NS2 static/dynamic models for structural purposes
Browser_ICE
Join Date: 2002-11-04 Member: 6944Members
Join Date: 2002-11-04 Member: 6944Members
<div class="IPBDescription">What is planned related to this </div>Looking at the colored concept made me realized something.
How much effort will be put into NS2 for realism in terms of map structural details vs static/dynamic models ?
I mean, just looking at this corridor, you see right there a high volume of poly amount if you want it to be that detailed. Now doing this with map brushes is a bit of pain and you loose a bit of quality doing it this way. Doing it as a model static or dynamic, you can achieve a high quality detail. Maybe having some of those pannels or corridor componants as models would be a better approch but then, I don't know what is your budget in terms of model limits.
So what is the NS2 developement team comments on this ?
What kind of structural models are planned for NS2 ? Static or dynamic ?
If dynamic, how dynamic will they be ? Moving parts ?
Having structural models would enable re-usability and modular design for NS2 maps and even give the NS2 future community something to play with when designing their own maps. You could have endless combinations of corridor types just by assembling different pannel models.
Maybe a friendly competition amongst the community to bring some of the models static or dynamic would be a good idea ? The chosen ones would be included as a standard package to NS2.
What do you think ?
How much effort will be put into NS2 for realism in terms of map structural details vs static/dynamic models ?
I mean, just looking at this corridor, you see right there a high volume of poly amount if you want it to be that detailed. Now doing this with map brushes is a bit of pain and you loose a bit of quality doing it this way. Doing it as a model static or dynamic, you can achieve a high quality detail. Maybe having some of those pannels or corridor componants as models would be a better approch but then, I don't know what is your budget in terms of model limits.
So what is the NS2 developement team comments on this ?
What kind of structural models are planned for NS2 ? Static or dynamic ?
If dynamic, how dynamic will they be ? Moving parts ?
Having structural models would enable re-usability and modular design for NS2 maps and even give the NS2 future community something to play with when designing their own maps. You could have endless combinations of corridor types just by assembling different pannel models.
Maybe a friendly competition amongst the community to bring some of the models static or dynamic would be a good idea ? The chosen ones would be included as a standard package to NS2.
What do you think ?
Comments
This is good for workflow, but it's also essential with the Source engine since creating complex geometry with the BSP slows down the engine.
Max
Don't forget a cocooned human props models! some technician operators traped in webs or something like it, as seen in Aliens.
<!--QuoteEnd--></div><!--QuoteEEnd-->
mmmm... noooo...
To much of a direct rip-off
-Sheepe
Creating complex brushwork is more taxing on the engine than creating complex models?
Don't forget a cocooned human props models! some technician operators traped in webs or something like it, as seen in Aliens.
<!--QuoteEnd--></div><!--QuoteEEnd-->
this was done waaaay back in duke nukem 3D, i dont think it would be suitable for NS...
A good example of how this works is the UT2.5 engine (The engine UT2004 runs on) where BSP is used for the basic walls/floors/ceilings etc of the maps, but all of the detail in the map is used through static meshes, basically prop models which are cached in memory and are then rendered straight from the memory. That means they're rendered a lot faster than having to be rendered through the CPU/GPU as it's essentially only rendered once and then copied to wherever else the prop is used in that map.
I would imagine a similar method is used by Source to render prop models.
And if you add onto that the fact that the actual geometry building tools in Hammer are pretty bad compared to an external 3D app such as 3DSMax or Maya ....
Original Half-Life style BSP mapping, using materials in interesting ways
Reusable props to enhance the detail of the world, architecturally and otherwise.
'One-shot' props that are made on a per-map basis and are unique to that setting/environment.
i think thats true, but im not positive. i know for a fact though that complex brushwork can never look as good as models simply because you cant really make that complicated of brushwork. its much more difficult to work with from the modelling and skinning standpoint, and doesnt look near as good.
<!--QuoteEnd--></div><!--QuoteEEnd-->
Thats not really true, unfortunately in HL2 the only clear advantage in using props is ease of creation and speed of rendering. For everything else brushwork will pretty much look better, since its actually lit properly. It receives shadows and is lit correctly in comparison with the environment, things which props dont do. The props in the source engine are lit from a single spot, they dont even have true vertex lighting. This means for instance a long pipe that goes from a light area to a dark area for instance will often look terrible, either glowing in the dark at one end, or looking like its in the depth of shadow at the other. As far as I know this is how it still works, and its not that much of an advancement from HL1. This lighting style works *quite* well with "realistic" maps with lots of ambient lighting, but not so well for contrasty light and dark areas as we may be seeing in NS2. Hopefully static prop lighting will be getting a serious overhaul after episode2 though! At least I have heard that that may be the case.
