Doom 3 Vs. Half-Life 2 redux....

Originally posted by EvilEwok
There are times to use physics and times to use other means. No the hallways arent littered with toilets and barrels for you to pick of and throw like hl2, but physics IS intigrated into gameplay.
Click to expand...
Are you saying that in a city being destroyed by aliens, the combine, and the resistance there isn't going to be any debris or other objects lying around in the streets?

I'd like to see your vision of a war-torn city...
 
quote:
--------------------------------------------------------------------------------
Some parts look like a low detail room with lots and lots of nice shiny textures they rely on to provide detail.
--------------------------------------------------------------------------------

If I didn't know better, I'd think we were discussing Halo! ;)
 
Originally posted by EvilEwok
It does in places where it is efficient to do so. Might aswell make the most use of a feature you can, and if you can make a 7000 polie model look as good as a 10,000 polie model, doesnt sound bad to me.
Click to expand...

No, it tries to replace a 20 polygon tube or pipe with a flat normal map.
For most monsters it works fine, but it makes geometry look so fake and flat. A lot of the bigger normal map details (im not talking about little screws and bolts here) can be easily done with a very small amount of triangles with a much much better result.


Are you one of those who freaks out when they see a poly edge in a screen shot? It looks fine in motion, because the outline is always in motion. Sure if they increased the poly count there would be less visible edges in screenshots, but in motion you really cant tell so there is no need for extra polies.
Click to expand...

I've seen it in motion (the alpha and the video's) and monsters look ok, won't hear me complaining about that. But the abuse of normalmapping on things that don't need it or try to replace entire objects with it is just shit.


And this is different form other games how? No game uses polies for all the detail, they use textures for detail. Look at the buildings of hl2. 90% of them are just boxes with a brick texture, and a window texture.
Click to expand...

Actually, houses are pretty complicated geometry. But my point is that normal mapping + polygons looks stunning, but Doom 3 tries to do it with normal mapping, without the polygons.

The difference is that doom3 doesnt look like a box with a texture, it looks like geometry. You cant get all the detail that doom3 has with polies, there would be too many. And it would look shitty to just use a regular texture, because the details arent flat. Normal maps make it look like geometry, so it works.
Click to expand...

It works partially, from low angles it's as flat as a pancake, it doesn't selfshadow either as far as I know. And I'm not asking to replace all those small screws and bolts in a metal place, those can be done with a normal map, but the larger details look awful with normal maps, just because it's flat with a 3d simulation.
In a map the metal plating reliefs can be done with normal maps, the bolts and screws too, bricks and stones are fine as well. But please make those pipes in the ceilings and walls 3d, a lot of 'em arent.
 
"I doubt that normal mapping is going to be abandoned anytime in, let's say the next ten years. Yes you get more detail out of polygonal models built from displacment maps, but you can then use normal mapping techniques over the top of those and add even more detail to a particular scene or object. Human facial features will be massively realistic within the next five years using such a technique."

Ten years is a long time in the gaming scene... enough to go from Labyrinth and Doom to HL2 and Doom3.

In Doom and Labyrinth we had no hardware accelerated video and a total on-screen polycount of a single HL headcrab model (maybe less).
... and that was just basic single-textured polygons and sprites.

Now something as small as a headcrab could have 3000 polygons, a shader, a normal map, a specularity map, and an alpha map.

... and hardware support for displacement maps allows higher polycounts at a fraction of the performance hit.

I can see people totally abandoning normal maps in 10 years and using a single heightmap to calculate the lighting that normal maps handle and the actual displacement.

In ten years, I wouldn't be suprised to see individual character models with polycounts near 1 million (hopefully with a good dynamic LOD system by then)... or even a game with a renderer based on non-polygonal surfaces like NURMS.
 
Seriously, look at the progress in ten years, from Doom to Doom 3. It's amazing, imagine where we'll be in 2014...
 
Originally posted by OCybrManO
"I doubt that normal mapping is going to be abandoned anytime in, let's say the next ten years. Yes you get more detail out of polygonal models built from displacment maps, but you can then use normal mapping techniques over the top of those and add even more detail to a particular scene or object. Human facial features will be massively realistic within the next five years using such a technique."

