Physics Tuning Related

[relaxation]

A comment was made on Twitch suggesting that the Jack*Bot the TPA streamer was playing on was slow compared to a previous stream of the real machine. After I acquired a copy of the IFPA footage and attempted to reenact the scene, a couple times, I created this side by side video

The descent from the visor bank is roughly 25~29.5% slower than it's real counter part, by a quater of a second. Is tournament &or personal footage of real machines used as reference material?

 

[Shaneus]

That could just be table angle, though. IFPA tables are likely to be set up steeper than a regular home user would, I'd say.

 

[Ben Logan]

I love the spirit of this video. Respectful way to gently point out that TPA is far too easy (Paragon is an exception)! We need more videos like this one. Thanks!

 

[Ben Logan]

Also points to a simple feature request: Let the TPA user adjust the slope! If tuning is an issue, then maybe limit to three settings: semi-level, midrange, and steep.

 

[relaxation]

I'm not certain it's the slope of the playfield, here's some context I put in the videos description

Acknowledgements: IFPA sourced material (720p60) has duplicate frames so it's not 100% exact, my shot wasn't an exact copy in TPA.

First round: disengaged flipper to passing the black line from orange to green below the slings.
Real:20
TPA:24

Second round: from across that black line to contact with the visor bank.
Real:22~23 (hard to tell with the blur)
TPA:20

Third round: from the visor bank to the black line between the orange and blue which is below the slings.
Real:42~43 (duplicate frame right before line)
TPA:54

The rebound from the visor takes noticably longer, the bank targets may unintentionally absorb some momentum [54 / 42.5] 27% speed reduction.

Here's when the balls make contact at roughly the same time against the bank, their disengage/ascent differences canceling out

Then moments later, .266th of a second, the real machine has a solid lead

If I was asked, which had the most contributing factor above.. slope or squishiness? I'd choose squishiness.

It's only a guess, we could say real was 3 steel ball lengths ahead so how much faster was real going and what kind of degree increase would be needed to match real on TPA?

 

[Ben Logan]

Great analysis.

 

[relaxation]

Well my bad math came up with a 1.78 degree increase, to match a real table. iirc jack*bot was set to 6.5 degrees while the higher optional setting was 7 degrees.

Consider the following: If...
Mass is 80g
Acceleration down slope is 43.7 inches/s, when incline is 6.5 degrees
Time frame of 16 frames, or .2667
Distance to make up is 3 ball lengths, 3.188 inches

43.7*.2667, 11.65 is our place on the playfield, we want to get to 14.84 by that time

so 14.84/.2667, we want a down slope acceleration speed of 55.64, using the calculator it would be 8.28 degrees
http://www.calctool.org/CALC/eng/mechanics/slope

So perhaps I shouldn't discount slope too much but I'd still stick with squishiness.
edited, 2 ball lengths was a mistake

 

[dave950lam]

To make this a better comparison, you need to use the actual machine that Farsight used to make the virtual game.

 

[Kolchak357]

I know Jeff asked for difficulty settings ages ago. So don't hold your breath on that one. It would be great though. Ball spin is the biggest change I'd still like to see.

 

[Ben Logan]

Agreed about ball spin as most necessary physics improvement. Lots to like about TPA, but it still plays like pinball flavored pong.

 

[shutyertrap]

While you guys were looking at the ball speed, I was looking at how much higher the ball bounces off the flipper in TPA compared to the real table. But yeah, we don't know how FarSight has their machine set up, and a lot depends on that.

 

[relaxation]

To make this a better comparison, you need to use the actual machine that Farsight used to make the virtual game.

Part of the question I asked, does personal footage get compared? Now it doesn't have to be every table in every situation. We players have been aware of Vaccuum ramps (which have been fixed?), springy spot targets (genie special/eb targets, think I heard they fixed those), suction drop targets (horizontal drop target walls, heard certain tables fixed but not all), acceleration one-way gates (cyclone has it bad).

Besides ball spin that gets brought up which they're unlikely to do, what other anomalies do real players see in farsight tables? besides the nudging which they also have been mum on

 

[relaxation]

I was looking at how much higher the ball bounces off the flipper in TPA compared to the real table.

I tried a little harder to replicate the angle

I think it looks darn close, flippers were actually disengaged/engaged at same time* (cant embed 2 videos) as the last comparison.

As for the bouncing, TPA flippers soak up a lot of energy.


*this is due to a variation in rolldown speeds, muscle memory doesn't help when ball speed varies from a trap. I ran into this before on Frankenstein (ball separation gamble on the right flipper) and F14-Tomcat (specific rebound to inlane)

 

[dave950lam]

How did you tweak the physics/programming? That almost matches the real thing.

 

[relaxation]

Didn't adjust anything on my end, the variable rolldown speed from a trap just happened to be slower than the first video.

spot the subtle difference, the frame before the flippers engage

So there is a degree of randomness coming from a trap, so things like the Fish Tales backhand not working 100% may contribute to this.

Another 'sometimes' backhand is on Eldorado EM, grabbing the 4th from the left of the 5 bank from the right flipper.

.. I have doubts about hitting the same 60th of a second, looking into that later.

 

[Rittless]

Just want to reiterate that I would be cautious against benchmarking against IFPA videos that will prefer steeper tables. At the very least, there is no guarantee their table is 'stock'.
It's an even muddier question for older games where table angle is not even specified in the manual.

However the last set of videos is a great example where table angle is not going to be an appreciable part of the speed; it more reflects the tuning of their simulated targets, which are plastic and metal on the real thing. Of course rubber tuning is a factor too.

I would reckon that Farsight do this kind of testing/comparison themselves internally. Things like vacuum ramps etc were crutches to improve playability in the early days but their physics model is beginning to mature up to the point where they can dispense with them.

Of course ball spin if implemented will really separate the men from the boys...