Update Speculation thread

Can't wait to drift the T series around that

 

I'm curious as to why there are two mappers working on two separate maps, especially with one so close to completion and an update so close. Why do hutch and mitch not combine forces to finish WCA for 0.9.0, then Mitch goes on to help Hutch finish italy. Is distance an issue?

 

They probably are, but sekritly

 

I guess they will. I mean it looks like it could come out fairly soon. I mean they can always leave a couple of empty spaces and finish them later (like JRI)

 

Or the ETK map. It had holes on release.

 

It is not quite that simple for two people to work on same map, because it is more of an art than building a brick wall, they may have very different style to make things happen and combining different methods to single map might add more work than save.

With some aspects surely working together is more easily achieved, but with some aspects it is not so easy in practice.

With some bits like objects that are worked outside of world editor, sharing those might be possible (still, textures might be of different artistic style etc.), but working on world editor, how do you combine two edits to map made in world editor?

Also it is not enough to get map done, it needs to be tested thoroughly with many different kind of hardware to ensure quality and performance level being where it should be, collision stuff especially needs to be tested out carefully etc. So even if map looks complete, it can still be far off from release ready.

 

Motionblur? It already exists, albeit in a buggy way. Go to the postfx folders and there is a document call motionblur which you can edit to be true. It looks glitchy as f.

 

I think collaborations could be done easily in the game. One person works on models and assets, while the other one makes the terrain, roads, and building placements. There is no reason that Hutch can't make buildings while Mitch works on the map.

 

I don't see them teaming up. Each map is their own, and they want to follow through, and make it THEIR OWN ways.

 

I know, but for the d-series which has been here since the beginning it was being a child a Christmas everything that the d-series got better offroad parts, and with the drivetrian update it was awesome.
and now i'm spoiled and want big-ass tires, optional v8 and all the other little goodies right away

 

Might be that they get paid for doing specific map, that would make it bit difficult for teaming up too if payment split needs to be figured out.

I still think that some day they might pull off map generator stuff, then most of the work would be doing those buildings and other objects, clever code would slap them to generated map.

 

Im really excited for the new camera things

 

Now Im not sure what you're talking about.. there was some info about new camera stuff?

 

I'm late to the whole 'procedural map' discussion, but I do think it needs to be given a dose of reality. Map generation does have more of an art to it than just making the assets. Sure, that's a big (hard, annoying) part, but which items are placed where is a more nuanced process that more or less requires a human. Or far more money and time to be put into code than BeamNG would spend in developing five regular maps in the current workflow.

Spoiler: On the Nature of the Limits and Expenses of Procedural Tech

At even the start of it, landscape generation, which is arguably the easiest type of procedural generation, would need some serious work to be done completely automatically and with a short loading period. Most landscape generators that are more than just fractal image generators do some serious and time-consuming calculations in order to approximate erosion, which is integral to making believable mountains, cliffs, canyons, and even beaches to some extent. One would also need to ensure that the map has the appropriate facilities for the landscape tile: no major landscape features, such as mountains, should begin halfway across the map's edge. The map would also need to generate the walls, obstacles, or infinite oceans needed to prevent the player driving off the world. This isn't crucial if you just want the map itself, but if major features weren't kept away from the map's edge, you could conceivably have major landmarks or areas of interest (towns, rivers, fun stretches of road) being abruptly cut up the middle. There's also a real-world logic about where things go: Lighthouses, as an example, can't just go into any map at any random point. Again, code could cure this, but it would come at the cost of 500+hrs. of development time, and untold amounts of player time during map generation.

Post-note: I now do realize you'd meant that terrain generates as the player moves along, but as we can only have one landscape object that can properly interact with the currently-simulated vehicles.

I was kinda hoping it would be a feature that would answer my prayers: https://www.beamng.com/threads/paths-and-ports-the-digital-continent-proposal.27049/

 

I'm thinking it more as a tool for mappers, but I don't see it impossible to have map generation at load time (rather long load time), however it would be simpler to have generation for base maps where roads get carved on map automatically, I mean if map maker now puts road in, brushes along the road to level the map to road, then brushes again to paint decal road, that is whole lot of work that can be automated.

Creating terrain and road can then be seen as bit like how A* finds path, you have to tell what would be limits, code finds possible solutions that you accept or throw away. Yeah it would be lot of work, but it is much of the future, might not happen in BeamNG, but it is just matter of time when someone makes it happen.

Each objects needs a tag and code needs to have ruleset for each tag, tag tells if it is a house, railing etc.

I'm also thinking that providing simple 2d drawing that would be top view of roads and maybe some village size elements could be useful to allow more control.

On other point, we already can import heightmap, why not to have roads carved based on 2d drawing that gets converted to simple spline roads that will follow height of terrain and set limits of grades etc. But of course roads could be made completely automatic way.

Placing village or railing can be easily automated too, when you have objects that have tag as village buildings, you can get random village quite easily, we know how city blocks are made, click insert village here and software adjusts heights, builds streets and places buildings.

For guardrails, we know that they get placed when there is bigger drop next to road etc. Just have program know that too, same way as AI driver knows to stay on road, slow on curve, map making can be set up so that editor knows rules, there is grade larger than set amount for longer than set amount next to road, guard rail is needed, poff, there is guard rail Yeah, takes bit more to get that guard rail to actually follow road, have in correct height etc.

Bob's track builder was really good beginning of more advanced editing, sadly piracy killed the product, but it still is quite advanced in many aspects. If world editor would have even features of that, it would be miles forward, but I'm thinking there is a lot more automation that can be had and automated map creation is huge project, but not quite as impossible as many thinks, random generation just is not correct way to think about it, it needs to be more of generation based by rules and it has to walk trough the road and terrain to know what is possible next and what is not possible next, like A* finds out it's path.

Doing it 3D is then going to be much more interesting

Update: Now I remember what game I have been in mind all the time, Transport Tycoon Deluxe and open source version of that.
Really long long time ago (That was released 1995), it had "3D" terrain (you could rise and lower the land and it was kind of isometric 3D), road and city generation, surely there are only 90 degree turns and only one kind of grades, even terrain resolution is quite low, but that with better resolution and more limits for generation to keep things closer to reality would work just fine for me.

I mean, we have lot of maps with unrealistic roads, having generator do roads within some close level of realistic grades would already be improvement.

Surely human made maps might be better, but also some would be worse than automatically generated, but if in 22 years old game could do what it could do in terrain generation, surely today's machines and advancements in programming should allow same in driving game.

 

This:
Devs work on things that gives better speed sensation, obviously effect in this video is way too much exaggerated
@crazikyle I think that this feature is possible to be in new update, so you may add it to "orange" list + we know that new version will be 0.9.

Simple and rough visualization of roads on Italy map:



@Post update
To give you guys sensation how big Italy map is, just look:


Brigh area is the same size as Utah or Jungle Rock Island (2x2km)
Also the airstrip will be 1.3km (precisely 1320m if my calculations are correct ) long and 50m wide - JRI airstrip have just 650meters and is 30m wide.

 

Obviosly that's not how it's going to be on the game anyway, but I was curious, that was that orbit camera and if it was, it was shaking way too much. I myself prefer having zero shake on the camera, but the orbit behaviour could be polished a bit, since it's a bit too loose and slow to turn, when reversing.

 

Ah, okay, I'm not running on full cylinders now, so I didn't get the joke.

I see, well let's hope that there will be some upgrades to that side. Camera modes are not bad, but they need some polishing definitely.