Doom 3 - Architectural Renderer of Choice?

[Devin Johnston]

I wan't to be able to do this with architectural models, why hasen't anyone developed a plugin for Max that would allow you to use the Doom or Half Life engine!

[abicalho]

You can do that with HL2. There's a source for a MAX plugin in their MOD code. I compiled it. I also did that with Unreal.

 

The MAIN problem is that it's not that simple. You can't simply throw geometry out. Geometry has to be optimized for the game, else it won't display it or work at all. Geometry has to always be closed, like a solid. It has to fit their unit paradigm, you have to add a couple extra objects that define regions, etc for occlusion. It's a complicated process!

 

I failed badly to bring just a single room into UT 2004. Tried all different exporters, spent a LONG time with it. It's just not easy. If it were, you would have seen tons of exporters, scripts, tutorials out there.

 

Alexander

[Devin Johnston]

It seems to me that there is a built in market for who ever can develop a relatively easy way to export models into these gaming engines. Which ever software maker does it stands to make a tremendous amount of money, which is why I'm surprised that someone hasn’t tried to make it work.

[Richard McCarthy]

I seem to remember seeing that Valve show off a "HDRI" demo (Cubic Mapping + Target Light Bloom...easy trick ) running in real-time in Quicktime Move format.

 

 

What HAPPEN to that feature of Half-Life 2's Source engine?!!

Wasn't that the main selling point of Source Engine?

 

Also wasn't Source also touting the "Random Genetic Face Generation" sort of technology (so every character in the game look unique) ?? I don't see that in my retail version of Half-life 2......strange?!

[Richard McCarthy]

Some "wierd" structures in Half-life 2 that I thought maybe game designers are incorrectly informed......

[oluv]

hey richard!

 

you seem to be as fascinated by the game as me ;-)

 

the game even runs smoothly on my shitty laptop with ati mobility 7500, although without all the "special" effects like reflections etc. still it looks very impressive.

 

i couldn't resist to also attach some maybe more "architectural" screenshots.

i think the high quality precalculated radiosity is what makes it look so real.

[IC]

Those screenshots would beat 50% of the architectural renders I see day in day out

 

When I first got Tomb Raider I used to spend hours just wandering around looking at the Architecture/Scenery.

Game took twice as long to complete as it should have.

 

Haven't seen a noticeable leap forward in any game since then until I saw these (and that includes Half Life which I loved.)

 

Oh look its lunchtime-think I'll nip out to Game.

[RyanSpaulding]

I actually despised the Doom3 engine. I felt it was EASILY surpassed by the HL2 engine. Doom 3 stuff was so low poly and too dark. Also, graphics on everything look good until you're up close where the normal mapping is obvious.

 

Interesting project tho.

[Pablo Zurita]

I actually despised the Doom3 engine. I felt it was EASILY surpassed by the HL2 engine. Doom 3 stuff was so low poly and too dark. Also, graphics on everything look good until you're up close where the normal mapping is obvious.

 

Interesting project tho.

I think the Doom 3 engine was great for what it was used for. Given the hardware constraints, time of development, etc. I think id Software made a great engine for the game they wanted to make. If they used a lighting and shading solution similar to what Valve used in HL2 then Doom 3 would be much worse since Doom 3 to create the atmosphere it has it required highly dynamic lighting and shadowing. And that's something that that HL2 doesn't have. Of course the Doom 3 engine would never work for architectural presentations due to the fact that there is no way to specify lights with diffuse inter-reflections of light.

[Richard McCarthy]

I felt Doom3 engine represent more like the last generation then the next generation. Even though it is said to have "unified lighting model", if you think about it, it's really more like the first generation CG... raytracing. Dark black shadows. On the other hand, Half-life 2 which represents the "last generation" is polished to a golden shine, the irony is, it uses radiosity in it's lighting, which in CG history is "more" advanced. (Even though it's baked Lighting)

But Doom3 engine definitely paved the way for the developer, now I think the next step would be to achieve realtime radiosity.

[Pablo Zurita]

I felt Doom3 engine represent more like the last generation then the next generation. Even though it is said to have "unified lighting model", if you think about it, it's really more like the first generation CG... raytracing. Dark black shadows. On the other hand, Half-life 2 which represents the "last generation" is polished to a golden shine, the irony is, it uses radiosity in it's lighting, which in CG history is "more" advanced. (Even though it's baked Lighting)

But Doom3 engine definitely paved the way for the developer, now I think the next step would be to achieve realtime radiosity.