Ten years is a long time in the gaming scene... enough to go from Labyrinth and Doom to HL2 and Doom3.

In Doom and Labyrinth we had no hardware accelerated video and a total on-screen polycount of a single HL headcrab model (maybe less).
... and that was just basic single-textured polygons and sprites.

Now something as small as a headcrab could have 3000 polygons, a shader, a normal map, a specularity map, and an alpha map.

... and hardware support for displacement maps allows higher polycounts at a fraction of the performance hit.

I can see people totally abandoning normal maps in 10 years and using a single heightmap to calculate the lighting that normal maps handle and the actual displacement.

In ten years, I wouldn't be suprised to see individual character models with polycounts near 1 million (hopefully with a good dynamic LOD system by then)... or even a game with a renderer based on non-polygonal surfaces like NURMS.
Click to expand...

NURBS I think you mean, I thought there was one game that actually used something like that. It used mathematical calculations to make curves, it was an RTS I can't remember the name anymore.

But a 1 million poly count will not be necessary, 50.000 with normal mapping, a single 3072x3072 texture with bump, specular and alpha channel should do the trick :)
Maybe some subsurface scattering to make nice skinshaders :cheers:

EDIT:EvilEwok, I'll try and find a nice screenshot which shows the lack of polygons. But I guess even then we won't work it out because it's just a matter of what you find acceptable in order to spare polygons. I just like to see walls with big heightdifferences and thing sticking out to be modeled, which too often doesn't happen. But it seems you find this acceptable.
 
Originally posted by PvtRyan


And you're acting like HL2 doesn't use any normal mapping, in fact, a lot of surfaces have normal maps. It uses normal mapping for relief on bricks and walls, used to add detail to objects, not to become objects themselves (or try to become).
Click to expand...

Thank god someone said this. EvilEwok fails to understand that a lot of the features that Doom 3 has Half-Life 2 has and vise-versa.

And just to get my 2 cents in . . . I think I'll enjoy HL2 better simply because of the story. I can't imagine enjoying Doom 3 as much because I don't want a fear-based immersion. I want a real-world immersion and an emotional connection with the characters kind of immersion. For example, having a world detailed enough that you can look at virtually any object and expect to be able to throw it, destroy it, etc. John Carmack himself said that he didn't want to put all those extra things in Doom 3. He called them "gimicks." I don't think they're gimicks. In fact, I think they're important.
 
If you want the ability to manipulate your environment in infinite ways (using almost any object, etc.) in either of these games I would prepare for some dissapointment. You can really only use what the designers decided you could use nothing more nothing less. For example you would think that a heavy oil barrel flung at a filmsy window would shatter that window. Well, you'd think that anyway.

And I don't think the Doom 3 alpha (as impressive as it was at the time) was anything indicative of the product we'll see when Doom 3 ships. I mean the alpha only had some very, very basic physics in it and only demonstratable on certain objects. The videos I have seen since then are much, much more impressive.

Like I said, give me the ability to shoot out lights and do some sort of permanent damage to the environment and you'll have me hooked.
 
Originally posted by PvtRyan
NURBS I think you mean, I thought there was one game that actually used something like that. It used mathematical calculations to make curves, it was an RTS I can't remember the name anymore.

But a 1 million poly count will not be necessary, 50.000 with normal mapping, a single 3072x3072 texture with bump, specular and alpha channel should do the trick :)
Maybe some subsurface scattering to make nice skinshaders :cheers:
Click to expand...
NURMS = Non-Uniform Rational MeshSmooth
NURBS = Non-Uniform Rational B-Splines

Either one would be interesting to see in a real-time game engine...

... and the 1 million polygons was an extrapolation of the increase in polycounts in games over a 5 year period.

I wouldn't actually expect to see a character with a polycount of 1 million... unless it were made up of tons of parts that each needed a lot of polygons.

In ten years I would expect that we would have already seen things at least as detailed as this being used as actual game models instead of models used to generate normal maps.

Lighting done with normal maps doesn't look as realistic as lighting done with polygons... and normal maps don't look correct at low angles.

