(Probably) Stupid Question for Map/Object Makers

[TG-55Panthercules]

I played around with maps a little bit in RoF (to create my mod that got rid of the zig-zag forests along NML on the Western Front Map), but I really have no idea what's involved in making maps or buildings or other map-related objects, which is why this is probably going to wind up being a stupid question.  But, I was flying around over a city on the Arras map and saw something that reminded me about something I've been wondering about for a while now regarding why the no man's land in RoF (and now FC) looks so flat and relatively unimpressive.

 

Specifically, I saw this:

 

and it got me wondering - are these buildings each separate objects, or one large object?  And regardless of the answer, is there some size limit on how large a building-like object can be, or how many such objects you can have on a map?

 

The reason for these questions is that I saw somewhere that someone was saying that (one of) the reason why we couldn't have 3-D trenches was that the game engine won't let you put things down below the surface of the map/ground.  But that got me wondering whether it might be possible to create an interesting-looking 3-D NML out of a series of objects that would basically be large one-story "buildings" placed next to each other, whose roofs would look like mud and craters and stuff that you now see on the surface of NML, with some barbed wire obstacles along some of the edges, where the spaces between these "building blocks" (like the spaces between the buildings in the screenshot above) would be the trenches themselves.  Of course, the land on both sides of the front would have to be configured to slope up slightly so these NML building blocks could sit down in the valley that would create, but if done gently enough maybe that could work?

 

There's obviously something wrong with this idea, or I'm sure we'd have 3-D trenches by now, but I'm curious as to what the problems would be.

 

 

 

 

 

 

[Raptorattacker]

Now THAT'S what I call thinking outside the box!! Interesting idea is that.

The only possible snag I could see with that woulld be the lighting cast being oposite to everything else.

[AEthelraedUnraed]

Right, it's a good question, and I honestly like it when people come up with unconventional answers - worst case is that someone learns something.

 

On 4/4/2023 at 2:56 AM, TG-55Panthercules said:

[...] it got me wondering - are these buildings each separate objects, or one large object?  And regardless of the answer, is there some size limit on how large a building-like object can be, or how many such objects you can have on a map?

They're separate objects within one "game object". Whether this requires one additional matrix multiplication ("sub-object space" to object space) or if they're baked into one actual object (or at least a shared space), I don't know. I *think* the size limit of sub-objects is 128, but I could very well be wrong here. I don't think there are any specific limits on the world space size, but if it gets too large you could naturally run into engine or shader limits, and besides that it might hinder performance.

 

On 4/4/2023 at 2:56 AM, TG-55Panthercules said:

There's obviously something wrong with this idea, or I'm sure we'd have 3-D trenches by now, but I'm curious as to what the problems would be.

Honestly, I love comments like this Most people suppose, without any prior knowledge or experience working with either/all of 3d modeling, shading, AI programming or even programming in general, that because something works in [insert completely unrelated game], it should also work in IL2, and that with only minor work. Not only do you explicitly keep open the possibility that you're wrong, but you also actively express a desire to gain knowledge in case you are. I hope I can at least answer some of your questions regarding the problems - if not, just let me know and I'll gladly clarify

 

On 4/4/2023 at 2:56 AM, TG-55Panthercules said:

The reason for these questions is that I saw somewhere that someone was saying that (one of) the reason why we couldn't have 3-D trenches was that the game engine won't let you put things down below the surface of the map/ground.  But that got me wondering whether it might be possible to create an interesting-looking 3-D NML out of a series of objects that would basically be large one-story "buildings" placed next to each other, whose roofs would look like mud and craters and stuff that you now see on the surface of NML, with some barbed wire obstacles along some of the edges, where the spaces between these "building blocks" (like the spaces between the buildings in the screenshot above) would be the trenches themselves.  Of course, the land on both sides of the front would have to be configured to slope up slightly so these NML building blocks could sit down in the valley that would create, but if done gently enough maybe that could work?

There's a couple of problems with the "building" setup:

- You've got the terrain underneath anyhow, resulting in at least one additional Z-buffer write.

- The terrain is likely highly optimised; you won't be able to do similar optimisations for actual 3d models.

- Likewise for additional terrain features such as grass. Enabling that on some objects could very well mean making either the terrain or grass shader less efficient overall.

- You'd need to align the several UV coordinates (one option would be to use world vertex positions).

- Likewise for vertex positions.

