RELEASED-wip Kyalami 2016

[Prototype]

This NurbsPath tool is kinda fun! :-D

Messing around trying to figure out the toolset here.

I can see that you could probably get everything you need out of it using this..
Will just take finessing to understand it completely and where the best positions for the points would need to be ...

Is this how its generally done in Blender?
I *think* this is what @Johnr777 said I should use?

 

[Kwagga705]

AI testing feeling ok.

Nice to see one of my favourite tracks being remade, how far is it?
Btw, can't wait to put in some laps with the rss f3
Love from Pretoria <3
EDIT: Didn't know it wasn't released yet, but great project

 

[Kwagga705]

This NurbsPath tool is kinda fun! :-D

Messing around trying to figure out the toolset here.

I can see that you could probably get everything you need out of it using this..
Will just take finessing to understand it completely and where the best positions for the points would need to be ...

Is this how its generally done in Blender?
I *think* this is what @Johnr777 said I should use?

When are you doing a Zwartkops remake xD

 

[luchian]

This NurbsPath tool is kinda fun! :-D

Messing around trying to figure out the toolset here.

I can see that you could probably get everything you need out of it using this..
Will just take finessing to understand it completely and where the best positions for the points would need to be ...

Is this how its generally done in Blender?
I *think* this is what @Johnr777 said I should use?

When there is no reference data to conform to (e.g. LIDAR & the like), it's better to use a bezier curve (mid-line of the track) and then align a plane to it to build the asphalt. (see THIS for some very good advice from Lilski)

If/when you have some sort of reference data, then yes, John's method is better, because you can conform to both left and right track limits.

 

[Prototype]

Thanks @luchian --OK, cool. Got it.

Ive been tracing the edges of the track on either side of Google mesh but wasnt sure how you would create a mesh between the splines ....

So then tried this and it just felt super intuitive and easy to do, but a little hit and miss, so wasnt sure.

Any help /pointers with how to do what @Johnr777 suggested?

---------------------------------------------------------------------

@Kwagga705 its going to be a while still

 

[Prototype]

QUESTION:

BLENDER:
Is this correct? In order to have a double sided mesh, you have to duplicate the face and flip the normal ??

So that "double sided" checkbox pretty much does nothing. ^^

Duplicated object with flipped normals ^

This works .... but hope theres a better way to do this??

 

[luchian]

Any help /pointers with how to do what @Johnr777 suggested?

I'm afraid you're not going to like it . I do not know of a "best way" because I am a fan of the single curve method. For when I've done some tests, I just converted the 2 curves to mesh and then tried different ways to fill that. Something along the lines of this:

..but was not happy with the "smoothness" and ended up doing something different: using an excel to calculate mid value and angle, from the left and right measurements. See HERE. HOWEVER, that might be tedious in your case (you'd need to take measurements every 10-20 m or so, on both sides, along entire track) and probably loose some of the detail..

Is this correct? In order to have a double sided mesh, you have to duplicate the face and flip the normal ??

YES, this is the correct way.


PS: there is a nice plugin, but it's not free.

PPS: I am by no means a Blender expert; just an user

 

[Prototype]

Thanks @luchian ! I appreciate the help. Thanks for indulging me in my blender noobery.
Ill check out the vid now !

 

[luchian]

Thanks for indulging me in my blender noobery.

oooooo, I am not that far ahead !

 

[Prototype]

Oh wow ! That Curves to mesh tool looks like a great solution! Just picked it up! :-D
Thanks for the heads up on that one.

 

[LilSKi]

When you have solid data where you can trace road edges. I usually make a curve for each edge.

I then convert that to a mesh and fill it in with alt-f in sections.

Once filled you get the road but it is flat between each edge it helps to return to quads as much as possible with alt-j

I then select the edges and invert the selection so it is only the inside edges and subdivide. This will give you a "center line"

You can then take that center line and turn it into a curve or make a new curve that is shrunk to the center line. Then you can make a new quad mesh on that curve and you can also make the quad road slightly wider and shrink it to the flat road you made and it will make it follow the width and camber of the original traced edges.

 

[Prototype]

Thanks man! @LilSKi Appreciate you taking the time to explain that to me.

 

[Prototype]

Inner and Outer curves done.

Needs another polish pass to double check everything and potentially tighten up, and then on to trying to make the mesh!

 

[Kwagga705]

Inner and Outer curves done.

Needs another polish pass to double check everything and potentially tighten up, and then on to trying to make the mesh!

So pumped to play kyalami, especially with the IGTC coming in November

 

[Prototype]

9 hour race !

 

[Kwagga705]

9 hour race !

That's gonna be awesome, heard Mika Hakkinen is gonna be there

 

[Prototype]

QUESTIONS
Track mesh

QUESTION 1:

• Is there a best practise methodology for mesh structure?
  (Grid fill)
  
• In the section outlined in pink, I cleaned up the verts manually (using GG slide and dissolve verts) to make a cleaner mesh.
  
• In the section outlined in blue I just left it as is.
  
  Top Un-subdivided, Bottom: One subdivision

My concerns regarding "cleaning up the mesh like this":

• Potentially damaging the accuracy of the edges by removing verts?
• Potentially Creating too large a gap between edges?
• Lots of unnecessary work if its not an issue.

My concerns if I don't clean up the mesh:

• Mesh management will be harder with a messy mesh?
• Unnecessarily higher geo count on mesh.
  
• UV mapping might be problematic?
• (Uv mapping tools seem to be a bit limited in Blender?)

-----------------------------------------------------------------------------------------


QUESTION 2:

• Seemingly too high resolution on TURNS comparative to STRAIGHTS?
• When I subdivide the mesh it really gets dense?
• Is what I'm getting here reasonable ?

-----------------------------------------------------------------------------------------


QUESTION 3:
​

• Numerical transforms.
• Hitting G Z to move the track mesh up by 0.1, or 0.05 etc ... How do we do this in Blender?
• I cant seem to find much on accurate, numerical moddeling?

Thanks in advance!

 

[LilSKi]

I usually prefer it to be all quads. What I would do now is make a curve that is shrink wrapped to the center line. Then make a new quad mesh to follow that curve. Roughly adjust the curve radius to always be slightly wider than the base road you have there. Then add the shrink wrap mod again to the new mesh and apply it to the existing road. It should then follow the existing mesh and be all quads.

The road can be non quad and it will work it is just more difficult to UV map. If it is non quads you will need a quad mesh at some point to UV and then use the data transfer mod to transfer UV coordinates.

 

[Prototype]

OK thanks. Ill research the shrink wrap stuff .. Appreciate the quick response !

 

[luchian]

@Q2: what I did use:
- a 2m segment for straight and slightly curved portions
- subdivided once or twice, in the turns

This is for the visual only (and as a rule of thumb, just use whatever looks smooth enough ingame, avoiding to see those "segments", while trying to keep it as low as possible. E.g. if a turn looks good with 2m subdivisions, no need to go under it).

Physical is another story. And mine looks like this (might actually be too dense, but I like the feel):

All info on how to do this, is in lilski's posts, HERE.