IF what I have said is the case, which as far as I know it is, it could prove quite tiresome working around these problems. In the original NS the theme often involved heavy use of sharp edged shadows and spotlights, unfortunately the way Source's current lighting system is, as soon as you put a prop accross one of these sharp lighting edges you will get an eyesore. I'll be crossing my fingers that we will at least get correct vertex lighting in the update.
Have you tried compiling your maps with '-staticproplighting' as one of your HLRAD parameters? It <b>drastically</b> improves the lighting of static props. I'm not sure why it isn't on as default. Props that fall between two contrasting light styles are privy to looking rather ugly still, but for the most part it makes them much more accurate at detecting subtle light positionings and such.
<!--QuoteEnd--></div><!--QuoteEEnd-->
thanks for that, it seems to enable correct vertex lighting at least, I cant expect much more than that I suppose <img src="style_emoticons/<#EMO_DIR#>/wink-fix.gif" style="vertical-align:middle" emoid=";)" border="0" alt="wink-fix.gif" /> I still find it worrysome in a way though, looks like I will have to take into account the lighting aspect when building props and designing their poly distribution based on possible lighting conditions.
And 6john, your right, props have a lot of benefits, Its just that with source and the way its lighting is done you have to be pretty careful about what you do to make sure everything looks cohesive. Large portions of fairly poorly lit props really stick out next to the crisp realistic light bouncing and shadows on brushwork. I was mainly pointing out that lighting is a huge part of the feel of NS and its maps, and lighting isn't one of Source's particularly strong points at the moment.
Speaking of which, does anyone know some details about what engine changes are coming along with episode 2?
Speaking of which, does anyone know some details about what engine changes are coming along with episode 2?
<!--QuoteEnd--></div><!--QuoteEEnd-->
<a href="http://en.wikipedia.org/wiki/Half-Life_2:_Episode_Two#Technology" target="_blank">http://en.wikipedia.org/wiki/Half-Life_2:_..._Two#Technology</a>
I have seen some examples (videos) of the episode two engine enhancements, but I can't remember where.
However, the benefits of Source's other features greatly outweighs its lackluster prop lighting, in my opinion. And the great thing is that the Source engine gets better with time, whereas the Doom3 engine, so far as I understand, is pretty much decaying already.
thanks for that, it seems to enable correct vertex lighting at least, I cant expect much more than that I suppose <img src="style_emoticons/<#EMO_DIR#>/wink-fix.gif" style="vertical-align:middle" emoid=";)" border="0" alt="wink-fix.gif" /> I still find it worrysome in a way though, looks like I will have to take into account the lighting aspect when building props and designing their poly distribution based on possible lighting conditions.
And 6john, your right, props have a lot of benefits, Its just that with source and the way its lighting is done you have to be pretty careful about what you do to make sure everything looks cohesive. Large portions of fairly poorly lit props really stick out next to the crisp realistic light bouncing and shadows on brushwork. I was mainly pointing out that lighting is a huge part of the feel of NS and its maps, and lighting isn't one of Source's particularly strong points at the moment.
<!--QuoteEnd--></div><!--QuoteEEnd-->
From when I used to map with the HL2 engine it seems to light models/meshes perfectly well, and definately a lot better than the UT2003/4 engine does. Plus, obviously it has bounced lighting making for some realistic lighting.
BSP for small details is what you did pre to 2003. Engines nowadays should be model based and BSP shouldn't be used for small details. That's why there are no custom maps for HL2/CS:S that look better than the best UT2003/4 custom maps (from what I've seen), since everyone seems to use brushes for everything. It's kinda sad actually.
Here's an extreme example of meshes/bsp, but it's not the best example since some of these models were probably pre-lit in maya.
<img src="http://img63.imageshack.us/img63/4382/staticmesheshq6.jpg" border="0" alt="IPB Image" />
I'm glad they're going in the model direction.
Just for example run through ravenholm again, with its extensive use of shadows and light and dark areas there are lots of instances where things just look a little "odd" with the way props are lit.
Either way, source was by far the best choice of engine for NS, and with its updates it will no doubt start to look better and better!
To demonstrate this further, it would be so much easier if someone would create screenshots that same colored concept image with wall models combined in different ways. Having a modular design would speed up the maping process of the community and therefore bring more diversity. The more modular modeled walls there are, the better.