Update Speculation thread

Nice,,but I still think they really are postponing Italy.

 

I still think my flawless logic holds up.

 

Nest, is it like nesting update then? With hatchlings and all, that would mean dragons, which game desperately needs

Probably, maybe, who knows

 

I fixed it just as you posted that.

 

Or convertible roofs, because it kinda looks like an egg hatching when you open one.
--- Post updated ---

Agree to disagree?

 

Awwww, no dragons update

But on serious note, with improved gravity, if you have a planet that is 500-1000 meters diameter or so, it might work quite nicely, you could use some underground for more roads, skybox might provide some challenges though, but it could mean going around in different way than we are used to.

If one would build a city of such, it could appear a lot bigger than it actually is.

 

*dun dun*

 

being open about things like this, is a good thing imho.

 

I see you can not grasp my intellectual capability of tying s'mores to Mussolini to whether or not Italy is coming out next update.

 

In regards to @tdev 's post regarding the Italy status...

I'm a little bit confused. Italy should, if it's using LJFHutch's method of map design, run faster than West Coast USA, as it would properly use LODs and forest objects.

Also, it seems that the "Autobello" car would be released alongside Italy, because it is likely Italian, so we might not see that in 0.14 as well.

Following the "Normal Update Schedule", if the next update is around November/December, Italy might come in March. I'll admit, I'm pretty disappointed to say the least.

But with both the "Autobello" and Italy likely out of the running, what could the next update include? There really isn't much to go on.

The Bluebuck has just been restarted, so there's a snowball's chance in hell for it being ready in time for the next update. We have gotten no word on the status of the Articulated bus. So, at this point, there really isn't much to go on. It's a bit disappointing to have been waiting for a major BeamNG related update since 0.12, and I suspected that there was other content being worked on alongside the Automation collaboration, mainly Italy and the "Autobello", but as of now, there's really not much to go on.

 

We will probably see performance and/or physics improvements, which is always nice to have, maybe even some more sounds to go with that? Especially seeing the latest update post about those... Not saying I'm disappointed about new vehicles/maps not being added in 0.14 since I'm really not, but it would be nice to see atleast some more parts for the current cars we have ingame

 

The issue is that the last update only added a smaller map based on another video game. The last "original" BeamNG map was released nearly a year ago.

The big factor is the "Autobello", and if it will be released on it's own or pushed back to release alongside Italy.
--- Post updated ---

There was no evidence. Don't cite speculation as evidence.

 

why does WCA run so poorly? is it a LOD issue or something? My PC should easily run that map, but once I get within viewing distance of any of the city parts, the game lags and has frequent slowdowns and fps spikes. I rarely use it solely for that reason. that being said, hopefully Italy will run better, more optimized at least. I wouldnt even mind at all if it takes another 3 or 4 months, as long as the extra time is used optimizing it and perfecting it. No sense in rushing out content, especially official content, if it appears to be half-assed.

 

The city is mainly made of 2 meshes, all of which are always loaded 24/7.

 

isnt that wildly inefficient in terms of gpu load?
--- Post updated ---
i mean I run a 1050Ti, so not the most xtreme card but definitely every other game I play with ultra graphics loads them 50-60fps no problem

 

It is extremely inefficient, but that's just how the map was made. /shrug

 

and so I guess the way hutch is doing italy, it will have LODs etc, so that I am excited for, but that inherently takes probably 2 or 3x the time to complete. will definitely be worth the wait.

 

It is bit more complex than just number of polygons, issue is not so much of GPU performance, but CPU, which can actually be made worse by LODs.

While CPU IPC has had some improvements recently, also clock speeds in practice have declined, there is more cores, but those are not using turbo clocks too much, base clocks are not really increasing, more of decreasing.

In other hand GPUs have improved steadily, which creates situation where games have been facing more challenges of CPU bottlenecking by graphics than by GPU for years.

20 Million polygons is nothing for good GPU of today, fastest ones can do 50 million polygons or maybe even more, but only if GFX engine and DX/Vulkan is not causing CPU to bottleneck first.

It is very big challenge to keep GPU fed with this stagnated CPU IPC improvement.

So, what they are going to do is finding ways to keep GPU fed, even it would mean graphics being more expensive for GPU budget, it is really only way to open up more performance.

It is also possible to improve rendering so that less is rendered, which would help both CPU and GPU, but it might not be enough, especially on hilly landscape that offers great overviews over map, only way is to make sure hardware gets used properly.

LODs I think might be thing of the past soon, or at least they might get used in a future much less visible way as there is not much benefit of going very low LODs, it is much better to ensure stuff gets batched up more. For this of course DX12 and Vulkan gives little help, but still modern fast GPUs want lot of polygons.

I'm not exactly sure what is the limit, but it used to be something like 300 polygons per drawcall 10 years ago and back then it was increasing really fast with new GPUs, today it might be something far more, so LODs can be really just waste of time and performance.


WCUSA is running something equivalent of visual collisions, if you turn collisions off, CPU load decreases a lot, but it would not run on any machine if it would not be combined to such large objects, which I believe is temporary measure to make it run for now.

Those large objects are not a problem in WCUSA, lack of LODs is not a problem either, sure 1050Ti does not like it, but 1070 and 1080 level cards run that map just fine.

I speculate that eventually WCUSA gets proper collisions, when they can put buildings back to single pieces, or if they determine it is fine to keep them combined, that probably helps some, but effect on GPU load is not very huge really.

Also what complicates matters is that it is more of surfaces, UV mapping, textures that affect performance than objects, but having more in one object can help to have more shared surfaces, which again is good for performance.

With current engine, put enough many textures, poor UV mapping (performance wise) and even under 3 million polygons you can have a map that is not going to run on any CPU in existence, LODs are not going to help at all, probably just hurt performance in that case, that is something people don't often know or realize.



Anyway point is that old fashioned less polys = better performance thinking is just not working anymore, WCUSA is not poorly made in terms of performance, sure it's current temporary state is having bit of performance impact, but it really is made to perform best possible with available CPU power on current engine, you can't have it running without tapping on GPU bit more than you would like to.
With all separate buildings, LODs etc it would not run even near as well as it does now.

Still even with trickery it is CPU heavy too, but at least we can run it with current CPU technology.

 

you forgot to carry the one...