[QUOTE=Legend286;26973795]I don't think Valve are going to use something like that, all the excess geometry created will bottleneck the fuck out of any cpu.[/QUOTE]
I don't fully know, like im not an expert on tessellation technology or even models; but i really think they used it for the torso wounds in L4D2, i mean sure you could make a model that complicated, but tessellation probably helped smooth it out.
Also: there are a few tessellation commands in l4d2 such as enabling it, its scale etc.
[QUOTE=glitchvid;26980753]I don't fully know, like im not an expert on tessellation technology or even models; but i really think they used it for the torso wounds in L4D2, i mean sure you could make a model that complicated, but tessellation probably helped smooth it out.
Also: there are a few tessellation commands in l4d2 such as enabling it, its scale etc.[/QUOTE]
They used stencils in L4D2 and some clever models, it's nothing complex. It's kinda like a more elaborate bodygroup system.
[QUOTE=Legend286;26973795]I don't think Valve are going to use something like that, all the excess geometry created will bottleneck the fuck out of any cpu.[/QUOTE]
The purposes for tesselation go beyond just better visuals. The latest tesselation techniques are very effective at scaling model detail without affecting UVs and critical geometric details. Think about creating one model that you can scale across different platforms with tesselation, rather than editing it to make 2 or 3 models. It's a huge time saver. It's also useful because scaling does not suffer the "popping" effect you get from LOD model exchanging.
I'll give Valve benefit of the doubt. They did pretty well making Left 4 Dead look realistic and it shows that Source's capabilities are measured by the artist's capabilities. When Valve does add something to the engine, it's usually done extremely well. For example, the SIGGRAPH papers revealed the enhanced gore in L4D2 used only 13% of the memory used in the gore of L4D. So in that fact, there is only few things I could actually wish for Valve to add to the engine. The problem is that Source is one of the most used engines for modding, you would think Valve might develop an engine and tools that would allow individuals or small teams to work faster. I guess we just have to dig our heels in the ground until Valve adds more stuff. It will inevitably happen.
In order of importance to me, what I would like to see most in Source:
1. Ability to scale props, most notably like in Natural Selection 2, so I'm not constraining my world for them. I'm tired of staircases that are not quite big enough, or a plank of wood that is just too long
2. Improved material system. Blends w/ more than 2 textures. Parallax & specular maps on world geometry. You would be amazed by how much more uniform and connected an environment feels with directionally correct specular surfaces.
3. A faster a way to see lighting and an addition of self-shadowing to models. It would also be nice to be able to control how much light a surface can reflect through hammer, rather than editing materials.
4. More official imp/exp plugin support for a greater range of modelling software. Do away with editing QC files with a text editor, make a program for the SDK that lets you check a few boxes, fill in a few fields, and automatically create the QC.
5. Creating terrain is tedious. Creating features such as cliffs & caves often require lots of preparatory brush & subdivision work & countless hours of work to create something believable. I'll make the request simple, fix it.
6. Optimization for larger worlds. We don't need a 20 square kilometer sandbox to play in, but something bigger than 7/10 of a square kilometer would be better. Source's limits are a tight as well and require some creativity to prevent hitting them. The ability to stream the world would be nice.
While a lot of you may not connect some of these with making maps look better, it does. The faster tools can allow a person to work, the more time that person has to make the game look better. Source and the SDK has the ability, it needs the speed that individuals require.
4 has been done to limit success... google Ducttape + Source or something.
Terrain creation is awesome, it's so un-limited. UT3's is quite limited using one big flat terrain you push up and down and then ptaches.
[QUOTE=Lord Ned;27009066]4 has been done to limit success... google Ducttape + Source or something.
Terrain creation is awesome, it's so un-limited. UT3's is quite limited using one big flat terrain you push up and down and then ptaches.[/QUOTE]
You know you can just use models for your terrain, this is pretty much exactly what Epic do, they even said themselves that the terrain in UDK currently is shit.
[QUOTE=Legend286;27009202]You know you can just use models for your terrain, this is pretty much exactly what Epic do, they even said themselves that the terrain in UDK currently is shit.[/QUOTE]
Modelling terrain would be a nice alternative. It wouldn't run nor look to good though in Source. A high poly count could keep the lighting acceptable, but performance would be lackluster.
[QUOTE=Hostel;27039795]Modelling terrain would be a nice alternative. It wouldn't run nor look to good though in Source. A high poly count could keep the lighting acceptable, but performance would be lackluster.[/QUOTE]
I was talking about UE3, the performance hit is almost nothing. As long as you cut your level up into chunks there won't be any issues regarding visibility checking and lighting.
As far as detail in maps, I think things like trim and molding always look good but sometimes left out.
If more people used -staticproppolys and those other compile commands that affect what casts lightmap shadows.
It's so simple to do, and yet not everybody uses them, but it adds so much detail and i've never seen a performance hit from it.
[QUOTE=Kuro.;27041687]If more people used -staticproppolys and those other compile commands that affect what casts lightmap shadows.
It's so simple to do, and yet not everybody uses them, but it adds so much detail and i've never seen a performance hit from it.[/QUOTE]
There is no performance hit because it's still just using the same old lightmap system. All it's doing is making more light traces during the compile time, which results in a more accurately lit map.
I use -staticproppolys, I can't live without it actually ever since I switched to the Orange Box engine.
