Hi All:

   My old Citroen GS got a brand new engine and suspensions, thanks to
Koppi's Bullet Physics Playground, but still doesn't steers... :)

   We added bindings for constraints and they seem to be working
decently, within the limitations of the Bullet library itself:

https://www.youtube.com/watch?v=bXTT1uGcpks

   I will commit the final LUA script for this "GS test bed" this 
weekend, as I'm still trying to find a constraints structure that works
convincingly in different situations.

   Regards,

--
Jaime

Post a reply to this message

Le 19/12/2012 13:53, Jaime Vives Piqueres a écrit :
> Hi All:
> 
>   My old Citroen GS got a brand new engine and suspensions, thanks to
> Koppi's Bullet Physics Playground, but still doesn't steers... :)
> 
>   We added bindings for constraints and they seem to be working
> decently, within the limitations of the Bullet library itself:
> 
> https://www.youtube.com/watch?v=bXTT1uGcpks

Yummy yummy yummy!
You're nearly ready for the next 007 movie!

It seems to slow as if someone was actionning the brakes once landed, or
is it the turn it took ?

Only point I would dispute: that colour, I cannot believe that Citroën
ever made a GS of that blue. (White is probably the most common colour
for the GS; Beyond that, you would need the official publication for the
available colours... Black-reflecting could also be common, but
raytracing a black paint is not that great for details)

Post a reply to this message

El 19/12/12 15:45, Le_Forgeron escribió:
> Yummy yummy yummy!

   Thanks!

> It seems to slow as if someone was actionning the brakes once landed,
> or is it the turn it took ?

   Yes, the LUA script is (sort of) braking it at step 150 into the
simulation, during 50 more frames. I said "sort of" because I just
inverted the angular motor power during 50 frames, then disabled it.

   But the suspension system broke after that... and so I shortened a bit
the animation to hide this fact. :( Fortunately, I changed now to
another type of constraint that seems to working better and it's more
resistant.

> Only point I would dispute: that colour, I cannot believe that
> Citroën ever made a GS of that blue.

   You're right: the texturing on my GS include files (which had a more
believable blue tone) was not working for some reason, and I just wrote
something quick (and not very good) to get the animation tests running...

> (White is probably the most common colour for the GS; Beyond that,
> you would need the official publication for the available colours...
> Black-reflecting could also be common, but raytracing a black paint
> is not that great for details)

   Well, to be frank, the model isn't a real GS variant either, so having
a non-official paint job doesn't seems out of place... :) I modeled it
looking at photos from all sorts of GS and GSA models.

--
Jaime

Post a reply to this message

Wow, wonderful work again. Yes this blue isn't really Citroen but the model is
close enough to go for a GS. I only remember a brown metallic one (break
version) a friend lend me ages ago for two weeks...

And I'm still driving a ten years old Xsara Picasso... (I didn't dare to model
it so far)

Best regards,
Michael

Post a reply to this message

https://www.youtube.com/watch?v=CGYPHWOAt94

After some hundreds of variants, I found a combination of constraints
which work decently for most situations. The problem is that it's a bit
unstable, and only works as expected half of the times you run the
sim... I suspect is not BPP, but the Bullet library itself. Anyhow, when
it works, you can play with it as long as you want, and it will never
break, as you can see on the above video. :)

--
Jaime

Post a reply to this message

On 03/02/2013 8:53 AM, Jaime Vives Piqueres wrote:
> https://www.youtube.com/watch?v=CGYPHWOAt94

Impressive.

-- 
Regards
     Stephen

Post a reply to this message

Jaime Vives Piqueres wrote:

> After some hundreds of variants, I found a combination of constraints
> which work decently for most situations.

looks great! The only point where it looked a bit jerky was when driving
up the stairs. With the steering / braking system, do you mean you set
the position / forces of the front wheels and the car follows it
based on friction with the ground?

Post a reply to this message

Le 03/02/2013 09:53, Jaime Vives Piqueres nous fit lire :
> https://www.youtube.com/watch?v=CGYPHWOAt94
> 
> After some hundreds of variants, I found a combination of constraints
> which work decently for most situations. The problem is that it's a bit
> unstable, and only works as expected half of the times you run the
> sim... I suspect is not BPP, but the Bullet library itself. Anyhow, when
> it works, you can play with it as long as you want, and it will never
> break, as you can see on the above video. :)
> 

Great. Awesome.

To be even better IMOH, could you somehow pattern the wall and may be a
bit the black ground. A large checker on the wall, and maybe some lines
on the ground ? Aim: to ease my view of inertia / non-drifting rear
wheels on sequence where only the car, the ground and the wall are. So
far on such sequence, I cannot assert that the car is not gliding.

The first step of stair has an unrealistic front wheel move (jumping
from one rotation to another).

before the ramp, there seems to be some stick/shadow under the front
right wheel, as if something was broken.

Post a reply to this message

Am 03.02.2013 18:33, schrieb Le_Forgeron:

> Great. Awesome.
>
> To be even better IMOH, could you somehow pattern the wall and may be a
> bit the black ground. A large checker on the wall, and maybe some lines
> on the ground ? Aim: to ease my view of inertia / non-drifting rear
> wheels on sequence where only the car, the ground and the wall are. So
> far on such sequence, I cannot assert that the car is not gliding.

Seconded. (I had the impression/expectation however that the car /is/ 
gliding, bu with the current ground it remains inconclusive.)

Post a reply to this message

> looks great! The only point where it looked a bit jerky was when
> driving up the stairs. With the steering / braking system, do you
> mean you set the position / forces of the front wheels and the car
> follows it based on friction with the ground?

   Thanks to Koppi addition of definable keyboard shortcuts, I was able
to assign a key for acceleration, another for braking, and two more for
the steering to left and right:

   + the acceleration key just applies some angular speed to the tires,
which rotate freely around a hinge constraint. Using high friction on
the wheels and the ground, the car starts rolling...

   + the braking key cuts the power and also blocks the free rotation of
the wheels hinge. Again, friction does it's work and the cars stops
reasonably well.

   + the left/right keys turn on the motors of a pair of hinge
constraints attached to the wheels, on the appropriate direction. I
didn't need to implement any sort of differential... still don't know
why, but the car acts as if there is one. For the rest, I totally
disregarded any other things like caster, camber, toe, ackerman, etc...

   Apart from that, there is a simplistic suspension joining the steering
hinges with the chassis: this is where most of the failed attempts were
wasted, trying to find a combination that worked well and didn't break
under the forces involved.

   Oh, ...and the jerky moment at the stairs was me pressing left/right
to find a better attack point, as the power was not enough for a frontal
approach... :)

--
jaime

Post a reply to this message