Read this post if you haven't already: https://gtm.steamproxy.vip/groups/vtmod/discussions/0/1693785035830353150/

The short answer is that, with all of the proposed mod tools, we could pretty much add anything to the game if 1) it's supported by the Stingray engine, and 2) allowed under the terms of the to-be-announced EULA.

What makes it into 'approved mods' is also yet-to-be-decided, but we know it won't include cheats and griefing. Unapproved mods could still be anything covered by 1 and 2, they just won't talk to Fatshark's backend for items, cheat protection, and unlocks.

Originally posted by Aussiemon:

Read this post if you haven't already: https://gtm.steamproxy.vip/groups/vtmod/discussions/0/1693785035830353150/

The short answer is that, with all of the proposed mod tools, we could pretty much add anything to the game if 1) it's supported by the Stingray engine, and 2) allowed under the terms of the to-be-announced EULA.

What makes it into 'approved mods' is also yet-to-be-decided, but we know it won't include cheats and griefing. Unapproved mods could still be anything covered by 1 and 2, they just won't talk to Fatshark's backend for items, cheat protection, and unlocks.

Yeah I'm pretty optimistic. I would definitely be excited with a level editor or entire map-making being possiblt. Other than that I think if you look at all the mods for Vermintide 1 that people have gotten super creative with, there's definitely a lot of potential, especially if the lobbies being split!

So the whole game source will be available, or do I misunderstand?

Originally posted by Chiptune Classic:

So the whole game source will be available, or do I misunderstand?

Everything but the engine code. Essentially all gameplay is written in lua, so all that would be available. I have no ETA for this though. Hopefully it will be when the Workshop goes live (aiming for late april), but I'm starting to get worried it won't get done in time.

Thats excellent. I'm a little crazy, so I'm curious if it would be possible to generate levels at run time.
Is that stuff too deeply nested into the engine to get into via lua?

If I can create a mesh object from a set of points, I can probably go ham from there

Originally posted by Chiptune Classic:

Thats excellent. I'm a little crazy, so I'm curious if it would be possible to generate levels at run time.
Is that stuff too deeply nested into the engine to get into via lua?

If I can create a mesh object from a set of points, I can probably go ham from there

That's not quite possible, though it is possible to spawn units (models with physics, materials, etc) wherever you want during runtime.

So you could in theory then, use that to build simplified levels, but I'm guessing you'd be severely limited on things like lighting which is probably a show stopper in itself. It would certainly defeat some of the advantages of trying to generate levels for VT I was hoping for, like the excellent rendering and lighting systems it has.

Well, this dims my hopes a little, but I'll certainly be keeping an eye on mods and seeing what others are doing

Well if models can be spawned then you could "simply" make yourself a tile set and write a hefty logic in lua to govern how the tiles are supposed to be generated and connected.

Real question is how efficient would it be to do that at runtime in current engine, plus whether AI can actually handle such levels without pre-generated navmesh.

I imagine that best viable option would be to have level randomizer features akin to Left 4 Dead 2 where certain parts of the map had roadblocks that opened/closed randomly during each playthrough giving people _slightly_ different path each time.

It's a good idea, the question about efficiency is an important one.

Whether or not the AI can handle a dynamic environment for pathing is a good question.

The problem with level randomization like L4D2 is that it still felt the same, I never noticed the different paths to any significant degree.
It's certainly worth investigating.

Also, consider this my official request that you guys investigate if it would be possible to allow run-time level generation as part of the modding api. I understand if its not possible though, its not small undertaking, and even in engines I play in there are limits for what you can do with level generation.

I wouldn't be surprised if the level generation required a series of baking steps, which wouldn't be accessible from the game engine itself.

If its simply not possible to ever allow it, then I am sad, but I can look at these other lesser approaches and see how much of it would be viable.

Define "investigate" even with no support whatsoever you could spawn players in a premade black box, fade in and teleport them to proper start area then fade out once level is finished generating.

Given that would be 100% on lua side you'd know when it's done so it's okay to "start" the level.

Well, I suppose I'm being a little presumptious. Perhaps the level system is all based upon models, and if so, I'd imagine the lighting system works fine with it. I didn't get that impression though.

While you could certainly approach it the way you suggest, I suspect there will be loss of visual quality.
Vermintide 2 is a beautiful game, the models, textures, and lighting are all on point.

It would be ashame to lose that quality for dynamic levels is all.

As far as i can tell it's a mix of both dynamic and prebaked lighting.
If the engine/tools are any decent or flexible it would be entirely possible to get decent visual quality using only dynamic lights and whatnot however performance cost of that would be way higher than for a traditional map.

Global sun shadow has both dynamic and static shadowmap. I assume dynamic one is created once at runtime for static objects at given resolution and then just kept in memory.

Dynamic sun shadow applies for everything that isn't 100% static so probably it would be "possible" to compensate using it if you had your game set up at high values for the shadowmap resolution.

Local ligths are packed into two atlases, one i assume is current one which consists of lights are being updated for given frame and cached ones which aren't being updated. Resolution is determined for entire atlas so each light only takes a portion of entire atlas resolution which depending on amount of visible lights can severely cripple their fidelity.

Technically it should be possible although it's not something i would really recommend.

Also it occurred to me that game is using Umbra. I'm not sure but usually it requires a separate scene to be exported to support occlusion properly.
Firstly i don't know whether creating that will even be supported in mod tools due to licensing and secondly even if it is, i doubt it will work well with stuff generated at runtime.

You can definitely do runtime level generation, just not at the mesh level very efficiently. You can defintely spawn sub levels or props completely dynamically through lua scripts. There are a ton of other really hard challenges you'll run into though, like how to stitch together navmesh runtime if you want ai to be able to move between different prefab sections. The Ambient Occlusion solution is currently prebaked, but you don't have to have AO. All other rendering techniques that I can think of are done runtime.

Procedurally making levels isn't impossible, it's just a very big challenge, otherwise we would've done it already :) (we have some procedural elements, but they're very limited and rely heavily on handcrafted logic and hand placed assets to work).

...oh yeah, you wouldn't be able to use umbra for occlusion. But if your map is procedural you should in theory be able to calculate occlusion yourself and just not spawn/show stuff until the player is there.