said.
[/ QUOTE ]
No you got it all wrong!!! TEH BEST ENGINE EVER IS:
SAUERBRATEN, JAWOHL!
Realtime, multiplayer mapediting FTW!
And it is a damn nice oldschool FPS, too.
However I think a lot of the technology it's using is fairly overrated, and mostly the only thing that looks interesting are their shadow maps, and that's pretty much standard DX10 stuff, I think. Certainly all the high-res screenshots you see floating around on the web, and their demo videos, are all running on ridiculously fast PCs and videocards you probably can't even buy yet.
I know that when it was demoed at a German game convention late last year, it was running on very high-end machines, and was generally quite slow and jerky, averaging 15fps. I assume they've streamlined it somewhat since then, but if you consider that the Unreal Engine currently runs everything very smoothly on existing hardware, and without a really huge difference in visuals, I'd probably go for UE3 at the moment.
Time will tell though.
About hardware, i got some feedback from inside that i'd run it fine, not on the best looking sittings but it would run kinda nice on my current system.
Correct me if I'm wrong but I've yet to see a UE3 game using that on vegetation.
AZ: while I think the lighting is better in the Crytek engine, you can still pull off a very nice outdoor scene using the U3 engine. Example:
[/ QUOTE ]
While that is a very nice looking scene, and while it is outdoors, it isn't really what I would call a wide open environment. I was speaking more about large, wide-open, mostly flat terrain. Those buildings are being used to occlude what may be behind them or possibly to give the illusion that there is something behind them when they may very well simply be facades that the player never gets close to.
It does look very nice and works well, though. In fact, I would say that it is a great example of the kind of thing Unreal is great for. What game is that from?
doom 3 = best engine ever, nuff said.
[/ QUOTE ]
Personally I still think Doom 3 is the best looking Next-Gen game released thus far . I don't know if it is the best engine but im willing to bet ID's new engine will blow them all away I just hope you guys don't become 3D realms ..I mean Carmack did say right after D3 was released that they would be showing off the new game shortly and well 2 years later ...lol
Quake Wars is teh new game...
[/ QUOTE ]
id isn't even developing it.
Carmack has claimed to be working on a new IP.
While that is a very nice looking scene, and while it is outdoors, it isn't really what I would call a wide open environment. I was speaking more about large, wide-open, mostly flat terrain. Those buildings are being used to occlude what may be behind them or possibly to give the illusion that there is something behind them when they may very well simply be facades that the player never gets close to.
It does look very nice and works well, though. In fact, I would say that it is a great example of the kind of thing Unreal is great for. What game is that from?
[/ QUOTE ]
Huxley, a FPSMMO on the U3 engine. I see what you're saying about the open lands though. I'm sure that programmers could come up with workarounds and illusions to make the U3 engine render beautiful landscapes, but just like Farcry, there seems to be no performance hit from looking at a building 10 feet in front of you and then turning around to see a whole row of islands.
I know a place developing a FPS at the moment using a great engine with superb lighting model, but the streaming they are getting is poor, since the engine was designed around portal based interiors.
Yes the cry engine can do citys and inside areas, its not just a island making engine. Its just that it can make islands very fast with less work than citys.
Their foliage stuff is really nice too but theres some pretty big open levels in UT and theres a comparable what you see is what you get / ability to use old shaders to make new ones and a similar set of features between both engines that people are mistakenly thinking is the domain of one side of the other.
I think the real difference between the engines, for those making games atm, is that Cry runs on vista and UE3 runs on 360,PS3,XP so choosing cry means you limit your target audience by a few million as people are going to be slow to pick up vista, as slow as they were with XP.
r.
In UE3, after everything is placed, you get to build lighting. You even have to rebuild the map to update the positions or cutouts you've made to BSP objects. It just seems like that stuff should happen on the fly, not after some sort of "build" process.
With that said... UE3 material editor, FTW:D
I'll learn any toolset atm I prefer UE3.
Although no rendering engine in the world is gonna make shit-tastic art look good, it's a balance of both the engine and the art quality.
[/ QUOTE ]
Quoted for truth. It is a balance. That said, imo If you're smart about what you do and how you use your resources, and depending on the type of game, it *is* possible to make a very good looking game with somewhat average tech, but it's *much* harder to make a very good looking game from very good tech but shitty art. Bethesda pulled off a graphically critically acclaimed game in Oblivion with Gamebryo, largely considered to be an average 3D engine (although its noteworthy that Oblivions performance is less than stellar).
When I was At EA we always managed to make good looking games, but we were always *way* behind the technology curve in terms of 3D engines. In a way, the poor tech dictacted the kind of games we could make. EA's action games like the Bond and MOH stuff ended up being these formulaic affairs that lead the player down a strictly pre-determined path, haunted house ride style. And I think that was partly due to the Hollywood folks coming into the co. but also I think that those are the only kind of games you can really make with a mediocre engine because of the smoke and mirror shit you can pull off. You just cant do more open world, ambitious stuff with a poor engine.
There is no 'best engine'. It's been said, but it depends on the kind of game you're making. Brushes, bsp, streaming, heightmapped terrain, there are all sorts of things that you might need an engine to be best at specific to your game. There really isn't a one size fits all yet.
Cryengine is looking amazing though. In game editing has to be the way forward. Baked in lighting really needs to just DIE a horrible death now as far as I'm concerned. If you can rip out the whole shadow map bake procedure and have that shit be truly dynamic, the saving in time and effort and ease of iteration is collosall. Plus of course you get dynamic night & day cycles for free. I wish I could show you guys the stuff we're working with but alas......
