Thanks for testing everything out so well. It still has some bugs that need ironing out for sure. The small rocks that don't collide are just part of the groundcover system, but most big rocks close to the road will do damage. Im going to reduce their size, so the close ones look more like large gravel bits. The graphics performance is a relief. I used a similar scheme as my other maps, so trees and other details only extend out to the treewalls.
I see you are looking for hillclimb group B car. Let me do a shameless self promotion: https://beamng.com/resources/pessima-hillclimb-gravel.3673/
I don't have exact stats, but my meager 960 is running triple screens about as well as the Utah map, somewhere around 60fps at mostly low settings. Jimmy Broadbent also just released a video of this mod, so might get some more interest in. Is this mod otherwise dead? The roads still seem a bit to unfortunately bumpy. Are you saying there's not enough precision to smooth them out more?
I do prefer bumpiness over unrealistically smooth (pretty much most maps have way too smooth roads compared how they are in reality), but closer to top some road edges are perhaps bit unfinished and my understanding is that Ewanc is going to smooth those out more. Any road that I have driven 100mph or faster IRL has felt like being on rough seas, quite often not so much corners have limited speed, but roughness of the roads. While purpose built race tracks have been much better, those have still had lot of up and down movement, as much as that effecting lines to choose. Update: As always part of my though is probably not coming trough, bumpiness I think is good, certainly there are then kinda sawtooth noise on road surface, which probably is because of limitations in terrain resolution, but that seems not to affect most of the vehicles too much.
Hey, pretty neat map(understatement), as pointed out by others, and aknowledged by you, some rocks have no collision. On the gravel map, it doesn't show any dust when I'm driving on the road. I feel like that kinda breaks some of the immersion sadly.
The mod was dead, but its cool how you crack a project back open and immediately find solutions. The bumps really worked out perfect for the gravel version. I'm working on a mesh road for the tarmac version, but its going to affect performance adding 250,000+ more hittable polygons, and will take some time to repave the whole track again. The benefits will be 3-fold though; the new mesh road snaps to the LIDAR very tightly, and will reproduce much more of actual road width, banking and undulations, along with the ability to model curbs and walls built along the road. The small rocks near the track and the boulders far away are procedurally generated in the groundcover layer. They scale with your graphics settings in the grass density setting. Being a rally level and the upper section is so sparse, its an easy way of adding immersion. I'd like to add more hittable rocks where cars are likely to slide off the track. The part of devils playground where this car fell off the mountain, I added a bunch of them. I agree about the dust, and had intended on making it dusty gravel, but couldn't find a terrain type for that. The gravel terrain alias I used (dirt_loose_dusty) is only dusty in naming. I was uncertain whether rally tires handle 100% on dirt_dusty so I had to chose gravel to be safe. Ill do some testing to see what the difference is traction-wise. They are hopefully setup so any off-road tire will handle equally well on any non-paved terrain.
You might want to use the roads+properties the devs used on the gravel/dirt section of the automation track, I love the feel and the look of those roads, might need to tweak it to suit the map but it could be a better starting position. Keep up the great work! I kinda want to make some sort of rally/racecar now because of these maps... Too bad that I don't have much free time now. Something like a fictional rally/racecar with an electric motor would be cool, think of an ID-R or maybe a Group S car, so many ideas, so little skill and so little time, and currently motivation
Currently it doesn't feel like bumpy asphalt, but a low quality rFactor-track, where lack of polygons (or somesuch) makes it very uneven, sharp undulations everywhere. Smoothing needed for sure. I do appreciate the overall effort though!
I'm okay with a bumpy road, provided it should be bumpy. Pikes Peak is a relatively new road however and should be fairly smooth, also think about the cars that race up it, very low to the ground with massive splitters. Just think the roads a little too bumpy. My only reference is my own experience with midnight runs down a mountain doing about 75 mph and it's pretty much smooth as butter. Even when I tried regular street cars they were bouncing all over Pikes. And at the top on the hairpins the inside corners get really jagged. Well good to know it'll be reworked, the trackside and the vistas are otherwise pretty exceptional. I think Blijo mentioned it, but the dev's just released... or are working on? a tool that lets you adjust grip and depth parameters for all the surface types so you can really fine tweak it.
Sounds like you already understand that your mesh road will perform very poorly... that's the main thing I came here to say. Others have tried that route and given up on it, the devs don't use it, etc. That and... I do have a suggestion. I'm sure it will upset some people. You should consider folding the map up! As you know, 1m resolution just doesn't cut it in all cases. LIDAR means very little for a racetrack if you can't race on it. One fold would let you double your resolution to 0.5m. Here is an example, I've rotated the bottom of the track -93 degrees using GIMP. The yellow tinted area is a 4096x4096 area, meaning that you can crop your heightmap to 4096x4096 and then rescale that to 8192x8192 and reduce your scaling to 0.5m. If you have better than 1m LIDAR then obviously it would be best to start there, fold the track up, and then correct the dimensions to 8192x8192 @ 0.5m. I'm not very familiar with the track and I didn't check the actual heightmap in this area, but I'd say that the negative effects here are minimal... the chosen area wasn't super straight to start and you can easily fix it up with a 93 degree sweeper (I only added a medium-sized turn in black). The attachment is full resolution but intended as a proof-of-concept only.
