Poll: how many cars would you trade for a better game/engine sounds?

[danielc]

Load plays a big part too, doesn't it? At least for real cars throttle and load are linked. It is relative to intake volume(cm3 not db)

 

[Griffith500]

Load plays a big part too, doesn't it? At least for real cars throttle and load are linked. It is relative to intake volume(cm3 not db)

Throttle position and rpm (plus boost via lag etc.) determine the output torque (and hence power) from the engine, and the "load" on the engine through the drivetrain, coming from the road via the tyres, determines the rate of rise (or drop) in engine speed - but that's all determined in the physics engine, and is reflected in the rpm value.

Applying more load to balance the extra power from opening the throttle, as they often do when testing on engine dynos, results in an implied correlation, but the change in sound is entirely due to the throttle opening.

 

[danielc]

Not true, Engine load by formula is:
Engine _ Load = Current _ AirFlow divided by
(Max _ Airflow(Rpm)multiplied by ⋅ (Baro pressure divided by 29.92 ) multiplied by sq root of 298 divided by (amb temp +273))
(The formula didn't translated well into the post...)
What you said is a relation between transmission losses/traction/pw/tq available at a given rpm.
When open 100% throttle engine load =1 or 100% at any rpm, altitude, temp, etc. Irl some tuned/boosted/high powered cars Go over 100%. This is to calculate fuel enrichment and whatnot.
It has to do with volumetric efficiency and this value could be used to calculate the ammount of air entering the engine and give a precise direction to the intake sound to be acurate.
(You've entered one of my alleys of knowledge now... Muahahah )

 

[Griffith500]

Where does "Current_AirFlow" come from? It must surely be dependent on the throttle position and rpm. Everything else is either constant or a slow-changing, non-interactive environmental variable. Also, Max_Airflow(rpm) seems like a hack, probably referencing some "full throttle" and standard conditions (29.92"Hg, 298K) throughput?
The pressure and temperature terms are negligible in effect relative to the air flow term, which will be primarily affected by throttle position for a given rpm (scaling between 0 and 1).

How is the dimensionless number named "Engine_Load" supposed to be used in the sound? To me it looks more like it should be a physics input. Multiply it by the torque curve to feed into the physics, which will return a perturbation in rpm, which is fed back into the sounds (and the next physics frame).


Volumetric efficiency is a measure of dynamic supercharging, such that all boosted cars go over 100%, by definition (of "boost"). Many NA cars do the same by exploiting pressure wave dynamics in the intake tract, timing the arrival of "positive" waves to "over-fill" the cylinder (relative to ambient conditions).

The sound generated by the intake is a function of the pressure wave dynamics, just as the volumetric efficiency is. The throttle plate serves to reflect pressure waves in different ways depending on its position - this is why the intake is "silent" (from the outside, not in the intake tract itself) with the throttle shut.

Changing the ambient pressure and temperature alters the speed of sound, and that alters the pressure wave dynamics - a physical model would just react to those changes automatically, but samples produced at one set of conditions will have to be tweaked, primarily using spectral techniques (many numbers) because of the nature of intake tuning, not simple volume scaling (one number).


Finally, the intake would still make a noise with zero (net) air flow.

 

[danielc]

This can be used as a guide to any conditions, being easier to make, be it part/full/off throttle.
There is no 0 air flow, if the engine is on, there is air flowing. There is also exhaust scavenging effect, vacuum and negative value loads, and using this formula you can correct the sound to sea level/high altitude conditions,on/part/off throttle just adding more or less as the condition requires. If a sound is captured at one particular condition, you can use over/under 100% to adapt to altitude/temp diferences. And would sound true to life. Its one of the variables, not the noise correction factor itself.
Vtec would probably sound like vtec should because the second cam would be accounted for.

 

[sumbrownkid]

This is more of a getting better/more engineers to handle this than getting rid of cars.

 

[HuskyGT]

Dupes and VGT cars.

 

[Griffith500]

This can be used as a guide to any conditions, being easier to make, be it part/full/off throttle.
There is no 0 air flow, if the engine is on, there is air flowing. There is also exhaust scavenging effect, vacuum and negative value loads, and using this formula you can correct the sound to sea level/high altitude conditions,on/part/off throttle just adding more or less as the condition requires. If a sound is captured at one particular condition, you can use over/under 100% to adapt to altitude/temp diferences. And would sound true to life. Its one of the variables, not the noise correction factor itself.
Vtec would probably sound like vtec should because the second cam would be accounted for.

It's basically a manipulation of an equation of state. You haven't answered my questions, though. How does it "correct" the sounds? What is the "load" value actually used for? Where does "Current_AirFlow" come from?

Am I right in thinking this is the "possible better metric for engine sound volume" that the physics guys came up with to better approximate engine performance effects of different atmospheric conditions for Project CARS? If so, it's a physics thing, that is being jury rigged to add some variation to the sound. It is also just one developer's approximation, not an actual fact of nature. That's especially since the important part, "Current_AirFlow", the bit that contributes the majority of the expression in that equation, is not defined in any way at all, and is unlikely to be anything but further approximations, as is only understandable for real-time calculation.

Regardless, it's only useful for sound if the samples are stored normalised relative to each other to have the same peak volume. Even in that case, it's not perfect because sound is fundamentally a pressure fluctuation phenomenon, not mass flow - a scaling based on the latter is not physically correct. What should be happening is normalising to an actual reference volume level obtained at the location of recording (analogous to "colour checker cards" for photographs of cars / tracks nowadays). Physically modeled sounds can give the correct relative volume scaling automatically, e.g. Sonory.


