37 replies to this topic

[Ben Logan]

Good Evening VP Friends,

 

I had a fun discussion about ball physics this morning with one of our awesome table authors. Being a mere player, all I have to offer are observations, unfortunately!

 

One area for improvement we seem to agree on is with regard to the pattern of fast/slow/fast/slow ball movement inherent in pinball. The ball rockets around a real pinball table with ever-varying speed -- a big part of what makes for the "ball is wild" feel, of course. On a pinball machine, the ball seems to...

 

* blast quickly off the flipper

* have all or part of the energy of its movement absorbed by an object upon impact

* significantly alter or altogether reverse course in reaction

* leave the impacted object at a different speed

 

All these collision based energy exchanges result in a change in ball speed. With VPX, the speed of the ball does change as it crashes and careens around the table... but not enough, imo.

 

On a real machine, the ball seems to travel quite dynamically in terms of velocity in a pattern of fast / slow, as encountered objects either re-energize, or absorb energy.

 

On VPX, the "fast" is fast. But, the "slows" aren't always so slow. The perceived result for VPX is less polarity between between fast and slow, and instead a more linear-speed progression around the table.

 

I'm pretty sure it's been discussed before, but I'm hoping we can revisit a discussion about rubbers in VPX. I'm guessing they're the biggest culprit here.

 

VPX Rubbers aren't squishy enough!

 

If you've ever played Barnstorm Games' Pro Pinball ("Timeshock" is my favorite) you can kinda feel the difference in the way their physics engine handles ball to rubber contact. The ball seems to "sink into" the rubbers more. In this one particular regard, I think they may have us beat.

 

Is this ball-sinking-into-the-rubbers behavior emulated in VPX? If it is, can we dial up that setting quite a bit?

 

Here's one more illustration. We've all seen "squash and stretch" on 1930s cartoons. If you've played "Cuphead," you know what I'm talking about!

 

VPX feels more like this when the ball comes into contact with a rubber:

 

 

 

When a ball contacts a rubber at high velocity on a real machine, the feeling is more like this (unfortunately this animation doesn't loop -- You may have to click on it):

 

 

This is a total exaggeration in effort to make the point, of course! Obviously it's the rubber that squishes and stretches, not the ball, but you get the idea.

 

Here it is in action on a real machine: Swords of Fury. The effect is best seen as the ball hits the top of the right sling's rubber. It's a subtle thing, but it's there if you look for it. The ball "squishes" into the rubber, freezes for a split second, and then sproings back out onto the playfield at a reduced rate.

 

 

https://www.youtube....eature=youtu.be

 

 

The video clip above is just one little example. But, think about that ball squishing into rubbers hundreds of times over the course of a five minute game -- that majorly impacts the feel of the fast/slow, push/pull feeling of real pinball.

 

I know I'm not pointing out anything that we don't already know. I've just been thinking, man -- so many things have been fixed so effectively:

 

- Latency drastically reduced

- Ball spin -- awesome!

- Ball hop -- badass!

- Flipper trajectory majorly improved

- Flipper tricks!

- Saucer swirlies

- Scoops that the ball actually falls into

 

The list of improvements goes on and on! So, why does my real-pinhead brother still say, "Um... sorry, this looks great, but just doesn't behave quite like real pinball"? I'm thinking we're down to the nitty gritty fine-details. My money is on improved rubbers for the win!

Edited by Ben Logan, 23 April 2020 - 03:41 AM.

[Schlabber34]

Good points Ben!

 

I've also been after one special problem that buged me ever since i started with VP and i think it also belongs to the rubber physics! When playing, people might think it's more a PF friction thing but i believe it's not! The problem is the following:

 

We all know that if we crank up the table friction to something that makes the ball roll / spin forward when rolling, just like on a real table, we get those ugly curve ball / bowling hook effect. For proofing my theory i need a real pinball machine and some time for high speed video capturing. The few places where i can play a real pin wouldn't allow for that. What i, or better we need is a high speed video of a rolling ball bouncing off a rubber post. Even better when that ball has some dots on it so you can see the spin before, while and after the impact more clearly. I can capture the same high speed impact video on my 144hz monitor at home.

My theory is, that if you compare both spin behaviours on impact and especially after the rebound, you might see (I HAVE NO PROOVE FOR THAT!!!) that the rubber on the real table is stopping the rotation of the ball almost dead on impact and sends the ball back onto the table with almost zero spin. I fear that in VPX the ball rotation isn't stopped on impact and the rubber sends the ball back onto the table with some heavy backspin that leads to the curveball behaviour. Right now we work against this by making the ball skid on ice which leads to even more physics problems! 2nd "proof" is, the ball in VPX is not jumping when hitting a rubber. We have to script that. On a real table the ball jumps because the rotation and the rubber friction in combination is making the ball "crawl" up the rubber. So it seems that there is something missing in the physics engine that calculates the rubber/impact thing like higher impact = more friction! Maybe???

