Jump to content
Search In
  • More options...
Find results that contain...
Find results in...
EsH

Any point for developers to make engines anymore?

Recommended Posts

I mean seriously, why bother? Even if a developer could construct an engine as complicated and robust as id's, why not just use that engine? It would probably be much cheaper in terms of both time and money. Is this the end of the graphics programmer? Is the new Doom engine going to be the "Windows" of the game programming world (not comparing quality, obviously, but market penetration)?

Personally, as a novice graphics programmer, I feel a bit daunted to say the least. I kind of get the feeling that the days when you would program an engine because it had some feature the other didn't, are over. I mean, just out of the box, most of you mod developers would be able to construct a game in a matter of a month that will look and play an order of magnitude better than Unreal2, which has been in devopment for many years by industry verterans.

So, I kind of get the feeling that the new engine is cementing the new trend, where games won't be constructed by "coding" something, but be based on the same paradigm that we have with highend 3D modeling and animation packages. When you want to make a 3D rendered movie, you don't go out and "write" a rendering software package. You pull one off the shelf, and write plugins for it, and basically work off of that.

This engine seems the same way. It seems completely unreasonable to go about "programming" a game anymore in the traditional sense, if we're going to working on the level of complexity that Carmack has put us on, now. I really thing the days of wide-spread 3d programming are coming to and end...

Share this post


Link to post

There is still a long way to go, DOOM 3 is only the beginning of one generation...

Share this post


Link to post

I would agree with your post if I was one of the types who thinks that the "latest & greatest" is best.

But I'm not.

Why would people code their own engines? Because it's fun. Because it's a learning exercise. Because it's accomplishing something. Because they can.

I am not daunted by new technology. Even if there was an engine that did it's entire game universive on an atomic scale, I'd still play my old NES ROMs. And I'd still code an engine that would run a game I'd want to play. Regardless if it looks better than the most advanced engine or not.

Share this post


Link to post
Lüt said:

Even if there was an engine that did it's entire game universive on an atomic scale, I'd still play my old NES ROMs. And I'd still code an engine that would run a game I'd want to play.

If you had an engine that did its entire universe on an atomic scale, I'm sure you could run all your NES ROMs and even code your own games from within it.

Share this post


Link to post

Obviously this engine is not meant for classic adventure games, strategy games or flight simulators. It's obvious anyone who can't quite code something like this will license a next-gen engine, not because it needs "the latest and greatest" in graphics, but because things like lighting, physics, scripting and the like aren't simple eye-candy features, but definitive steps towards inmersing the player.

Because once you experience this, other games don't quite pull you in. If the game is about you taking the role of a character in a somewhat complex world, it doesn't matter if it's a RPG, FPS o TPS, developers need to make the game not only fun, but inmersing. You have to care about the world, the characters, the plot. Otherwise you'll lose interest in a matter of hours.

I suspect there will be less attempts at coding an all-together new engine, since it's getting too complex these days. People will license technology from id, Epic, LithTech and the like.

Still, if you're planning to make a Starcraft clone you won't go for the Doom3 engine. If you're working on a Tribes clone, a Freespace clone or a Civilization sequel, Doom3 just doesn't works. You have to find a piece of technology that fits your design.


At least there will be far less half-assed attempts at coding new engines. You know about those could-be great games that have floppy code on the back.

Why would people code their own engines? Because it's fun. Because it's a learning exercise. Because it's accomplishing something. Because they can.

Not everyone can. Or have the money/time to do so.

Share this post


Link to post

Zaldron is right, yet still, new engines for the FPS genre could be done - the Doom 3 engine seems capable of making movie quality environments and characters (As technigues for making models and general art improve, we'll get to see something which will look even more real than regular cgi graphics IN the Doom 3 engine), but there's still one issue: Severe limitations to amount of characters on screen.
New engines might not improve much in terms of graphics and general looks (maybe soft shadows and the like), but new engines could improve on the Doom 3 engine in terms of the amount of enemies on-screen.

Share this post


Link to post

New engines might not improve much in terms of graphics and general looks (maybe soft shadows and the like)

The DOOM3 engine produces far from photorealistic renderings...

Share this post


Link to post
Fredrik said:

The DOOM3 engine produces far from photorealistic renderings...

It's a big step forward. But still, we're lacking things like radiosity, extensive object-based light sources, color bleeding, realistic caustics, sub-surface scattering, deformable geometry and oh so much more.