To see an example of this go to http://mirror.ati.com/vortal/r350/flash/9800educational/index.html
Click on Truform 2.0... then go to the 3rd page in that section.

So eventually, when models can have high enough polycounts, normal maps will almost definately be abandoned (you have to make a hi-poly model to generate the normal map anyway, so just use the hi-poly model).
 
I think D3 looks better graphically than HL2, but HL2 scores points for the use of detailed material physics (wood acting like wood) and the facial expressions. D3 looks more real, but HL2 feels more real. Personally I think that Stalker: oblivion Lost is using quite a nice engine as well (X-Ray Engine), which seems capable of rendering large areas and incorporates a good physics engine, dynamic lighting and an impressive level of detailing and textures.

Ultimately the really exciting engines are those that are going to be with us in a few years time, that incorporate the positive lessons from all of these engines.
 
Doesn't Sony have some NURBS hardware in the PS2? I thought they were big on that technology.
 
Originally posted by Lavrik
Sonys technology is old news for the PC
Click to expand...

Huh? I wasn't aware of any NURB acceleration technology on the PC.

I remember an interview that some game news group did with John Carmack and Tim Sweeny where they asked where the future of 3D graphics was heading, and they specifically asked about NURBs and both JC and TS shot it down saying that it offered no real advantages over the current polygon rendering styles. In fact they both seemed to think that finding new ways of pushing more polygons with less memory bandwidth was where everything was headed. Tim Sweeny seemed particularly excited about being able to do procedural texturing. I would love to see that at work in a game but I have a feeling it's going to be another few years before I see some cool stuff come of that particular use of pixel shaders.
 
Originally posted by InsertNameHere
Seriously, look at the progress in ten years, from Doom to Doom 3. It's amazing, imagine where we'll be in 2014...
Click to expand...

http://graphics.stanford.edu/~henrik/animations/

Take a look at the first animation, "The Light..." download the larger one, and prepared to be awed by the lighting. THIS is some of what I look forward to seeing happen in realtime gaming environments - maybe ten years, maybe less who knows. But the future is bright --- Soli
 
Sweet! I imagine that some game engines will be able to do that in about 5 years time. We'll have to wait for Unreal 3, Doom 4 or Half-Life: The Next Generation.
 
I'm not so sure... Photon simulated rendering in a FPS?...

Perhaps when we get to the GeForce 12..
 
Originally posted by Syphoon!!
I'm not so sure... Photon simulated rendering in a FPS?...

Perhaps when we get to the GeForce 12..
Click to expand...

It's more of a CPU thing than a GPU one. Look at what's happened in just the last few years with physics in games. They had to make a lot of tradeoffs in physics because the CPUs weren't powerful enough to do the game and physics at the same time. Now we have realtime force calculations based on the velocity and weight of objects!

The photon calculations would probably be done on the CPU and passed onto the GPU for the proper lighting calculations and final rendering. It's a lot closer than you think.
 
Originally posted by Unnamed_Player
Doesn't Sony have some NURBS hardware in the PS2? I thought they were big on that technology.
Click to expand...

Supposedly the new FPS game Killzone that was announced just a short while ago is using NURBS. It looks pretty damn good for a PS2 game and it kinda reminds me, artistically, of HL2 in some respects. Very little is known of it so far though, they're keeping a tight lid on it for the time being.

www.killzonegame.com
 
Eventually there will be real time global illumination in games it's just a question of when. To produce something of say current movie CG standards in real time would take a decrease in rendering time of about 10,000 fold (from 500sec per frame to about 0.05sec per frame).
Assuming that rendering time decreases linearly with processor speed that's a 10,000 time faster processor. Going by the law that processor speed doubles roughly once every 18 months that means we will have to wait...... 20yrs a fair while but not that long.

(2^13=10000ish, 13 18month cycles is 19.5, near enough 20)
 

On the sound front, I am totally thrilled that Doom 3 is going to have 5.1 sound.
Click to expand...