 

Like i said, IT'S JUST A THEORY!!! I have no proove for that!!

[wrd1972]

Great discussion. Glad it came up.

 

The MAX setting for friction on a rubber seems to be 1, and its my gut feeling that the friction associated with 1 just is not near enough. It just seems to me that the ball bounces off of rubbers like the rubbers were made of plastic, and it looks very binary most of the time. On real PB, the ball can hit a rubber and rebound away in a binary looking way, or it can just nearly stop, or fire off in a very unexpected direction. You just dont see these random like behaviors in VP, and scatter angle does not seem to be the cure either. Just seems to me that the rubber friction should have a much higher influence over the balls behavior than what it currently does.

 

I also believe that flipper friction, or lack there of, is partially responsible for the wider flipper trajectories that we tend to see, that we try to correct with hacks. On real PB, the flipper rubber tends to grab the ball as the flipper moves up. Its almost like the ball is sunk into the rubber, or planted to the rubber if you will. On VP, I think the ball is not grabbed by the flipper, but rather it still rolls a bit across the flipper face as the flipper moves up, thus the wider trajectory.

 

I have also played with elasticity fall-off setting for the rubbers, and still cant wrap my head around what that really does. I have tried different fall-off setting on a test table, but I really dont see any different behaviors between lesser fall-off, or more fall-off.

 

I think it would be VERY interesting to see what would happen if the friction setting for objects could be dramatically increased within the engine. I have also thought about why we dont have a friction setting for the ball. We have a friction setting for everything else, but not the ball. Maybe that is a missing factor here.

[toxie]

A rubber is in general a very wild beast for physics engines. Cause in order for it to really work well/exact it would have to deform and actually "remember" that it is deformed. Otherwise one would have to employ some kind of spring hack or something to emulate this behavior (like one can do for cloth).

[wrd1972]

Toxie,

Thanks for chiming in.

 

Is it possible to give us a larger "friction" effect, than what is currently provided with a setting of 1?

[Schlabber34]

@toxie

When a ball "bites" into a rubber on impact it deforms it and this leads to more of the balls surface being in touch with the rubber and in addition to that the weight of the ball puts alot of preasure on the rubber for a very short time. I think the rotation is stopped at that moment within milliseconds. Like a parabolic curve that calculates the friction on impact. Is something like this possible for VPX?

Stopping the curveball would enable us to use a higher friction for the PF. I think that this would affect the flipper behaviour big time. Right now you don't really know which direction the ball spins when it skids onto the flipper. A propper rolling ball that is connected to the PF surface would always spin in the same direction and the rpm's would always be connected to the speed of the ball which makes a shot much more predictable!

Gesendet von meinem SM-G955F mit Tapatalk

[batman444mab]

@Ben logan

Isn't it what you're looking for ?

https://youtu.be/vVq8wu9DgRk

I also am interested by ways to improve VPX. Or at least noticing it. VPX is amazing but when you're out playing an actual pinball machine, you're noticing that we need some physics features as mentionned earlier.

 

That's Adams family in slomo with dots on the ball as you request.

[Ben Logan]

@Ben logan

Isn't it what you're looking for ?

https://youtu.be/vVq8wu9DgRk

I also am interested by ways to improve VPX. Or at least noticing it. VPX is amazing but when you're out playing an actual pinball machine, you're noticing that we need some physics features as mentionned earlier.

 

That's Adams family in slomo with dots on the ball as you request.

 

Nice find! It was actually Schreibi that requested that particular footage. I'm sure he'll be happy you located it! I've actually seen randr's cool marked-ball videos, from back in the day when devs were trying to capture ballspin behavior -- the addition of which was an absolute game changer.

 

 

A rubber is in general a very wild beast for physics engines. Cause in order for it to really work well/exact it would have to deform and actually "remember" that it is deformed. Otherwise one would have to employ some kind of spring hack or something to emulate this behavior (like one can do for cloth).

Yes -- Thanks for chiming in toxie! If it's just too much a beast, I totally understand.

 

I think you've played BarnStorm's Timeshock. What are your thoughts on the way their engine handles rubber physics? Would you agree that they've captured something special in this regard? If so, any idea how they might have done it?

 

https://www.youtube....h?v=BaDcltvaD-k

 

I think I understand what you're saying insofar as rubber "deformation memory" is concerned. I'm guessing rubber deformation would account for what Wrd is describing with regard to unpredictable scatter angle from ball to rubber contact on real table. It makes sense: If a rubber is deformed (indented) in a particular spot, that would certainly send the ball in a different / unanticipated direction in comparison with what a pristine, brand new, undeformed rubber would deliver in terms of rebound trajectory. 

