FPS Boosting.

Wouldn't making a race out of stunt pieces be a good way to test the cars on a flat road? Just make a giant drag strip with the flattest prop and test from there. I definitely don't have anything valuable to add because I run on a kektop and am having to make do with a slightly fluctuating 40-50fps with various degrees of frame drops. However, I'd like to throw that idea out on the table.

Eliminating curbs and turns from the speed testing somewhat gives us a starting point with the frames discussion.

Fps gives nothing on a flat surface, there is no difference in speed or acceleration.

I already did a test on a city straight. But steve wanted to see something on a flat stunt road.

Having a 100% controlled environment will let you narrow down exactly what is happening, where it's happening, and when it's happening. For example, in Kasumi's city straight video, which curbs did he hit? Which ones did he miss? Did he hit the same ones on the next run? Did he hit them with all wheels or just one side?

If people are going to try and narrow down what's happening beyond a basic statement of "FPS makes the game engine wonky AF", then you have to have some kind of control where the only variables in play (or as few as we can get them to be) are the ones we are looking to test for. So you make a 5 mile long stunt road straight without bumps, then with a 1 "value" bump every "short straight stunt road" piece. Then you do one with a 2 "value" bump at the same distance markers as the 1's, and compare. Then change your FPS and do it again.

And repeat until you get the info you're looking for, which you can then apply to a racing situation with the added variable that you can't really tell where curbs are at and just have to pray to god that you're on the "voodoo line" going down the Spa straight and getting every invisible bump possible, plus all of the visible ones.

scenic I do understand why people prefer footage or at least pictures of the test results, that is a fair point I can not deny. But what raises red flags for me is that going by tests that require a siginificant amount of driver input, giving you too many uncertainties and factors to take into account. Questions like "did they steer too much and negated a boost?" or "is all the tyre screeching neglible or slowing him down?" and "did they hit that strangely angled bump in the road in the same spot every time?". I'm not trusting Story's video because his fastest laps include some number of mistakes, meaning that human error is having a big say in the results. If you want to figure this out properly you have to do these seemingly "duh, that's obvious" tests so that you can rule out certain factors with 100% certainty.

Also what I've been saying with my previous two posts isnt completely going against what others have been saying. Sarah and MindlessRiff do poke their science sticks at the right things though, although I felt the need to correct Mindless on a few things he still got the core concept right. In my posts so far I havent given away the right answer, because I want people to think for themselves. I have realized with all my car tuning shiz that giving people the answer right away kills curiousity, they will go with whatever you told them is supposedly "right" or "better" and never question it and tell others how their differing ideas are wrong, which is very bad.

So what's the damn answer then? Bumps, elevation change, dips & physics update frequency. ...And some mechanics knowledge.

Curb/Bump boosting works by hitting the suspension limit of the vehicle, or in the case of tyre clipping cars hitting the rims rather than the tyre, will seemingly "de-ground" the bumped wheel for a split second upping the engine RPM somewhat if it's a driven wheel. Once the wheel settles back down that RPM increase gets translated back as additional acceleration for a moment. Do note that you cant hit bumps too hard or too frequently or you wont give the car enough time to resettle and speed up. This, I suspect, is where these odd framerate "sweet spots" come from, whatever that ends up resonating with the physics engine and its mechanics.

So how can Story be so damn fast around Spa and Serenniel's speed test in the same area show such differences? It's a combination of everything at play, really. That stretch of road isnt as smooth as it might appear, the Nero has bad suspension so it boosts really easily and in Serenniel's case the 100+ framerates were never stable, so it's hard to tell if the advantage comes from fps alone or that slight "jitter". If you've ever done stunt races with tubes then you would know that essentially wallriding the tubes is the fastest way to gain speed and shoot ahead. Cars with bad or limited suspension tend to win out, like the Nero or Zentorno with their AWD further aiding the boosting as you bump into each tube segment.


Summary:
So the factors we have to consider when reviewing footage/results are these:
 * Elevation change - does the track have a lot of chances to hit the suspension/rim limit?
 * Car configuration - is it AWD or simply 2WD? AWD cars will gain far more from bumps than RWD or FWD. ref
 * Human influence - can the car drive itself for most of the test or is the driver required to apply a lot of precise input?
 * Were there any oddities? - Was there a large amount of deviation in car behaviour from one run to another? If so, the samples may have to be discarded because of noise unless the differences are directly related to what's observed.

With that last two factors Story's comparison falls flat, as his 60fps run showed less precision on the driver's end and even slid at least once losing a ton of speed. His comparison is greatly skewed in the wrong direction because of large amounts of human error, the comparison is thus invalid. That leaves us with Serenniel's big comparison still showing strange behaviour but is unfortunately also in an area of the map where it's hard to predict surface changes, also making that test a little dodgy to analyze but it could potentially still be valuable.