SOLVED How can i add bumps and better cambers to my track on 3ds max.

[LilSKi]

Well as I said it really isn't very hard but maybe I say that as I have done it a few times already. My Bridgehampton track was my first track ever and it took a good amount of experimentation to get the surface right. In the case of Bridgehampton the road is very wavy and bumpy so there is multiple levels of noise in the surface. The physical is originally a duplicate of the visual. I apply the noise and subdivisions to the physical. Then I shrink the visual back down on the new physical. Without the final shrink of the visual on Bridgehampton there were times it would swallow almost a half a tire of a car. This is how I continued to do it with both Riverside and now NJMP which is LiDAR based.

For a basic noise yes you can leave the visual alone and the visual of the car going in and out of the track is minimal. I have done it for a few conversions and it works fairly well. But as with Bridgehampton there are cases that it is way to much to be tolerated.

If you are not familiar with the track take a look at this video.

This one is good too

 

[Mr Whippy]

Ah well yes, you're adding large noise there, up to say 25cm worth.

So deviation from the visual mesh means you need to take the visual back to within the range of the physical.

But in essence you could have added that large scale noise to the visual (big noise) and added the finer noise to wheels again directly, not via a fine separate physics mesh mesh.


It's not really that important. Like I said, it'll have been ACs approach due to early lidar builds.
They want to show all that detail of their scans so they run with a dense physics mesh.


It's just this thread show how complicated a concept it is to grasp for those without that mesh... and how it feels like extra work for nothing when all they do is smooth their visual mesh then add noise.


Maybe it's a feature AC could add in future.
I'd personally not bother given the choice but only will because I have to.

 

[LockeNessMotorsports]

We're getting there now I'd say aha only thing is I need more promenant cambers the ones I have by dragging the verts down aren't working the best.

And if you add too many edges in it gets all chaotic and if they don't line up it makes squares on a meshsmooth pretty much.

I was looking on youtube and found a video on blender and they were using ctrl t or a tilt tool but idk if that exists in max.

 

[Mr Whippy]

You could try soft selection.

Or select the edges on the inside or outside of a bend, convert to spline, optimise the spline, then tweak the spline to suit the curve.

Then manually set verts to that splines 'smooth' z info.

Ooor, rebuild the corner using a spline cage and 'surface' modifier.

That way you have some nice control and can sub-divide it to any smoothness level you want.

 

[Pixelchaser]

I'm quite lucky as real roads are really quite smooth. however some time ago i tested a function in RTB. It has a randomiser function for every vert on a road. it will displace randomly over the road piece by a factor of 1 cm minimum(X&Y). this is how I'm going to deal with this malarkey, my visual will be attached to the physical exactly. this is what my visual attachment looks like. 9 mins onwards I drive on a section I was testing this. I think its pretty good. this was a test at 2 and 5 cm bumps. obviously way to bumpy. it needs more design into it which will happen. but its a mammoth undertaking over my length of course so struggling to get to that. but between this rtb function and further max subdivide I hope to get a nice feel but it will be heavily optimised to just give a feeling as I'm working with extraordinary amount of poly for it.

 

[Mr Whippy]

I'd do physics once everything is finished.

That way you can add in details for decals, cats eyes, pot holes, whatever you want.

Trying to do physics as you work along on an iteratively designed course is a waste of time I'd say.

Unless you have real data you're conforming to of course, but then they're not iteratively designed.

 

[LockeNessMotorsports]

We're getting there now I'd say aha only thing is I need more promenant cambers the ones I have by dragging the verts down aren't working the best.

And if you add too many edges in it gets all chaotic and if they don't line up it makes squares on a meshsmooth pretty much.

I was looking on youtube and found a video on blender and they were using ctrl t or a tilt tool but idk if that exists in max.

I'll try , never heard of a spline cage before. The spline method sounds like it might work I'll try some methods out.

Edit, it won't let me convert edges to a spline and what did you mean by smooth selection on z axis sorry still learning

 

[Pixelchaser]

I'd do physics once everything is finished.

That way you can add in details for decals, cats eyes, pot holes, whatever you want.

Trying to do physics as you work along on an iteratively designed course is a waste of time I'd say.

Unless you have real data you're conforming to of course, but then they're not iteratively designed.

I agree. but whats gets me in this conundrum is the fact that at the end of the day, that physics layer is the most important thing in the track for our purposes. yet it cometh last. its seems not logical.

live long and prosper.
.
.
.
and never worry about when the physics layer comes .

 

[LockeNessMotorsports]

So playing around with high and low poly on the track, i like experimenting anyway look at this difference with this low poly is allot better after a mesh smoothing is added Here's a comparison between a high poly/ more edges in a corner and a lower poly one each of these has 3 iterations of smoothing but look at the difference. High poly first then low underneath.

 

[Pixelchaser]

that is indeed the edges you require. there is one still cornered at the top and left side. the trick is doing that without smoothing and getting those edges as curves thus retaining the low poly nature.

spline and loft and you would get to dictate your spacing properly. the method you are using will take you twice as long over spline and lofting.

 

[Mr Whippy]

When you select edges, there is a button in the edit panel called 'created spline from selection' or something.