 

What about the "spring hack" you talked about? That sounds great! It's just this springlike "compress / sproing back" behavior that seems to be missing from virtual pinball rubbers. Real rubbers seem more like miniature trampolines. Personally, I think we could live without the virtual deformation memory algorithm. Spring hack would be a super welcome addition, especially if it were optional and didn't affect legacy tables unless switched on.

 

All that said, I can totally see how these questions could be beyond the scope of what you have time or inclination to explore. You and fuzzel do so much for the community already! Maybe rubber collision physics aren't worth digging into...

Edited by Ben Logan, 23 April 2020 - 10:41 PM.

[rothbauerw]

I'm guessing, because I don't really know, that rubbers are simulated as rigid bodies with some approximations in the calculations to account for energy loss on high velocity impacts (elasticity falloff).  But why couldn't these be modeled as springs?  You'd still have to account for losses, but I don't think that would be difficult.

 

One of the things I enjoy most about VP projects is using real world physics equations to simulate various behaviors.  For example, some of the funnest parts of the TOTAN project were figuring how to animate the lamp and the genie. 

 

I used elastic collision equations on the genie to determine how much it would move based on the velocity and mass of the ball.  If you pay close attention, the ball will continue to move forward into the genie once it starts to swing.  With standard vp physics, the ball always changes direction on an impact.  I also used bouncing equations to accurately simulate the decaying movement of the genie after it's struck by the ball.

 

For the lamp, I added spring equations to simulate the two switches under the playfield that the cause a kinda herky-jerky rotation or a spring back as rotation slows and the switches are hit.

 

I don't know the inner working the VP physics engine, but I have to think this is doable.  I'd be willing to help tackle this when the devs have time.  While I can make my way around a script, the level of coding in VP is more than I can handle and it's not something I'm currently capable of doing alone.

[Ben Logan]

Lamp and genie animations are wonderful, Roth. I had no idea how much work and creative brilliance went into that animation process. I’m totally curious to look at that TOTAN script now (I think I can understand just enough to appreciate your work). 
 

Speaking of work, Bord has a series of YouTube videos detailing his current table creation process. What you guys do for this hobby is totally amazing. Super inspiring, especially during these challenging times. 
 

Really want to emphasize that I completely understand that simply starting a discussion or adding a potential item to a coding wishlist isn’t anything like doing the actual work that toxie, fuzzel and development team members do. Same goes for our amazing authors. I could always hang with a response like “Nice idea. Why don’t you learn to code, Ben!” I recognize it probably takes a fair amount of patience to listen to development “ideas.” You guys are the best. VPX is already incredible. 

[rothbauerw]

Definitely keep the ideas coming Ben.  I remember when you first starting the flipper discussion.  These seeds grow.

[Rawd]

 ......Being a mere player, all I have to offer are observations, unfortunately!

 

 

It's never too late to open up the editor and start learning script.  I'd like to see a Ben Logan original. 

Edited by Rawd, 24 April 2020 - 01:22 AM.

[Schlabber34]

 

@Ben logan

Isn't it what you're looking for ?

https://youtu.be/vVq8wu9DgRk

I also am interested by ways to improve VPX. Or at least noticing it. VPX is amazing but when you're out playing an actual pinball machine, you're noticing that we need some physics features as mentionned earlier.

 

That's Adams family in slomo with dots on the ball as you request.

 

Nice find! It was actually Schreibi that requested that particular footage. I'm sure he'll be happy you located it! I've actually seen randr's cool marked-ball videos, from back in the day when devs were trying to capture ballspin behavior -- the addition of which was an absolute game changer.

 

 

A rubber is in general a very wild beast for physics engines. Cause in order for it to really work well/exact it would have to deform and actually "remember" that it is deformed. Otherwise one would have to employ some kind of spring hack or something to emulate this behavior (like one can do for cloth).

Yes -- Thanks for chiming in toxie! If it's just too much a beast, I totally understand.

 

I think you've played BarnStorm's Timeshock. What are your thoughts on the way their engine handles rubber physics? Would you agree that they've captured something special in this regard? If so, any idea how they might have done it?

 

https://www.youtube....h?v=BaDcltvaD-k

 

I think I understand what you're saying insofar as rubber "deformation memory" is concerned. I'm guessing rubber deformation would account for what Wrd is describing with regard to unpredictable scatter angle from ball to rubber contact on real table. It makes sense: If a rubber is deformed (indented) in a particular spot, that would certainly send the ball in a different / unanticipated direction in comparison with what a pristine, brand new, undeformed rubber would deliver in terms of rebound trajectory. 