Share this post


Link to post
Fredrik said:

The DOOM3 engine produces far from photorealistic renderings...

What do I know - it looks photorealistic enough to me.

Share this post


Link to post

besides, the entire 3d programming scene can't hinge on one person. what if carmack suddenly feels like getting into flower arrangement instead of programming? life would have to (weep weep) go on.

besides, as big a jump as doom3 is, real-time graphics still have a long way to go. super basic things that are a given in non-realtime cg like anti-aliasing is just NOW becoming a viable option. dynamic lighting? simple stencil shadows? that is nothing.

the 'potential' of 3d gaming is so staggering. people think gameplay is everything, but when you can make convincing, immersive worlds the gameplay is ENABLED to go to the next level.

:D

Share this post


Link to post
chumps said:

people think gameplay is everything, but when you can make convincing, immersive worlds the gameplay is ENABLED to go to the next level.

Exactly. Imagine that day when your body and mind reacts to a "game" just as it would in the equivalent real life situation. Think about the infinite possibilities of exploiting that technology while designing your game.

Share this post


Link to post
Zaldron said:

It's a big step forward. But still, we're lacking things like radiosity, extensive object-based light sources, color bleeding, realistic caustics, sub-surface scattering, deformable geometry and oh so much more.

*cough*
Please explain

Share this post


Link to post
Psyonisis said:

*cough*
Please explain

Gladly.

Radiosity : That's when lighrays bounce several times against geometry (generally 5 to 8 times), acummulating brightness in those places where the simulated photon hits. This allows you to pretty much completely remove pitch black shadows. This way it's more appealing to the eye, and really good looking. So much, in fact, that it fools the brain into thinking it's a photo from a real object (provided the model/materials are good enough). Obviously we can't do this in real time, in fact PCs barely make it for complex models when you're rendering a scene in 3d modelling packages.

Extensive object-based light sources : Everyone bitches about the shadows being sharp, but in fact you can't have soft looking shadows unless you have this (excluding that hack-job mentioned here a while ago). It basically means to have real geometry do the light casting, instead of point lights. Imagine a bunch of lightrays coming from one point, and colliding against something... say, a chair. Lightray 1 will hit the chair, so does Lightray 2, 3, 4, 5, but 6 completely misses it. Since there's no middle ground the light in the ground behind the chair goes from nothing to something in just 1 pixel. Extensive light sources can't be quite made in programming, but you can simulate them by placing lots (and I mean lots) of point lights scattered uniformly across the object's volume. Obviously this takes so much more time to render that it would be wiser to make the levels 20x more detailed to really give any good use to all that GPU power.

color bleeding : This is a personal favorite of mine. When dealing with lightray bounces, you can simulate the spectre found in both natural and artificial sources of light. A blue ball looks blue under the light of the sun because it absorbs the red, orange, yellow, green, violet components of its spectre and lets the blue bounce. So if we had Radiosity, and builded a big gray room and put a big blue sphere in a corner, the area near the ball would have a blueish tint.

realistic caustics : Lightrays that bend/reflect based on the refraction coeficients of the materials. The phenomena that makes all those funky electricity-looking patterns in the bottom of a pool when the water is full of ripples.

sub-surface scattering : This is when lightrays go trough materials not so quite opaque. The best example is a candle, where all the wax next to the flame is luminouscent, while the bottom is not. That's because lightrays blatantly lose energy when dealing with dense matter.

deformable geometry : Self-explanatory.

Share this post


Link to post
EsH said:

I mean seriously, why bother?

Because in 10 years times DooM III will most probably be the DooM of today.

Share this post


Link to post
Zaldron said:

At least there will be far less half-assed attempts at coding new engines.


Ah...yeah <minimizes MSVC, and whistles innocently>

It's just gaustly to think about, though... watching Carmack is like seeing a Godzilla film. One minute, the streets are filled with hords of happy companies devoloping "amazing" 3D games, then something like Doom3 comes along, and lays waste to everything. Unreal2... Kreed... all that time and effort and hype, and in the end, all they'll have to show is somehing average and second-best (que synth violin music). I felt really sorry for Epic... I think Gamespot had a tiny little article about Unreal2, with one screen shot. Haven't even heard about "Doom Killer" Kreed...

Share this post


Link to post

zaldron`s post

that souns interesting.but,most of those feature is not so remarkable(even if, how hard it will be implemented) IMO.
deformable geometry,that`s the next feature that should be worked on.

