Physics engine discussion

[vikingerik]

It's all about what is described as "railroading". There's some shot adjustments / helpers going on in the engine, that much is obvious.

The point of this thread is that we now know how those helpers work, thanks to [MENTION=446]invitro[/MENTION]'s discovery in the resource files. We see the numbers defining the regions on each flipper that will apply force in predetermined directions towards the shots.

We've learned that the railroading occurs not just because TPA's spatial or temporal resolution is too coarse. It occurs because the flippers are literally defined to railroad the balls.

 

[Zaphod77]

The railroads are put in as a COUNTER to not enough spatial and temporal resolution. To make sure that the shot that needs to be possible IS possible. Because their recreations are "exact" they can't just tweak the flipper strength and angles a little to make the shots possible like we all had to do in VP.

Without sufficient of both, the shots are too HARD, and need help from the physics engine to be doable at all.

When a shot doesn't go where you expect it to on a real table you can't complain. Real life doesn't lie. but when the same happens in digital pinball, you feel that the simulation robbed you.

PS. anyone who doesn't think zen doesn't the same thing, go play their spiderman table. nearly any reasonably aimed flip from the upper flipper will hit the side ramp or the orbit.

 

[vikingerik]

The railroads are put in as a COUNTER to not enough spatial and temporal resolution.

But that's the wrong counter, treating the symptom not the problem. The real solution is to improve that resolution. We know it can be done because Pro Pinball did twenty years ago. (Temporal resolution may be constrained by technical platform limitations, although I'm still not sure if PP or VP runs any faster than 60 Hz.) Farsight just isn't interested and doesn't have a need to be.

 

[Ben Logan]

I haven't been around TPA as long as a number of you veterans have been -- my post count will testify. But, I can't help thinking that what we're doing here is valuable to Farsight as they develop SPA and Physics 4.0. In the past they may have had the luxury to not be interested in Pro Pinball quality physics, but that may change with SPA. I'm going to risk a minor gamer's heartbreak and remain optimistic that they'll meet our request for an expert mode, at least.

Why? There may be a buck in it for them (and I do believe at least some of the FS team truly loves pinball and wants to honor the wild ball that is the game.) I believe a significant part of Stern's motivation in partnering with FS to release current production line tables like Ghostbusters is to create a "try before you buy" platform for prospective real pinball purchasers. If you live in the middle of Kansas (no offense intended) and don't have access to a hipster barcade, but want to try before you buy, SPA is your means to do so. If it feels like PONG, you're going to be disinclined toward purchasing the real table. For prospective real pinball table purchasers who are using SPA as a research tool, we need Expert Mode at the very least.

Also - while we're discussing consumer needs -- didn't Pro Pinball's Timeshock and other tables sell pretty well back in the day? As such, Pro Pinball is living proof that a high quality physics emulation is not only possible, but profitable.

 

[DA5ID]

I am all for any forward progression toward improving the physics. I would like to throw out a friendly reminder that this is essentially a video game and enjoy TPA, Zen, and Zacc because they are first and foremost FUN. The physics on all three are close enough that my disbelief is temporarily suspended and escape is thoroughly achieved.

The points about if this was an uber-realistic simulator you would be better IRL. ALL pinball tables in our physical world play differently based on the machines age, cleanliness, and wear and tear. I know guys in league and tournament that OWN the table we are playing together and don't dominate because the shots they have mastered at home don't always work on the table that plays differently. They, like everyone else, have to adapt their play to how the table flows. Thats why we come early to league/tournaments and "warm-up" with practice runs - we are testing the table for variances.

TPA has absolutely made be a better player in real life, because like the players that own the machines at home, I am able to get enough flight hours logged so that I know the machines ruleset and have developed a high scoring (game winning) strategy.

You'd be surprised how many games you can win by simply knowing what the skill shot is, or how to start a multi ball quickly.

 

[Ben Logan]

Totally agree that TPA is super platform for ruleset learning. PAPA videos are great for this, too. But, I'm with vikingerik: Pro Pinball makes me better at real pinball, TPA makes me worse.

It's a Pinoccochio style dream: TPA is a wonderful video game. Can it become a real simulation? I think so. And an "Expert Mode" can only attract more revenue. I'm spending the next two days at Pinagogo Pinball Conference in Dixon. If I thought I could drop say, fifteen bucks on Expert Mode tables in SPA / TPA that actually helped me to practice for a tournament, you bet I would!

 

[steven120566]

I am all for any forward progression toward improving the physics. I would like to throw out a friendly reminder that this is essentially a video game and enjoy TPA, Zen, and Zacc because they are first and foremost FUN. The physics on all three are close enough that my disbelief is temporarily suspended and escape is thoroughly achieved.

The points about if this was an uber-realistic simulator you would be better IRL. ALL pinball tables in our physical world play differently based on the machines age, cleanliness, and wear and tear. I know guys in league and tournament that OWN the table we are playing together and don't dominate because the shots they have mastered at home don't always work on the table that plays differently. They, like everyone else, have to adapt their play to how the table flows. Thats why we come early to league/tournaments and "warm-up" with practice runs - we are testing the table for variances.