- Each object would require its own LOD levels, resulting in sub-obtimal performance especially for faraway objects. (For the terrain objects, you could simply merge quad vertices to get a lower LOD, as it's one object/vertex grid. This isn't the case for multiple objects, each of which would require their own LODs.)

- You'd be very limited with regards to crater/trench structure; in essence, every possible combination would need to be made, including terrain where necessary.

- Basically almost all of the above again, but then for the ground colliders.

 

Now, there are a couple of much better options. One could, for instance, write a fragment shader that emulates a 3d effect without any actual real 3d vertex shader stuff going on. Or, one could generate additional vertices if the player is close enough, and then do the 3d stuff in the vertex shader. If necessary, you could even keep it out of the shader altogether and generate additional actual terrain vertices where necessary. Note that the current terrain consists of quads, which are quite easy to subdivide to get additional resolution - provided of course you've got some means to generate that!

 

 

Edited April 6, 2023 by AEthelraedUnraed

[Jaegermeister]

That idea seems like it would be possible with the existing terrain structure, but it would have to utilize objects like the cliffs on the Normandy map. Trees and other objects on ground level show through it. As with the cliffs, I'm sure there would be issues matching them to the existing ground texture. I think they would need to be custom built to the ground elevation, meaning you might as well have built in the trenches in the map in the first place. With the small amount I have experimented with editing ground textures, it appears that you can't make vertical surfaces like you can with the 3D cliff objects. 

 

Then there is the issue of the random explosions through the trench "objects" as well.

[IckyATLAS]

It seems that to progress further into better maps, the map technology will have to be changed.

I saw recent demos of the latest version of Unreal 5.2 engine (that would be excellent for TC) and map making possibilities.

We are now light years of this quality. I am afraid that if IL2 does not improve its maps it may loose the visual attractivity. And the no mans land simulation in FC is a blatant example. But there are others too. Simulation of destroyed cities building and true 3D map craters. Vegetation, trees etc. that stay nice and fresh even if you let a 1 ton bomb on it and burn the hell out of the place. The problems to have realistic connection between sea land and ships sinking in the shore like in moving sands, all these strange behaviors show the limits. The Kuban map remains superb, but the sea simulation, water, lakes, rivers all show the limitations. As gamers eyes and brains will become more accustomed to the visual quality of what recent sim technologies offer, what we get in IL2 may become a little dull and primitive and so less attractive.

When I compare IL2 Sturmovik original maps the technology was dated from 22 years ago. Rise of Flight was released 14 years ago and had improved map quality compared to the initial IL2. IL2 Great Battles dates from 10 years ago and its map quality improved a little over time. But the old engine limitations  put a limit to how much better it could become, before needing a complete technology change.

To give you a hint about where the map making technology stands as well as physical simulation here is a nice tech demo enjoy:

 

[Guest]

44 minutes ago, IckyATLAS said:

It seems that to progress further into better maps, the map technology will have to be changed.

I saw recent demos of the latest version of Unreal 5.2 engine (that would be excellent for TC) and map making possibilities.

We are now light years of this quality. I am afraid that if IL2 does not improve its maps it may loose the visual attractivity. And the no mans land simulation in FC is a blatant example. But there are others too. Simulation of destroyed cities building and true 3D map craters. Vegetation, trees etc. that stay nice and fresh even if you let a 1 ton bomb on it and burn the hell out of the place. The problems to have realistic connection between sea land and ships sinking in the shore like in moving sands, all these strange behaviors show the limits. The Kuban map remains superb, but the sea simulation, water, lakes, rivers all show the limitations. As gamers eyes and brains will become more accustomed to the visual quality of what recent sim technologies offer, what we get in IL2 may become a little dull and primitive and so less attractive.

When I compare IL2 Sturmovik original maps the technology was dated from 22 years ago. Rise of Flight was released 14 years ago and had improved map quality compared to the initial IL2. IL2 Great Battles dates from 10 years ago and its map quality improved a little over time. But the old engine limitations  put a limit to how much better it could become, before needing a complete technology change.

To give you a hint about where the map making technology stands as well as physical simulation here is a nice tech demo enjoy:

 

Very pretty. To run that in an air-sim (or any other actual game) you'll need an Intel i26 27900k and an RTX 12080ti.

[AEthelraedUnraed]

...aaaand we're back at discussing engines.

2 hours ago, IckyATLAS said:

It seems that to progress further into better maps, the map technology will have to be changed.