 

What about the "spring hack" you talked about? That sounds great! It's just this springlike "compress / sproing back" behavior that seems to be missing from virtual pinball rubbers. Real rubbers seem more like miniature trampolines. Personally, I think we could live without the virtual deformation memory algorithm. Spring hack would be a super welcome addition, especially if it were optional and didn't affect legacy tables unless switched on.

 

All that said, I can totally see how these questions could be beyond the scope of what you have time or inclination to explore. You and fuzzel do so much for the community already! Maybe rubber collision physics aren't worth digging into...

 

 

 

 

 

Thanks, that was exactly what i was looking for! The high speed cam was not quick enough for the hard shots but you could definetly see a tendency when the ball bounces off a rubber, plunger or a flipper. The ball seems to stop spinning on impact and the distance the ball is skiding after the rebound is very short with no real curveball behaviour. The skiding distance is just too short for a curveball! I would guess it's a bit more when the ball hits the rubber at higher speeds.

 

One thing i noticed in this video is that the ball seems to be more willing to rotate then the VPX one. Just a liitle bit of friction applied and the ball changes the rotation angles like it weighs only 5 gramms. Tapping the plunger made the ball rotate imediatly! To me the VPX ball always feels a bit lazy when it comes to speeding up the rotation. Almost like the mass / rotation thing is calculated "wrong".

 

Let's see if i can make some high speed cam shots on my 144hz monitor at the weekend!

[batman444mab]

@Schreibi34

I read from the new threatre of magic thread that physics can't be tweaked as we want since vpx physics engine as to be corrected through vbscripts. i know from my past knownledges in coding ( amiga ) that calculations has to be processed really fast. It's all about trade off. I prefered the version with SOSRAMPCOIL at 8.5 but users have prefered 2.5  due to CPU limits. 

From that , would you think that limitations that occured in vpx could be overcome with the new freezy project to use DOTS/Havok physics in unity ?

We can also look at FX3 physics too even if the last volume has lazy flippers but a better ball behavior like ballmass etc...

Can you post your cam footage for us ? Will you use a dotted ball ?

[wrd1972]

What if we had a "scatter elasticity" adjustment for rubbers? It would work similar to the existing scatter angle. That could possibly fake in the random rebounds off a rubber?

[Ben Logan]

Effects of both elasticity falloff and scatter angle changes made to rubber in the editor are tough for me to perceive in action. Not exactly sure what they're doing.

 

I'd assumed that elasticity falloff works as a form of "value compression" -- meaning low speed impacts will impart maximum bounce, while higher speed impacts will "compress" the amount of bounce. I'm thinking this is how we get little "micro bounces" at low velocity impact, without having the ball rocket off the darn table when it hits a rubber at high speed. Is this indeed what elasticity falloff does? Makes the elasticity "fall off" at high speed impacts?

 

If so, rubbers are already calculating the speed at which the ball strikes them. I wonder if we could consider adding a new parameter -- "rebound time."

 

What I think I'm seeing on real tables as opposed to VP is that the harder then ball strikes a rubber, the more the rubber "absorbs" the energy of the ball (greater trampoline compression). This takes time.

 

If we added "rebound time" as a function of speed of impact, that might simulate rubber compression. Not visually, of course. The visual effect is only secondary. It's the "slow the ball down" by stealing energy as a function of speed of impact that would matter toward capturing the character of a ball careening around table insofar as dynamics of speed change are concerned.

 

Slow hit to rubber, instant rebound. Hard hit to rubber, delay of 2 milliseconds before rebound. Medium hit to rubber, delay of 1 ms. I'm just making these values up to illustrate.

 

Would adding a parameter like this to flipper rubbers screw up trajectory? If so, maybe it should only be applied elsewhere (assuming anyone thinks this is decent idea!).

 

Just musing out loud here, as usual.

[batch]

Perhaps I'm wrong, but not everyone uses rubbers, many use walls instead with rubbers not collidable

 

And slingshots are walls too

 

But maybe I talk nonsense !

[Ben Logan]

Thanks for the info, Batch. Didn't know that! ;P

[batch]

Perhaps you're right, and we should try to use rubbers and test what you're talking about

 

And for slingshots, I remember Kiwi used flippers on one of his last tables release

Edited by batch, 24 April 2020 - 09:37 PM.

[wrd1972]

Perhaps I'm wrong, but not everyone uses rubbers, many use walls instead with rubbers not collidable

 

And slingshots are walls too

 

But maybe I talk nonsense !

Yep I use round prims and walls for rubbers, and I am pretty sure many others including JP do this as well. I do this so I can have different physics setting for the band where it wraps around a post, versus where it does not. Also setup in the editor is easier for me with these objects instead of rubbers. There are some other advantages for me as well, but they are not real important for this topic. Physics of those objects are controlled by a material.

 

I would ASSUME that the ball would react the same off of a wall, or a prim - just as it would off of a rubber object. If it does not, please say so.