I think the real difference between the engines, for those making games atm, is that Cry runs on vista and UE3 runs on 360,PS3,XP so choosing cry means you limit your target audience by a few million as people are going to be slow to pick up vista, as slow as they were with XP.
[/ QUOTE ]
I am pretty sure that Crytech said they also have a XP compatible DX9 renderer for Crysis. They just don't pimp it as much since it looks not quite as good.
You are right about the consoles though, but as far as the actual game goes though, I would consider that a good thing.
gold words
If your meshes are high enough resolution and built in the right way, you can do your AO bakes on the vertices.
But furthermore, I hate no shadows within shadows as much as the next guy, and in fact If you look at past threads about Cryengine, I've pointed that stuff out. Cryengine doesn't appear to have any as far as I can see.
But, really, comparing engines gets hugely complex because there are so many factors at play. Many of these have already been mentioned. But they include:
- what style of gameplay are you going for
- platform compatibility
- tools quality
- your minimum spec (for PC)
- how much money you have to spend on the engine
- physics
- AI
- Character Animation
- Vehicle systems
- seamless world streaming
- network code (and server capabilities within)
- licensor support
- code clarity, commenting and documentation
- your team and their familiarity with certain tech
- how many programmers you have available to work on extending the engine
etc.
That's true doc. If you're developing a game that requires dynamic night & day cycles, it forces you to question the accepted norm of baking the crap into everything. You just can't do it, or not in the same way at least.
Also, the perception of fancy graphics in the eye of joe public is an interesting thing and you might not necessarily need everything in an engine that you think you do. People marvel at Oblivions exteriors and interiors, but if you actually look at it, other than a couple of cheap hacks for tree canopies and characters, there simply is *no* shadowing system at all in Oblivion.
Never did I say anything about 'realtime lighting owning all' though. Clearly in the looks and performance dept. baked lightmaps 'own'. I was saying that a worthy goal in lighting for the future would be to have everything dynamic.
I would have to disagree that vertex AO baking 'simply doesn't work'. It works to a degree. It worked just about fine for every building in Oblivion. But it isn't ever gonna look as good as AO textures true (assuming you can get decent resolution maps which is questionable in a large open world game). But anyway, I'm fine with baking AO as lightmaps. But shadowmaps are starting to feel dated and limiting to me, but again, it all depends on what type of game you're developing and I think we're looking at it from different perspectives. Baking shadow maps is fine for many, many kinds of games, but as doc says, start messing around seriously with daylight cycles and huge open world stuff, and it's a new set of problems.
Here are both tech demo's if you guys want to compare:
UE3 tech demo 2004 - http://www.gametrailers.com/gamepage.php?id=2509
Crysis GDC tech demo 2007 - http://www.gametrailers.com/umwatcher.php?id=9091
Both are awsome, I dont think people would have been so excited about UE3 if theyre tech demo was just bushes and water. The art in gears is definatley easier to impress with. But ofcourse UE3 will be on more platforms, used in more games and therefore accessable to more people. So like people said, "best" is a subjective term because a engine performs a number of functions, and what functions are important to each individual must determine their choice.
In theory you could store one ao map in each channel of a .dxt1 texture
[/ QUOTE ] interesting.
Asmuel, I'm not so sure you can rely on GDC tech demos for a realistic indication of an engines true abilities quite honestly. Whilst the Art is unquestionably amazing, clearly the technology in Gears Of War falls significantly short of that '04 UE3 video. Either the transition from pc to 360 lost some horsepower, or more likely the reality of actual game code running in the background and having to fit everything into memory meant some bells & whistles had to be cut. (and that's just an observation rather than a negative comment about UE3. It's obviously a very capable engine). But that's pretty much why I'll believe half that stuff in Cryengine only when I see it in a *shipped* product (I did see it at E3 but it ran like a vwcamper van and on god only knows what hardware).
Either the transition from pc to 360 lost some horsepower, or more likely the reality of actual game code running in the background and having to fit everything into memory meant some bells & whistles had to be cut. (and that's just an observation rather than a negative comment about UE3. It's obviously a very capable engine). But that's pretty much why I'll believe half that stuff in Cryengine only when I see it in a shipped product.
[/ QUOTE ]
and thats pretty much it right there. both companies are selling their engines commercially so it would be in their best interests to show everything that the engines *can* do than what they will ultimately will do at runtime on a target system.
I think it mainly comes down to the type of game you are making, how everything is coded, hardware it is running on if it's a pc based title, and how many souls were sold to get the game to run good.
I also sat in on the Crysis presentation at GDC and saw all the neat little things they were doing and how they did them. I have UE3 right now and it's been fun to play with and see what I can do with it. Guess I will make my decision when I get my hands on Crysis.
Anyone got a copy? lol
You know, ever since Half-Life 2 was released, some realistic looking games make me motion sick. I would have to go lay down for about 10 minutes every 30 minutes or so when I played through HL2. I also got motion sickness from playing Doom3, Prey, FEAR, and SIN episodes. Surprisingly, a ton of people at work got motion sickness from Gears of War, but I didn't. It seems to me that the motion blur/depth of field would make me nauseas for sure, along with the realistic foliage, etc.
Does anyone else experience this?
But the general marketing pitch from developers seems to be "Hey our engine uses dx10! We are hot shit and our graphics are better." which I've never really understood. So you're using a new driver specification with random new features, are you doing anything new with them?
Same goes for shader models. Companies claim to have Shader Model 3.0 (tm)(r)(c) support like it's a magical ring you put on your engine for +7 graphics. It's a language specification, it doesn't mean anything to the consumer unless you can say "I needed feature X to pull off realtime AO" and even so I'd be much more impressed with the AO than the random implementation details.