Correction, Doom III now supports Dobly Digital 6.1 Surround Sound...
I read it in the last preview or Q&A @ gamespot..
I'll find the source for you later, as I'm @ work.. :dozey:
 
Originally posted by MooCow
Eventually there will be real time global illumination in games it's just a question of when. To produce something of say current movie CG standards in real time would take a decrease in rendering time of about 10,000 fold (from 500sec per frame to about 0.05sec per frame).
Assuming that rendering time decreases linearly with processor speed that's a 10,000 time faster processor. Going by the law that processor speed doubles roughly once every 18 months that means we will have to wait...... 20yrs a fair while but not that long.

(2^13=10000ish, 13 18month cycles is 19.5, near enough 20)
Click to expand...

You're assuming that realtime light rendering on a global scale will be done in games the way it is in movies. I highly doubt that. More likely game developers will find faster ways to do the calculations neccessary and devote CPU and GPU cycles to other things.

But look at how far we've come in just a short time. Quake 3 was released in what, 1999 and when the Doom 3 alpha was shown at E3 2 years ago people couldn't believe that it wasn't prerendered. Most graphics card manufacturers are saying that it will be totally possible in the next 5-6 years to have the CG Yoda from SWII on your desktop rendered on the fly. How cool will that be!
 
Originally posted by MooCow
Eventually there will be real time global illumination in games it's just a question of when. To produce something of say current movie CG standards in real time would take a decrease in rendering time of about 10,000 fold (from 500sec per frame to about 0.05sec per frame).
Assuming that rendering time decreases linearly with processor speed that's a 10,000 time faster processor. Going by the law that processor speed doubles roughly once every 18 months that means we will have to wait...... 20yrs a fair while but not that long.

(2^13=10000ish, 13 18month cycles is 19.5, near enough 20)
Click to expand...

While that theory is true, it is bound by the issue of physical dimension and tolerance. The limits of which are already been speculated upon with Present Chip design (you want to imagine the size of the heatsink on a processor 10000 times faster than your present one?).
 
Remember, most renders are done solely by the CPU... using a video card as well allows the computer to do a lot more in a given time by outsourcing a lot of the work to the video card.

Think of the CPU as someone that knows how to do a little of everything... he could take an extremely long time to do one complex job by reading the instructions, rush it and make it look like crap, or hire an expert to do it in a fraction of the time with exceptional quality.

After cool new rendering techniques are standardized and added to DirectX, the video hardware is built to do those new things (like pixel and vertex shaders, FSAA, AF, etc) very fast... and it is this specialized approach that lets a video card do the same thing much faster than a CPU could do even with the most optimized code possible.

Complex rendering calculations that have to be done by the CPU now (like things shown in those awesome videos by Henrik Wann Jensen) could possibly be supported by video cards within 5-10 years.

... and who knows?
By then we might even have specially designed processors dedicated to 3D Physics or AI (but that's less likely because they would have to have decide on a standard for AI on which to base the hardware) to work in conjunction with a regular processor to free up even more CPU time.

I can't wait to get all this stuff we are talking about and use it with a VR system.
 
Originally posted by Kadayi Polokov
While that theory is true, it is bound by the issue of physical dimension and tolerance. The limits of which are already been speculated upon with Present Chip design (you want to imagine the size of the heatsink on a processor 10000 times faster than your present one?).
Click to expand...

LOL. The case becomes the heatsink! :eek:

Actually Intel is already saying that there is a possibility that all their chips past Tejas will require watercooling. Air cooling just won't cut it for them anymore.
 
get some phat heaphones, you'll get fully immersed in the experience then :)
 
well i am kinda interested in those multiprocs, imagine they could double or triple the current CPU speed, btw all these things talked will probably be possible in a couple of years...just go 5 years back and look at a game...
 
sad...really.
There should be no need for liquid cooling if companys would design smart CPU's instead of inefficient heat producers.
Just think how closer we are to rendering light like this in games with 64bit CPUs. ;)
AMD's roadmaps points to a Athlon 64 4300+ (2.8ghz) by Q4 '04 while Intel's roadmap for Q4 '04 shows Prescott 3.8ghz and Tejas may be pushed back til '05. And both companys have dual cores in mind. I think they stated end of '05?
Ahh 64bit OS/64 programs/64bit CPU 4300+ by next year...yeah that sounds good. ;)
 