TPA has absolutely made be a better player in real life, because like the players that own the machines at home, I am able to get enough flight hours logged so that I know the machines ruleset and have developed a high scoring (game winning) strategy.

You'd be surprised how many games you can win by simply knowing what the skill shot is, or how to start a multi ball quickly.

The above post I quoted by DA5ID kind of makes a point to me that resonates.

As fascinating as this thread is... and boy is it interesting, I prefer to believe that they have perfected ball physics to an imperceptable variation from reality. In my case that works because I haven't played a real table in two decades. To me and probably most TPA players, this is what "pinball" is today. And as much as I try to project the game to reality, it seems easy to do for me. And, if I think about it being FAR from real as oppopsed to close, well then the fun of the game diminishes. The Pinaball Arcade universe is the games...digitized as close to the real thing as can be, the ambiance and the leaderboard communities And plus...these tables never physically break down, although it does go through console controllers pretty fast. The beauty of this game is that all things are equal to all players. Everyone on a level playing field making the leaderboards a very valid indicator of pinball aptitude. I wonder sometimes, can I equate the best of the best players of TPA to casino card counters, are they able (consciously or subconsciously) to predict exactly how the ball is going to behave every time? Or is it really table rules knowledge combined with impeccable hand-eye coordination....

 

[vikingerik]

Also - while we're discussing consumer needs -- didn't Pro Pinball's Timeshock and other tables sell pretty well back in the day? As such, Pro Pinball is living proof that a high quality physics emulation is not only possible, but profitable.

Pro Pinball didn't have TPA and VP and Zen to compete against. Pro Pinball stood out by leaps and bounds in its day against all the other sims that were crude cartoony vidya-games in comparison.

Pro Pinball also wasn't competing against the race-to-the-bottom oversupply on Steam and the app stores. Pro Pinball sold for $50/unit which was normal in that day and represented a much higher return on investment than TPA can do at $5 while competing against millions of 99-cent games.

PP's physics was a sufficiently distinguishing selling point then; it isn't now. Notice that PP Ultra has only been focusing on rendering and platform support and hasn't done anything with physics yet. Because that's what matters in today's marketplace.

 

[Kolchak357]

I love the way pro Pinball plays and feels. But we'd still be on the first table pack arguing about what the second pin release would be. So I'm happy with what we have. But I'd love TPA with PP physics if given the option.

 

[Ben Logan]

The above post I quoted by DA5ID kind of makes a point to me that resonates.

I prefer to believe that they have perfected ball physics to an imperceptable variation from reality. In my case that works because I haven't played a real table in two decades. To me and probably most TPA players, this is what "pinball" is today. And as much as I try to project the game to reality, it seems easy to do for me. And, if I think about it being FAR from real as oppopsed to close, well then the fun of the game diminishes..

Great points. This post really illustrates the need to address two markets: Casual gamer, vs pinball enthusiast. I can totally respect that for the casual player, who is probably fantastically skilled at say, first person shooter games, etc., the difference between TPA physics and real world is "imperceptible." But if you're passionate about pinball, or even play Pro Pinball or Zaccaria you just can't help feel that something major is missing from TPA physics -- on the scale of color tv vs. black and white. An Expert Mode would correct this.

 

[Ben Logan]

I love the way pro Pinball plays and feels. But we'd still be on the first table pack arguing about what the second pin release would be. So I'm happy with what we have. But I'd love TPA with PP physics if given the option.

Would you pay a couple of bucks extra on a table you loved to unlock it? I'm curious. Maybe it's just me. And I'm relatively broke by American standards!

 

[Jeff Strong]

I love the way pro Pinball plays and feels. But we'd still be on the first table pack arguing about what the second pin release would be. So I'm happy with what we have. But I'd love TPA with PP physics if given the option.

The Pro Pinball delays are mainly because of the pre-rendered graphics, not the physics.

 

[Jeff Strong]

Pro Pinball didn't have TPA and VP and Zen to compete against. Pro Pinball stood out by leaps and bounds in its day against all the other sims that were crude cartoony vidya-games in comparison.

Pro Pinball also wasn't competing against the race-to-the-bottom oversupply on Steam and the app stores. Pro Pinball sold for $50/unit which was normal in that day and represented a much higher return on investment than TPA can do at $5 while competing against millions of 99-cent games.

PP's physics was a sufficiently distinguishing selling point then; it isn't now. Notice that PP Ultra has only been focusing on rendering and platform support and hasn't done anything with physics yet. Because that's what matters in today's marketplace.

It also helps when you already have the best physics in town.

 

[mpad]

I am a big fan of pro pinball, but I think you can't just put one physics engine into the other game. the approach in game engines is totally different. Static vs dynamic 3D.
But the laws of physics and the math should be the same

 

[Zaphod77]

Once we start trying to increase temporal resolution, we end up running into timer resolution issues. one frame links are hard enough in fighting games, and anyone capable of making them will do very well in real pinball.