What the unified lighting model does is allow to light all the geometry the same no matter if it is static geometry or a dynamic model. The benefit of this is that all the geometry looks the same, but HL2 doesn’t have that.

 

While HL2 does have diffuse inter-reflections, its all lightmap based like Quake 3 or Quake 2. This has a number of problems that made id Software stay away from that kind of lighting for Doom 3. First the lighting is very static, you can’t do most of the effects that Doom 3 uses for example those dynamic red lights, the lights attached to moving objects, etc.

 

Another reason for not using an HL2 kind of lighting model in Doom 3 is the fact that in Doom 3 they had many moving objects that change the lighting a lot. While in HL2 there are dynamic objects, they are mostly small. In Doom 3 they have big objects that if they don’t cast shadows and are lit correctly, then the scene looks pretty bad. You can see this in HL2 when you are for example in a dark indoor environment and then you open a door and you are in a bright outdoor environment, if you look at the door you will see that the door looks pretty weird. In fact you can almost always determine what geometry is static and what’s dynamic. In Doom 3 you can’t because all the geometry is lit the same way.

 

And the last reason I would say is that since Doom 3 is very indoor based game, having a lighting model like Doom 3 has gives some nice benefits, the main one being the fact that there is no lighting compile time, and another one is that in the editor you can actually see how the map is going to look like, and its not an approximation, it’s the real look. All these benefits come without a really high visual cost, since Doom 3 is all very indoor with sharp lighting, the fact that it doesn’t have diffuse inter-reflections is not that obvious for the player.

[stefkeB]

Most of what you see in the commercial games is the quality of the level-designers: modelling, mapping, shading and lighting and optimised for low-poly environments.

 

The Unreal Engine and the Half-Life Engine and the Doom Engine can jump over eachother with each new release, but at the moment, Half-Life2 might have the edge.

 

But since I happen to have been playing a bit with doing architectural visualisations in the past (using Quake3 and Unreal Tournament 2003, which are both stupid games, but have a nice engine), I found that it's rather manual labour...

 

The engines are optimised for geometry that is only placed on the grid.

They have rather "fat" collision detection (you usually can't pass through a simple 90cm door).

The perspective view is usually a very wideangle view, which gives a rather distorted view.

Most engines favour Dungeons (enclosed environments).

 

You can make a clean level in the Unreal Editor with the included boolean-modelling tools. The visual detail is often modelled elsewhere (e.g. XSI for Half-Life, which is also used for most pre-calculated lighting).

 

One of our students used Unreal for a full museum model and got pretty far. He used lightmaps for the outside and vertex lighting for the inside which looked limited, but was more feasable then doing it all with lightmaps...

 

Oh, last time I heard, the cel-shaded graphics of XIII used the Unreal Engine.

[IC]

Eventually got round to buying HL 2 last week and I am totally amazed at the environments (and just about everything else for that matter). Okay the geometry isnt very complex but the mapping is terrific.

 

What struck me in the first task of the game is the way you can just about always see the imposing blue tower projecting up into the clouds. It's these little things that make it more than just a game, unlike the unengaging competition.

 

The number of buildings in the view at any time is quite mind boggling. It takes me ages to get anywhere because I look at everything in sight.

[mbr]

I am not a gamer and really never have time to play, even if I was. But I am interested in the real time potentials. I've used EON to create some interactive environments, but it's just not there yet for most applications. The time it takes to optimize is just crazy, then getting your maps correct, etc, etc

 

These screen shots of HL2 are fantastic! I may just go and buy it for the renderings! That's the advantage of having HUGE budgets - you can work to get it exactly right. With arch viz, it's just not that flexible and you don't work a year on one project, with one goal (how many times have clients changed the scopes? "Oh, you want an animation now that it's a super high poly model made for one view??")

[Richard McCarthy]

I am not a gamer and really never have time to play, even if I was. But I am interested in the real time potentials. I've used EON to create some interactive environments, but it's just not there yet for most applications. The time it takes to optimize is just crazy, then getting your maps correct, etc, etc

 

These screen shots of HL2 are fantastic! I may just go and buy it for the renderings! That's the advantage of having HUGE budgets - you can work to get it exactly right. With arch viz, it's just not that flexible and you don't work a year on one project, with one goal (how many times have clients changed the scopes? "Oh, you want an animation now that it's a super high poly model made for one view??")