Originally posted by Asus
sad...really.
There should be no need for liquid cooling if companys would design smart CPU's instead of inefficient heat producers.
Just think how closer we are to rendering light like this in games with 64bit CPUs. ;)
AMD's roadmaps points to a Athlon 64 4300+ (2.8ghz) by Q4 '04 while Intel's roadmap for Q4 '04 shows Prescott 3.8ghz and Tejas may be pushed back til '05. And both companys have dual cores in mind. I think they stated end of '05?
Ahh 64bit OS/64 programs/64bit CPU 4300+ by next year...yeah that sounds good. ;)
Click to expand...

Intel recently demoed a 4 core chip at the plant near my home. Pretty cool stuff. They're still working out a lot of the technical issues but it worked pretty well. The Linux OS they had running saw it as a 4 proc box.

I can't wait for a 4 core Athlon 64! :eek:
 
When a RTSC is found processors will no longer be limited by heat... and a few weeks ago I read that someone that has been working with carbon nanotubes might be able to produce a demo of a RTSC in as little as one year.

If that is true... wow.

EDIT: RTSC stands for "Room-Temperature Superconductor"
 
the doom 3 videos, screenshots and leaked version don't look -that- amazing, nice graphics for another first person shooter. the gameplay doesn't seem that exciting or tense, though hopefully the scare factor will make it rule.
 
nano technology is the way forward, but given it's in it's infancy the production costs might be prohibitive for some considerable time yet.
 
I am anticipating HL2 and D3 equally... but hopefully HL2 manages to get released before D3, as scheduled! Otherwise, at least I'll be able to blast demons while I wait another month or so for HL2 to come out! : )
 
I honestly think compairing D3 to HL2 is apples to oranges. Each game is targeting a different audience and is really its own type of game (both are fps but d3 is more "survival horror" vs HL2's "Hero" feeling) You cant really compair the two.

You can the technology behind them though. And in that respect I think they are about equal. Each excel's in certain point's and fails in other.
 
your joking right? They are both scare/action science-fiction run for your life FPS games. They both even have a very similiar story.

its like comparing green apples to yellow apples. And in this case hl2 owns doom3 in all ways. HOWEVER, i will get both cause both are still A+ title games. GFX wise everything ive seen from hl2 pre-alpha owns anything from that doom3 alpha and newer doom3 shots.
 
I dont totally agree with you.

Doom 3 is made to be scary.
Half-life 2's video's and screens do not inspire fear like D3's do.

They are both FPS's and that is about as far as compairison goes.

This is just my opinion tho.
 
Everything I have seen coming out of Id on Doom 3 lately looks pretty fantastic. They've improved the hell out of it since the alpha leak.

HL2 looked very cool when they were demoing it until I realized that they were doing the only things you could do. For a game that strived for so much realism as far as the physics system goes you can still only manipulate the specific objects that the designer lets you. I was kind of suprised to find out that you can't even shatter windows that aren't setup to be shattered.

I expect a lot from games that have had the hype that these two have. One of them had better deliver.
 
What's with the whole "You can't compare apples to oranges." saying?

Yes, I most definately can.

Oranges are usually more acidic.
Orange rinds taste worse than those of apples.
Apples don't spray juice everywhere when you cut them.
... and the list goes on.
 
You must log in or register to reply here.

Similar threads

Krynn72
You guys, what the **** is up.
Replies
11
Views
2K
Tollbooth Willie
Tollbooth Willie
ktimekiller
The "other" ValveTime website...
Replies
2
Views
674
Shem
Shem
Lizardizzle
  • Poll
A proposition: ValveTime Half-Life 1 co-op event
Replies
5
Views
1K
Tollbooth Willie
Tollbooth Willie
Lizardizzle
2019: The year of Half-Life SFX videos
Replies
4
Views
2K
Ennui
Ennui
Tollbooth Willie
HL2.NET HORROR SLAP
Replies
7
Views
2K
Tollbooth Willie
Tollbooth Willie
Back
Top