Using that formula for volume scaling will not automatically make VTEC sound right, because the difference is not just one of volume. Correct recording of the car at all engine speeds on both cam profiles will allow you to switch between the "VTEC sounds" at will - this is what I do with my synth, only it's generative, rather than from recordings. The "sound of VTEC" is inherent to the sound the car is actually making, unsurprisingly, and scaling relative to a consistent recorded reference rather than a pseudo-physical approximation will always be superior.

Engines absolutely can net-flow zero air over a complete cycle; reciprocating pumps with the input / output closed do just that, but they still make a noise.

 

[danielc]

So what if its jury rigged? Otherwise there would be no excuse for sim games to be imperfect by now...
It is a manipulation, and would be used for mixing/tying together intake and exhaust sounds to simulate what happens irl.
The load value is the manipulative factor to simulate conditions without having to do a sample for each tp/condition and would still be effective, not to mention it could be used to make tuned cars still sound true to the stock engine. Current airflow is what would give engines some character, because say a viper v10 sucks way more air than say a midget... And each car/intake mani/throtle body has its own characteristics and limitations in terms of cfm. Bigger turbo/boost would be very audible if it was the case.
Same peak volume? Why? A car is louder at peak rpm than at cruising/low rpm, isnt it? Also they sound diferent when you are no throttle and gun it vs an accel test, right?
Not necessarily the game physics treats air how it should.
Its not about the correct real life physics, its about simulating
And vtec was a bad example, i give u that....
Edit not just because it can that it means it does....
Edit 2. If you re-watch the fuzzy logic video, a negative load would simulate scavenging effect and the 5 sounds (muscle, f1,boxer, f40 and the original) are actually 1 set of samples(right?)

 

[Griffith500]

The load factor scales the volume, that's it - and I'm pretty sure it is SMS now. Volume changes alone do not achieve what you're describing.

A Viper would still sound different from a Midget even if they were "sucking" the same quantity of air, and that would be true also if they were heard at the same effective volume, no matter the air flow.

The samples are normalised because they're recorded. It's a pretty standard thing to do, otherwise the volumes could be all over the place as you blend through them all. You still have control of the volume in-game, it's just to "match" the samples to each other - ideally the samples should be normalised to a reference, as I said.

So if the engine is 2 dB louder at 3000 rpm than it is at 2000 rpm, it should be reflected in the samples - it's actually probably easier to still normalise the samples and just scale up the gain with rpm accordingly; but it's RMS pressure amplitude that determines volume, not "air flow"!

There is not a direct link between the two, because of the pressure wave dynamics that the engine is tuned around, and the same principles are exploited in any silencing / sound shaping devices. That equation doesn't actually scale the volume with rpm, though, due to the "air flow / max air flow for this engine speed" term (because it's meant to scale a torque value, not power).

 

[danielc]

What kind of reference? In gt case for example.
What i am saying is a real engine makes diferent sounds even if at the same rpm, because of the air flow and state(acel/decel/steady) .
for a particular airflow(maf/map sensors monitored) there is an ammount of fuel, not to mention timing and many other factors, and the explosions have diferent sounds. Load accounts for that and makes pertinent changes(more/less fuel) and its not current airflow/max airflow for that rpm, is rpm where max flow is achieved, wich is max torque point.
Many cars have flaps in the manifold and/or 2 paths to maximize torque via a longer route/ or smaller hole(bottlenecking for increased air speed) or maximize top end power by switching to a faster/less restrictive path.

 

[Chrunch Houston]

dumb

 

[GT Pro]

Polyphony is killing the franchise with a quantity over quality attitude. There are far too many poorly rendered cars, and the time spent on standard cars #500-1000 could be better spent on revamping the sounds. This is why I loved GT3; despite a low car-count, it felt very polished despite being PD's first PS2 debut (indeed, one of the first PS2 titles). GT6 is the first Gran Turismo instalment I haven't bothered to buy.

 

[MOV Games]

All the standards which shouldn't be there in the first place, and all non existent LM and Touring cars.

GT7 needs to be a complete refresh, far improved course graphics, all new accurate sounds, a MUCH better GT mode and way more Online options. Otherwise PD will become even more of a laughing stock, hell I've not even bought 6 due to it being way too disappointing.

 

[phil_75]

GT6 has been dead to me for weeks now, who'd have thought after the let down of GT5 that 6 would be worse than that?

It makes GT5 look like perfection now, although it did come on leaps & bounds and turned into a fantastic game but if PD think i'm waiting 2 years to sort this one out, think again!

Hopefully Project Cars lives up to expectations and gives PD the much needed kick up the arse they need, but i doubt it!

I never ever would have thought i could be so dissapointed with GranTurismo, my favourite game, the sole reason i owned a PS1,2 & 3 and have no faith in it recovering to the franchise it once was.

The sounds for me are the biggest problem and let down.
I just cannot ignore them any more, all i hear is that monotonous drone and it takes away all the good work they do on the graphics and ultimately the immersion.

GT6 comes on when a new seasonal is out, do that for as long as my ears can bear the particular car, never more than 2 hours, then its back in the case till next seasonal.

The only thing that would get me back is if they sorted the sounds.
The RedBull JR is fantastic, so they can and could have done it but we get moon missions, mapped moon, rover & stars for 3 tiny challenges that take about 15 minutes and thats it.

Well done PD, well done.

 

[remote_racer]

Surely trade a bunch of RX-7s Miatas, Skylines and Mitsubishi GTOs. The main reason Forza dosen't have the sheer number of cars GT does is the copy of models. And I think it is laziness or something to have cars with almost identical interiors and only 3 of 13 or so premiums.