Once created, you can adjust that to how you want it in the z axis.
Then you can tweak the mesh verts in z to match the 'smooth' spline.

 

[Mr Whippy]

So playing around with high and low poly on the track, i like experimenting anyway look at this difference with this low poly is allot better after a mesh smoothing is added Here's a comparison between a high poly/ more edges in a corner and a lower poly one each of these has 3 iterations of smoothing but look at the difference. High poly first then low underneath.
View attachment 1455



View attachment 1454

But your added edge loops in the high density mesh are all straight between the same few edges present in the low poly mesh.

Stop trying to do everything in one big stack.

You'll never achieve a full non-linear building process in Max for something like a track.

Sub-d/mesh smooth are a pita because verts aren't explicitly intersected.


My 2p.

Outline the inner and outer line of the track with splines. Get it perfect.

Create the opposing edge using outline, and tweak (no new knots ideally) to fit.
You can now create a mesh over this. Set the spline to 0 smoothness (iterations)
You can snap easily to that spline.

Then up the spline smoothness, add more edge loops and adjust to fit.
Keep going till you have about 2m intervals, maybe 1m.


Ahhh I dunno.

Loads of approaches. Some are better earlier, but worse later.
Some vice versa.

Simple reality is that there is no 'easy' way.
Get a solid fixed outline down in splines and just work at it till you have a nice smooth flowing mesh that works around the course.

You could literally draw each triangle one by one and get a good result if you wanted.

I've made about 15 full tracks and tens of WIP layouts and each one is done differently based on my needs.

They almost always have a really solid spline outline as a base though!


Dave

 

[LockeNessMotorsports]

Idk I'm still confused if you make a spline off the edges of the track it's just on it's own you can't tweak the track mesh using that spline, or did you mean to assign a mesh to the spline. That's the problem with the low poly in one way it's hard to get all the bloody cambers etc right now but i don't want to and add loads of edge loops and ruin the smoothing again so I'm stuck at the minute.

 

[Mr Whippy]

No the splines are a guide.

You can snap to them though.

Honestly you sound like you're over thinking things.

You'll *always* lose the smoothness adding loops, that's why you need a solid outline to move the new points to.

The idea is to create the right amount of loops in the loft, sweep, or whatever other initial process you use, for then making the next stage work.


Or how about creating a rough but close top-down sub-d mesh that is just smooth for z info, but has a poor flow for visual use... but good for tweaking camber etc.

Then conform a perfect top down mesh to that z.mesh.

That way you deal with z data on a mesh optimised for that (camber, height etc), and do the visuals with a mesh optimised for that.


You will *NOT* do everything in one uncollapsed mega stack of modifiers in 3DS Max or Blender.
Even if you could it's still generally too limiting for iterative processes.

The only tools I've seen that do it all for example, do it over point clouds and auto z snap while you draw in the ortho top down view, like the iRacing track videos show.



By building your own smooth z mesh for conforming with, you're kinda copying that approach as best you can.


Dave

 

[LockeNessMotorsports]

No the splines are a guide.

You can snap to them though.

Honestly you sound like you're over thinking things.

You'll *always* lose the smoothness adding loops, that's why you need a solid outline to move the new points to.

The idea is to create the right amount of loops in the loft, sweep, or whatever other initial process you use, for then making the next stage work.


Or how about creating a rough but close top-down sub-d mesh that is just smooth for z info, but has a poor flow for visual use... but good for tweaking camber etc.

Then conform a perfect top down mesh to that z.mesh.

That way you deal with z data on a mesh optimised for that (camber, height etc), and do the visuals with a mesh optimised for that.


You will *NOT* do everything in one uncollapsed mega stack of modifiers in 3DS Max or Blender.
Even if you could it's still generally too limiting for iterative processes.

The only tools I've seen that do it all for example, do it over point clouds and auto z snap while you draw in the ortho top down view, like the iRacing track videos show.



By building your own smooth z mesh for conforming with, you're kinda copying that approach as best you can.


Dave

Aha yeah I overthink everything aha part of anxiety I like things to be perfect. Yeah I know you can snap things to the spline I get what you mean now for conform I couldn't get it to work on the google map it self it hovers above it for some reason. I guess I could create my own but how would you go about getting exact.

 

[Mr Whippy]

The eternal question. How to get it exact.

I could write a book or two haha.

Are you local to the track?

 

[LockeNessMotorsports]

The eternal question. How to get it exact.

I could write a book or two haha.

Are you local to the track?

no where near aha, i'm in the UK

 

[LockeNessMotorsports]

The eternal question. How to get it exact.

I could write a book or two haha.

Are you local to the track?

Pm me if it's easier

 

[Mr Whippy]

Hmmm.

Anything like drone footage online?

Might be able to create a point cloud from it?

No aerial lidar coverage for it?


If it's only a small kart track maybe someone local would 'drone' it for you for beer tokens?


Where is the track? I can have a look to try find useful stuff?

In the end old fashioned 'hard work' will work.
I.e., referencing videos and stills and lining things up.
Yes it won't be perfect, but you'll learn loads doing it, which will be valuable for any track you make in future even if it's laser scanned.


Dave

 

[LockeNessMotorsports]

Oregon Pats acres, there is a bit of aerial but not a load of it.

Here's one but you don't see allot of it