Sounds interesting, basically cutting it into 2 maps, but of 2x higher resolution. It would solve some issues, like the jagged edges on cliff side roads and hairpins plus the main issue. I had thought of this option in the planning stage - to just do the Glen Cove to Devils Playground portion at 2048 map size. I do see some hurtles though (redoing decal roads, forest, ect) and I'm trying to keep this from going back to dead project status. Lots to think about and test. Thanks for the good advice.
You're welcome. If you decide not to take that route it will be entirely understandable. OTOH if you graduate from thinking about the idea into the planning stage I'd be happy to give you a hand with chopping things up and putting them back together smoothly. You're right about the decal roads, they would be an issue. I think that the damage could easily be limited to 50% though, the top half doesn't need to move. I believe that a mesh road would require you to do a 100% rebuild of the road, regardless of any folding or not folding. So folding and sticking with a decal road would still save a lot of work vs that. I think that you understood my suggestion just fine, but sometimes it can be hard to communicate on forums. I'll do my best to make 100% sure we're on the same page. I'm suggesting cutting it into two track sections and then rejoining those track sections smoothly. All of your elevation data for the road's path would still be LIDAR/DEM-sourced, except for one 'fantasy turn' you'd add to join the two track sections. The off-highway surfaces would be much less realistic, but we could easily blend the two areas in a pleasing way. This is something I've experimented with previously and I was able to produce good results. The resulting road layout might look like this attachment. EDIT: and the point of all this is to fit everything into a single terrain since the game only supports one terrain at a time.
Your right, I thought you had overlaid the 2 map parts to show how they would scale. Its an ingenious idea, now that I see what you mean, with the new pic. I'm not sure how folding a mile tall mountain would work out but it'd be interesting to say the least.
Awesome, glad we're on the same page now. I'm out of time for now, but maybe soon we can whip up a more detailed proof of concept terrain using all groundmodel layers and the heightmap. That last sentence sounds like a challenge.
But have you attempted to write something on piece of paper at that 75Mph speed? All I'm saying is appearances are deceiving, there are no smooth roads IRL, especially not when speed goes up or above 100mph! Of course I do admit that it would always be better if rough surfaces and bumps would be designed feature instead of terrain limitation etc. However I do like this rough surface with SBR4 as much I like roads at ECA.
I have a request: could you still leave the original, not folded maps in some other configuration or other mod? This map is pretty much one of my first contacts with pikes peak, so I would want to learn and memorise the actual layout rather than one containing the "fantasy corner"
Don't worry, if and when we build the folded version, it will be a bonus layout. I don't think it will even be possible to make it a proper mountain with vistas the way it needs to be configured. It will be your choice to drive the scenic bumpy road, or the tunnel vision high fidelity one.
Nice. I'm glad to hear that. I decided to go for gravel option as my main way of learning the track, so I am not really bothered about bumps.
(speaking as a modder here, not as a developer) I'm kinda skeptic about your idea @torsion, I can't really see how the mountain and terrain bending is going to work, and it will alter the scenery quite a bit. Also, this is the LIDAR version of the climb, which is supposed to be a really accurate version of the track. In my opinion, it wouldn't be really logic to get through all the bother of having it as accurate to at the end add a turn that wasn't there before. It will also slightly change the time needed to drive up to the summit, and kinda not be accurate enough to compare our times with real life. I would be in favor of your first solution Ewanc, aka making the top of the mountain (the more open part) a map and porting every other part of the terrain in a mesh. Yes it will probably make the map size heavier, but I'm not sure if would ruin performance that much. For example, here I have a road mesh I made, with 1 156 085 faces, and FPS is still at 60+ (granted, there is nothing else on the map, just a ground plane, a few cubes and this) : (For those curious, it's 50+km long and climbs at about 5200m (air starts getting so thin the pigeon stalls, normal road cars don't have the go to start from a standing start)) And it would still be possible to split it in multiple smaller meshes with lods to improve performance a bit. And as long as you put the new smaller map at the exact same place as the old one, I believe you could also keep the forest as it is and not have to do it again. The main problem will be the decal roads and the groundmodels, as I don't think decal roads work outside of a map area, and I think it will be a real hassle to specify the texture in the mesh... It's possible but very time consuming I think, or maybe there is an easier way to do it in a mesh I don't see. However, it would be very easy to port the decalRoad part of the existing map on the new smaller map area.