Share this post


Link to post
dsm said:

What do I know - it looks photorealistic enough to me.

People said similar things about DOOM nine years ago :)

Share this post


Link to post
EsH said:

I think Gamespot had a tiny little article about Unreal2, with one screen shot. Haven't even heard about "Doom Killer" Kreed...

i pay attention to all of the game that deserve it.this included NOLF2,Deux Ex2,republic and etc..

Share this post


Link to post
Dark-tenshi said:

that souns interesting.but,most of those feature is not so remarkable(even if, how hard it will be implemented) IMO.
deformable geometry,that`s the next feature that should be worked on.

They are just simple details that completely fool the brain into thinking it's photorealistic. Deformable Geometry is interesting yes, but there's no way for a computer to mantain a consistent level of detail. You can't destroy walls and computers and expect to see pipes and circuitry automatically. And while it seems it's not a big issue, it would really kill the inmersion in an engine that looks absolutely spectacular when you're not fooling around with surfaces.

Share this post


Link to post

yes, it looks hard to implement.that`s why Red Faction`s story is set in Mars Mining.it might need special piece of artwork/texture for destroyed geometry.but if the destroyed geometry do look all the same.the immersion will be lost.

Share this post


Link to post

@Zaldron
Im sure radiosity wont be shown up in a game in the next 20 years.
You have to fake details in a realtime-environment because the cpu power today (and in the next few years) wont be able to render a scene via radiosity. You cant speedup radiosityrendering with a better graphiccard, all the calculations are done by the cpu.

Some highquality radiosity renderers are Blue Moon Rendering Tools (used for Ants and Shreck), MentalRay, vRay, Brazil...
I think youve heard how long they rendered the movies.
Ok, using a resolution of 3800x2400 (or whatever the exact cinematic resolution is, its around that) is a bit more than the todays 1024x768 in games. But you can be sure that a PIV 2,6ghz is to slow for rendering in realtime with radiosity a old game like quake.

Share this post


Link to post

Well, who knows what we'll have in the future? Perhaps the graphics cards will have multiple processors, one separate for radiosity then...

Share this post


Link to post
Fredrik said:

Well, who knows what we'll have in the future? Perhaps the graphics cards will have multiple processors, one separate for radiosity then...


I dont think thats possible. Radiosity is much more than a OGL function like the dot3 bumpmapping.

Share this post


Link to post

Well, having in mind how quickly technology advances, we'll probably have computers capable of it in 5-10 years anyhow...

Share this post


Link to post
Fredrik said:

People said similar things about DOOM nine years ago :)

Heh, I remember my English teacher (he was a pretty cool guy) said the he couldn't imagine anything with better graphics than Doom.

Personally, I didn't find the graphics in the old Doom photorealistic - it didn't look like a movie, but I loved the graphics all the same, they were some of the best graphics out there for the time.
After having "digested" the impressions of Doom, I began to imagine that one day, FPS games would have movie-like quality graphics - seems like now in Doom 3 to me :-)

Sure, the next generation engine might come out looking like REAL cinematic quality (not quite CGI-like like Doom 3), but it will not have the same effect of awesomeness on me, because Doom 3 made the first HUGE step towards really looking like a movie (It looks far better than anything in Toy Story).

Share this post


Link to post

I wont disagree that we will get cinematic quality in 10 years in real time but im sure its not done with radiosity ;)

Its easier and more effective to fake some effects than to calculate it, for example the volume lights in q3. Its only a shader wich fakes the lightcone.
Even in doom3 the fire is still a sprite and looks kick ass.
Im using in 3dsmax maps instead of calculation a fire with plugins like afterburn wich take MUCH time to calculate and MORE time if rendered with a radiosity renderer. And it doesnt look complete different from what i will get after 3 hours more renderingtime.

I stand only for the point that we wont use radiosity in games in the next 20 years. Maybe in 30-40. Who knows what will happen...if anyone invent a revolutionary rendering methode or whatever...

Share this post


Link to post
Burzum said:

Even in doom3 the fire is still a sprite and looks kick ass.

Uhh, do we know that yet? We haven't really seen that much fire as of yet and it might as well change in the final release.
But oh well, I'm no tech guy so I know nothing about this, all I know is that Q3A's fire wasn't sprites, it consisted of two crossing, flat "plates" and they used a similar technigue for some of RtCW's fire.

Share this post


Link to post
Guest
This topic is now closed to further replies.
×