Pro pinball is supposedly wonderful and fully emergent, and yet in many cases, i felt that shots were too difficult. this is because while you can run the physics at subframes you are limited by the feedback of the visible frames, so it ruins your ability to judge timing compared to real pinball.

What TPA does is indeed the lesser of two evils i think. tha said there are ways they can get away from railroads without affecting flipper action much.

 

[Kolchak357]

Would you pay a couple of bucks extra on a table you loved to unlock it? I'm curious. Maybe it's just me. And I'm relatively broke by American standards!

A buck or two? Sure

 

[invitro]

Would you pay a couple of bucks extra on a table you loved to unlock it? I'm curious. Maybe it's just me. And I'm relatively broke by American standards!

I'm procrastinating on trying to figure out the PBT*.csv files some more... I looked for awhile last night, and even dug out gnuplot to make some scatterplots, and didn't get anywhere. I'm starving to find out exactly what those TransferPoint rows mean, as I'm not sure if we can go much more forward without that understanding. But in the meantime...

Since you asked , $2 is close to the max extra I'd pay for a table with Pro Pinball physics, if it also had some kind of expert mode that was able to keep the (say) #10 game under (say) three hours. And I'd pay that for about half the current TPA tables, but not all at once. If I weren't also relatively broke, I'd pay more, maybe $5-7... I'm guessing I'd pay $10 total per table for a "perfect" simulation. (I know, I'll be first in line to pay $10 for the new tables in SPA, knowing that they will be imperfect, and not being sure that they are worthy tables, but I've never played those as real machines. If a "lost" 1990's WMS appeared, I'd probably pay $25 or more for its simulation .)

 

[Mike Reitmeyer]

While I'm not going to go into the details of how our physics work, the number of transfer points doesn't mean there are only finite ways the ball can be shot. So 10 transfer points on the Funhouse upper flipper does not mean there are only 10 shots. There is an interpolation done between them, which comes out to theoretically max float possibilities. But in reality we're gated by frame rate.

The railroading I keep seeing mentioned, I think has more to do with 60hz refresh rate than anything. Picture the ball being cradled, and then the flipper drops. Every time you do this the ball starts in the same location, and gravity and slope of the playfield/flipper angle (which are all constants) pull the ball at the same speed toward the drain. Lets say it takes 1 second for the ball to roll past the tip of the flipper. This means there are only 60 possible shots you can ever do from that flipper because the ball will always be on those same 60 points as it moves each of the 60 frames in the second.

 

[invitro]

Thanks for your post, Mike. Well, different people posting here have different agendas. Personally, I think 60 possible shots is enough. The problem I see is that those shots are not assigned to targets correctly, and it's easy to see how that can happen if the assigning is done artificially (as appears to be the case) rather than naturally.

Now, we know that your assigning of shots was done incorrectly, from the previously impossible shots (AFM R ramp, HRC R orbit, et al.) being changed in version 1.41.7. The suspicion I have is that the errors go beyond this... various ramp "vacuums" being assigned more of those 60 shots than they should be, maybe to even a global overassignment of low-angle shots (backhands and flipper-tip shots) and underassignment of center shots.

I personally use "railroad" not to refer to flipper shots, but to kickouts and other ball returns from the upper playfield. Like the TZ Slot Machine return, and the Cirqus Ringmaster return that vikingerik mentioned in the parent of this post. I don't see how the current frame rate limit could possibly be responsible for those railroads.

I probably don't need to say this, but I am thrilled with TPA and am satisfied with its physics in general. I might have an agenda to try to get people on this forum to realize that a >60 frame rate would give less general improvement in physics than fixing the flipper shot assignments, which I suppose is called "tuning." Well, since I'm not sure that my claims are correct, my real agenda is to figure out what is correct, first , based on hard facts like those in the PBT*.csv files.

 

[vikingerik]

The railroading I keep seeing mentioned, I think has more to do with 60hz refresh rate than anything.

We've long known about the 60 Hz limitation. This thread is about realizing that that's not the whole story.

The evidence is that some railroads are more frequent than others. On STTNG, there's a railroad from the left popper that almost-but-not-quite makes the lock hole, which I can make happen like 90% of the time. But there are other railroads that occur much less than that; notably the lock release on TZ ricocheted off the held upper-left flipper into the piano, which happens for me about 25%.

The most straightforward explanation is that TZ gives you 1/60 second to make that shot but STTNG gives you a wider window, because TPA artificially consolidates multiple 1/60 second intervals into the same result.

I always suspected that was happening but could never think of how it would, until invitro discovered those "transfer points". As Zaphod said, single-frame links in fighting games have long existed and nobody can do them consistently. I can make the same railroads in TPA occur more reliably than anyone can hit a single frame timing window in anything.

Finally, how do we explain the temple shot on Ripley's? Before some recent tweaking, this shot would always railroad into the C lane the same way. Now there are a few different possibilities up that ramp, at least one that goes to B and two different paths into C. How is this possible if there was only one 1/60 input into the original railroad?