I saw recent demos of the latest version of Unreal 5.2 engine (that would be excellent for TC) and map making possibilities.

We are now light years of this quality. I am afraid that if IL2 does not improve its maps it may loose the visual attractivity. And the no mans land simulation in FC is a blatant example. But there are others too. Simulation of destroyed cities building and true 3D map craters. Vegetation, trees etc. that stay nice and fresh even if you let a 1 ton bomb on it and burn the hell out of the place. The problems to have realistic connection between sea land and ships sinking in the shore like in moving sands, all these strange behaviors show the limits. The Kuban map remains superb, but the sea simulation, water, lakes, rivers all show the limitations. As gamers eyes and brains will become more accustomed to the visual quality of what recent sim technologies offer, what we get in IL2 may become a little dull and primitive and so less attractive.

When I compare IL2 Sturmovik original maps the technology was dated from 22 years ago. Rise of Flight was released 14 years ago and had improved map quality compared to the initial IL2. IL2 Great Battles dates from 10 years ago and its map quality improved a little over time. But the old engine limitations  put a limit to how much better it could become, before needing a complete technology change.

To give you a hint about where the map making technology stands as well as physical simulation here is a nice tech demo enjoy:

Unreal's Nanite system looks fantastic, but you need the development capacity to create all those 3d models as well as implement them in-game. I expect little but triple-A developers to start using them. Which 1CGS isn't.

 

I agree that the IL2 terrain (and foliage) needs to be upgraded. But once again, I don't see how "old engine limitations put a limit." All that's needed is some shader rewrites, and some additional textures. Right now, the terrain uses a single texture with its alpha map used as a kind of pseudo-heightmap used to create shading effects. Kinda clever, but since this is the only texture, it's rather limited in what can be done. With one additional texture, you could use normal mapping (compressed into two channels) and then you'd have three additional channels for things like specularity/roughness, grass coverage or true heightmapping.

 

For the trenches; one could likely get reasonable results with parallax mapping since trenches are basically self-occluding. Again would require an additional channel, but the positive side is that it could retroactively be applied to the already existing trench system with little extra work.

 

Is there any indication whatsoever that the above improvements would not be possible in the current engine?

Edited April 5, 2023 by AEthelraedUnraed

[1PL-Husar-1Esk]

2 hours ago, AEthelraedUnraed said:

...aaaand we're back at discussing engines.

Unreal's Nanite system looks fantastic, but you need the development capacity to create all those 3d models as well as implement them in-game. I expect little but triple-A developers to start using them. Which 1CGS isn't.

 

I agree that the IL2 terrain (and foliage) needs to be upgraded. But once again, I don't see how "old engine limitations put a limit." All that's needed is some shader rewrites, and some additional textures. Right now, the terrain uses a single texture with its alpha map used as a kind of pseudo-heightmap used to create shading effects. Kinda clever, but since this is the only texture, it's rather limited in what can be done. With one additional texture, you could use normal mapping (compressed into two channels) and then you'd have three additional channels for things like specularity/roughness, grass coverage or true heightmapping.

 

For the trenches; one could likely get reasonable results with parallax mapping since trenches are basically self-occluding. Again would require an additional channel, but the positive side is that it could retroactively be applied to the already existing trench system with little extra work.

 

Is there any indication whatsoever that the above improvements would not be possible in the current engine?

I think they don't have any ideas how  to add those additional textures, or don't have resources, time, employees with knowledge. They focused only on planes  not environment ,excluding the clouds (which could have better definition, because up close they look to uniform , hard to judge the distance because lack of futures) . Look how artists polished ROF with those technology at time, it's big difference. 

Edited April 5, 2023 by 1PL-Husar-1Esk

[IckyATLAS]

13 hours ago, Hetzer-JG51 said:

Very pretty. To run that in an air-sim (or any other actual game) you'll need an Intel i26 27900k and an RTX 12080ti.

This is just a few years away. In three four years which means tomorrow we will have it.

11 hours ago, AEthelraedUnraed said:

Is there any indication whatsoever that the above improvements would not be possible in the current engine?

This is what goes around, but I do not think that the devs have formally stated that. Some in the forum maybe know the engine and its limitations. In any case I have no idea about the engine nor its limitations. What I see is very little evolution or improvement over time. There is some clear improvement like for the sky and clouds and rain, that has been a real step forward, but on the ground map nearly nothing has moved, except maybe the special section of the map for TC on the the Prokhorovka map where some more details were made in the ground and that was already long ago.