 

 

I think what your approach might be wrong. Modeling for high polygon with high resolution textures then optimise isn't the efficient way to do it. I think if you approach it like the game designers, start out just with massing study level of polygon count.. LOW POLYGON modeling all the way while keeping an eye out for texture resolution, it will save you a lot of steps later for optimisation.

 

I just recently bought the book "Making of DOOM 3" and it was very informative and entertaining book. It is filled with illustratioon and photos and I highly recommend anyone who's interested in real-time stuff to purchase it.

In one of the chapter, it was interesting to note that most of the ID Staff are NOT experienced in modeling for high polygon .. meaning they model low poly through and through, and it is because of DOOM3's new normal mapping technology that they had to start model in high polygon and then capture the detail into normal map and optimise the model to low polygon. That's why it's been taking them so much time.. they had to learn the new process and also learning things about "specular mapping" and "diffuse color map"

 

 

So, in essence, I think Real-time VIZ is actually simpler than Rendering VIZ, it is like the approach of using SketchUp, if you keep it simple, and approach the way game designers do, you should have no problem

[Brian Hay]

I agree with Richard.

 

Low poly is the way to go. Put the detail into the maps. The other advantage to this approach is render times. With low poly models and high res maps, rendering out to video (if necessary) and baking GI lighting to maps is generally much faster.

 

RPC has been very successful with their image-based rendering solutions. Just apply the same concepts scene-wide and that's essentially your workflow.

 

Brian.

[Richard McCarthy]

I forgot to mention another thing, I think low polygon has advantage over high polygon modeling. For example, now, every 3D app worth it's salt have Sub-division modeling feature, and because of that, you have more advantage to keep your polygon model low, and convert to sub-div later for high polygon presentation model. This is especially true in character modeling where a lot of interpolation of surface features comes into play.

[IC]

I don't know how relevant this is in Arch Viz.

It's ok when you're modelling, say, a rusted metal surface with old ruined windows but for a pristine roughcast (stucco) finish with proper window reveals and nice new shiny uPVC and glass it's another matter altogether.

 

You get away with a lot in the game environment because of the pace and variety.

Saying that, check out these images from a map made entirely by one guy:

 

http://www.cgtalk.com/showthread.php?t=227088

[oluv]

Saying that, check out these images from a map made entirely by one guy:

http://www.cgtalk.com/showthread.php?t=227088

 

wow! these screenshots look freakin' cool! did you download and play this map?

by the way, how do i apply this map? where do i have to copy it to let it work?

[IC]

I don't know anything other than what's on the CGTalk thread.

 

I had enough trouble just installing the game never mind any of that malark! Took me about a full day and I still can't get it to start up in offline mode.

[oluv]

I don't know anything other than what's on the CGTalk thread.

 

I had enough trouble just installing the game never mind any of that malark! Took me about a full day and I still can't get it to start up in offline mode.

 

 

 

-== POST EDITED by FORUM MODERATOR ==--

[Richard McCarthy]

Half-life 2 requires you to be ONLINE to play. You MUST connect to steam to play. If you are trying to play in offline mode, you are most certainly playing a warez version. Please refrain from talking subject matter relating to warez in this forum. It is strictly forbidden.

[IC]

Excuse me but I'm most certainly not playing a warez version.

 

As I said, it took me almost a whole day to register with Steam and get going. I am having to go online every time I play but it clearly says in the instructions that Offline Mode is available if you tick the 'remember my password' box which I did.

 

I assume this means you can then unplug from the internet once you start playing? Sometimes when I do that I get an error message-I must be doing it too early.

 

Quite a serious accusation/assumption you have made there Richard.

[Richard McCarthy]

I am sorry, I assume you are trying to play an illicit version judging from oluv's post previously. My apology. Judging from the Steam forum there seems to be a lot of players having problems to play offline mode right now after updating.

 

http://forums.steampowered.com/forums/showthread.php?threadid=273537&perpage=15&highlight=offline%20mode&pagenumber=2

 

Looks like you can only play online right now.

[IC]

I got Oluv's comment by email so I can understand why you thought that.

No harm done.

 

I'm a bit out of touch with game trends-I've never had to go online to play an offline game before but if it stops piracy that's fine I suppose.

 

I looked at the Steampowered forums on Saturday to find out why I was having problems and noticed others were too.