I'm going to post the following in links since I do not want to clutter too much up.
[url]http://www.empiresmod.com/drupal/node/48[/url] - One of their new additions is a new parallax shader that works on both models and light-mapped materials. It's stable enough to implement into a mod w/ huge maps such as their own. Seeing this also gives me a hunch that you could work specular maps onto light-mapped materials as well.
[url]http://www.youtube.com/watch?v=JYiVkHEzThk[/url] - Almost everyone has seen this by now. He states in his description that contrary to the video, the shadowmaps made from the sun go from a resolution of 3072² to 1024², allowing the shadows to extend to 10,000 units away. Given the additional shaders such as SSAO, volumetric light, DoF, etc. the sun shadows must run pretty well by itself too. This would also be useful even if the sun position was fixed. You could place a traditional light environment with lower brightness and a very large sun spread angle to simulate ambient lighting and radiosity.
My ultimate point is that there are innovators out there allowing Source to be the best it can, but they keep this stuff under the wraps, whether it's on purpose or they don't realize how important this stuff is to people like us. I wish these people would collaborate and bring this stuff together, allowing the entire community to benefit. It's wishful thinking but it is also common sense. Garry's Mod is so popular because it exhibits just a fraction of this idea. With resources such as the above available to everyone who wants to use it in Garry's Mod is just unthinkable.
[QUOTE=Hostel;27044075]I use -staticproppolys, I can't live without it actually ever since I switched to the Orange Box engine.
I'm going to post the following in links since I do not want to clutter too much up.
[url]http://www.empiresmod.com/drupal/node/48[/url] - One of their new additions is a new parallax shader that works on both models and light-mapped materials. It's stable enough to implement into a mod w/ huge maps such as their own. Seeing this also gives me a hunch that you could work specular maps onto light-mapped materials as well.
[url]http://www.youtube.com/watch?v=JYiVkHEzThk[/url] - Almost everyone has seen this by now. He states in his description that contrary to the video, the shadowmaps made from the sun go from a resolution of 3072² to 1024², allowing the shadows to extend to 10,000 units away. Given the additional shaders such as SSAO, volumetric light, DoF, etc. the sun shadows must run pretty well by itself too. This would also be useful even if the sun position was fixed. You could place a traditional light environment with lower brightness and a very large sun spread angle to simulate ambient lighting and radiosity.
My ultimate point is that there are innovators out there allowing Source to be the best it can, but they keep this stuff under the wraps, whether it's on purpose or they don't realize how important this stuff is to people like us. I wish these people would collaborate and bring this stuff together, allowing the entire community to benefit. It's wishful thinking but it is also common sense. Garry's Mod is so popular because it exhibits just a fraction of this idea. With resources such as the above available to everyone who wants to use it in Garry's Mod is just unthinkable.[/QUOTE]
The guy on the video said he was going to release a tutorial on how to do it, just once he gets it, because it doesn't work every single time he does it or something.
what do i think would make a map higher detail?
1. clever brushwork
2. even more clever texture application
3. running the game on a desktop pc with a video card worth its salt.
4. bump mapping textures
5. custome hd textures
6. propper lighting
Multiplayer Hammer.
Will never happen but it's a dream.
[QUOTE=Redcow17;27051371]Multiplayer Hammer.
Will never happen but it's a dream.[/QUOTE]
Instancing. I've done it.
[QUOTE=Hostel;27044075]I use -staticproppolys, I can't live without it actually ever since I switched to the Orange Box engine.
I'm going to post the following in links since I do not want to clutter too much up.
[url]http://www.empiresmod.com/drupal/node/48[/url] - One of their new additions is a new parallax shader that works on both models and light-mapped materials. It's stable enough to implement into a mod w/ huge maps such as their own. Seeing this also gives me a hunch that you could work specular maps onto light-mapped materials as well.
[url]http://www.youtube.com/watch?v=JYiVkHEzThk[/url] - Almost everyone has seen this by now. He states in his description that contrary to the video, the shadowmaps made from the sun go from a resolution of 3072² to 1024², allowing the shadows to extend to 10,000 units away. Given the additional shaders such as SSAO, volumetric light, DoF, etc. the sun shadows must run pretty well by itself too. This would also be useful even if the sun position was fixed. You could place a traditional light environment with lower brightness and a very large sun spread angle to simulate ambient lighting and radiosity.
My ultimate point is that there are innovators out there allowing Source to be the best it can, but they keep this stuff under the wraps, whether it's on purpose or they don't realize how important this stuff is to people like us. I wish these people would collaborate and bring this stuff together, allowing the entire community to benefit. It's wishful thinking but it is also common sense. Garry's Mod is so popular because it exhibits just a fraction of this idea. With resources such as the above available to everyone who wants to use it in Garry's Mod is just unthinkable.[/QUOTE]
I honestly never knew the "-staticproppolys", I'm going to start compiling with that on. Since it's such a slow compile, should I do that even when I'm just testing it, or really just for a final?
[QUOTE=Reactors;27055569]I honestly never knew the "-staticproppolys", I'm going to start compiling with that on. Since it's such a slow compile, should I do that even when I'm just testing it, or really just for a final?[/QUOTE]
-staticproppolys -staticproplighting -textureshadows should be used when posting screenshots or final.
Sorry, you need to Log In to post a reply to this thread.