[Jaegermeister]

40 minutes ago, IckyATLAS said:

This is just a few years away. In three four years which means tomorrow we will have it.

This is what goes around, but I do not think that the devs have formally stated that. Some in the forum maybe know the engine and its limitations. In any case I have no idea about the engine nor its limitations. What I see is very little evolution or improvement over time. There is some clear improvement like for the sky and clouds and rain, that has been a real step forward, but on the ground map nearly nothing has moved, except maybe the special section of the map for TC on the the Prokhorovka map where some more details were made in the ground and that was already long ago.

 

The detail enhancements on the Prokhorovka map mainly had to do with 3D modeling of the buildings and objects in the detailed area. They have a damage model and the tanks can interact with them without hitting a proverbial 2D brick wall with nothing inside. It is not considered necessary to have that type of model for all objects when you are flying a plane. The only time you would see them is when you crash, and that would be very brief I believe. The craters are still amorphous above ground patches that you can drive through similar to the trees.

[RossMarBow]

8 hours ago, Jaegermeister said:

 

The detail enhancements on the Prokhorovka map mainly had to do with 3D modeling of the buildings and objects in the detailed area. They have a damage model and the tanks can interact with them without hitting a proverbial 2D brick wall with nothing inside. It is not considered necessary to have that type of model for all objects when you are flying a plane. The only time you would see them is when you crash, and that would be very brief I believe. The craters are still amorphous above ground patches that you can drive through similar to the trees.

Ah didn't know that.
Good thread #nostupidquestions

[AEthelraedUnraed]

On 4/5/2023 at 12:50 PM, 1PL-Husar-1Esk said:

I think they don't have any ideas how  to add those additional textures, or don't have resources, time, employees with knowledge. They focused only on planes  not environment ,excluding the clouds (which could have better definition, because up close they look to uniform , hard to judge the distance because lack of futures) . Look how artists polished ROF with those technology at time, it's big difference. 

The shader techniques I mentioned are pretty common knowledge nowadays and well-documented online. I'm pretty certain the problem is resources and time rather than knowledge

 

15 hours ago, Jaegermeister said:

The detail enhancements on the Prokhorovka map mainly had to do with 3D modeling of the buildings and objects in the detailed area. They have a damage model and the tanks can interact with them without hitting a proverbial 2D brick wall with nothing inside. It is not considered necessary to have that type of model for all objects when you are flying a plane. The only time you would see them is when you crash, and that would be very brief I believe. The craters are still amorphous above ground patches that you can drive through similar to the trees.

I think it's also worthy of note that the Prokhorovka map uses a much more detailed heightmap, as visible in the image below (left: a gorge on the old Stalingrad map, right: a similar gorge on Prokhorovka). The better heightmap doesn't solve the problem of 3d trenches or even craters though, as even the Prokhorovka heightmap isn't nearly detailed enough to render those.

 

15 hours ago, IckyATLAS said:

This is what goes around, but I do not think that the devs have formally stated that. Some in the forum maybe know the engine and its limitations. In any case I have no idea about the engine nor its limitations. What I see is very little evolution or improvement over time. There is some clear improvement like for the sky and clouds and rain, that has been a real step forward, but on the ground map nearly nothing has moved, except maybe the special section of the map for TC on the the Prokhorovka map where some more details were made in the ground and that was already long ago.

Well, what goes around are mainly hyper-sensationalised comments from people who don't necessarily know what they're talking about and which often goes directly against past experiences - e.g. the relative speed with which the improved sky or even the improved clouds were developed suggests it isn't particularly hard to rewrite shaders in the current engine. Those very few who have any direct experience with the engine keep their mouths firmly shut.

 

As you say yourself, the sky and clouds have really improved a lot. I think the ground is/would've been next in line (depending on whether or not they plan to continue with (a hugely improved version of) the current engine in the next project). Likewise for foliage (both grass and trees/bushes) that also could do with some TLC. As an example of what could be done using relatively common shading techniques, I created this small proof-of-concept no-man's-land inside Unity using its standard shader and minimally edited game textures. Note that although it's a Unity shader, the techniques themselves aren't anything special. This is strictly a fragment shader, so no extra geometry is created.

[danielprates]

On 4/4/2023 at 6:24 AM, Raptorattacker said:

Now THAT'S what I call thinking outside the box!!

 

Thinking above the ground level!