Camera position changes car behavior

[DriftMethod]

If there is a noticeable physics difference between views, we should look at the fastest drivers. They can lap consistently near record times. Given enough time to practice, they would be able to see a difference in lap times. Only those that don't normally use the roof cam (if that is the supposed "fastest" view) would improve their times. Or do all the Aliens already use roof cam?

My unsupported guess is that it's just how you react to the different perspectives.

 

[eran0004]

So, here's video showing the strange pivots. Notice that the car, in every camera location, pivots at the rear axle MORE OR LESS, which is ok.

First run is cockpit cam, second is roof cam, third is chase cam, last is front cam. On that last run, it could totally be that the conditions are not controlled, but I swear I see the rear end sliding back and forth more than in the others.



Maybe I am reacting poorly to the front cam and giving too much input, but I am doing it every time and I can't seem to not do it.



As you can see in the video, that is not always the case in a video game. The front of the car is following something, and it isn't the front tires.

I can't see any difference between the runs. You should record it in chasecam though, so the camera is riding along with the car. Makes it easier to see how it actually moves.

And the video doesn't show that the car is following another object. At best it shows that there is some lag in the animation of the wheels when you perform rapid zig zagging, but the camera is so far away that it's hard to tell.

And the bumper camera is not a front camera. It's positioned inside the car, which means that if the car pivots around the camera, it should behave very similarly in bumper, roof and cockpit camera, while chase cam would be the odd one.

 

[thestig911]

Got to try this out. I always use chase cam, I was doing 1:40.5 in a WRX yesterday at Interlagos and could not get any more time out. Maybe roof cam is the key.

 

[miata13B]

And what's with the alternating camber on the front wheels when they go from left to right? I have never seen front wheels do that? Is that something new?

Could it be that they have implemented a dynamic camber into the model? That is the only thing I could think possible.

 

[Tou2istTrophy]

It really does not....

....


....



....

 

[outofspace]

Have you guys noticed that in the auto play video replay while in the main menu the cockpit view camera is different from the one we have? The camera shakes & i think it follows the head movement of the driver!

 

[Draggon]

I'm almost 2 seconds faster around Dragon Trail in the Gr.3 Huracan using roof cam vs. bumper cam. I chalk that up to ( for me anyway ) the fact that I can see around the turns so much better. I can see the exit of turn 1 when I'm a 3rd of the way into it on roof cam, while on the bumper cam I can't see it until I'm straightening up for the exit.

 

[Metalogic]

I always use the cockpit cam, it's what I'm used to with other racing sims and feels to me more like I'm actually driving the car. I use a wheel (T300)

 

[phil_75]

Physics doesn't work like that. The reason why you get this feeling of different pivoting centers is probably because when you're using "bumper" cam it's easy to believe that the camera is at the front of the car, when in fact it's in the middle of the car.

Its not.
It really does feel different.
Try it.

 

[JacksHerer]

I've noticed the car feels different in different views, but bumper cam gives the best visibility while still being immersive and feeling fast and exciting.
When I go back to bumper view after playing in VR and visa versa I do notice my car isn't where I think it is, but the experience is so different that I put it down to that.

I will certainly have a go at qualifying in roof cam, but expect I would still race in bumper cam.

 

[GTWelsh]

I have tried all the camera positions on Dragon Trail in the Lambo, and can't get anywhere near, or even remotely close to my cockpit times.

 

[pedrocor]

Different cameras give different perspectives how the car is behaving and is placed on track.

Roof cam gives the best balance between car placing precision and behaving awareness, IMO, so slightly better times.
Chase cam is too far to give you the same precision and the notion of under or oversteer comes slightly later.
Bumper cam gives you precision, but it's harder to judge oversteer.
Cockpit view is very close to roof cam but has sight slightly obstructed by the very same cockpit, so not so fast.

Racing against other cars is another discussion completely different.

It's "normal" your car seems stationary and the track seems to pivot around it, after all, the camera is related to the car not the track.
Same thing when your inside a car, it seems the landscape "travels" while you are stationary. That's why people get travel sick by the way.

 

[phil_75]

So I can literally press square to change cam to go around corners better in the middle of a race if your theory is true ?

No not really.
I can set the same lap times in each view, it is the "feel" that is different.

I am adjusting driving slightly to compensate for the different pivot point.
Everybody has their favourite view ie "feel" where they are most comfortable.

Cockpit feels the most neutral and balanced (roof cam similar but better view) whereas bumper cam feels like you are always on the edge of oversteer/more oversteery and chasecam feels more understeery.

If its just a placebo from where the camera is mounted then its a bloody good one.

(DS4 user)

 

[racer_xis]

Tested roof cam with a car Im used to and a track I can sometimes get top10 times... Felt different, but times were almost the same, even a little bit slower than bumper cam.
I think its a perception issue.

 

[baldgye]

To me it felt like the steering was different per view. Chase cam always seemed far stiffer to me than bumper, and bumper always seemed more twitchy than cockpit.

I have tried all the camera positions on Dragon Trail in the Lambo, and can't get anywhere near, or even remotely close to my cockpit times.

Yeah you me both. I think the real aliens are people who can set the same sorts of times in all views!

 

[eran0004]

Its not.
It really does feel different.
Try it.

Feeling different doesn't mean that it actually is different. The camera has absolutely no impact what so ever on the driving physics.

Here is how you can test the pivot point theory: Find a tarmac area that is big enough that you can do a full circle at low speed and full lock. While you drive in this circle, change the view a couple of times. Then watch the replay and look for any changes in the path that the car takes.

 

[ZHPbimmer]

This theory is ridiculous. Why would PD go through the hassle of changing the physics based in the camera view when they can’t even push out more tracks or have nearly as many cars as other racing games?

It’s your perspective, bottom line. Bumper cam has a much closer depth of field than cockpit or chase so it allows you to attack those corners and find the apexes much easier, as well as judge how close you are to another car so you can pass them or cut them off as close as possible. I’ve noticed this in all racing games years ago. It’s one reason why I️ don’t like bumper cam because it is not realistic at all. If you can see what is directly in front of you from the bumper it makes it a lot easier VS having to judge it from the cockpit.

 

[zzz_pt]

Feeling different doesn't mean that it actually is different. The camera has absolutely no impact what so ever on the driving physics.

Here is how you can test the pivot point theory: Find a tarmac area that is big enough that you can do a full circle at low speed and full lock. While you drive in this circle, change the view a couple of times. Then watch the replay and look for any changes in the path that the car takes.

Streets of Willow has a big run off with a circle in the surface. That might work for those who want to test this theory.

 

[Odyofil]

Tried 1000 times.
In cockpit view, steering wheel turns faster, reacts faster, and car is more snappy than chase cam. (Im a chase cam user btw)

And no, it is not an illusion.

 

[zzz_pt]

Maybe a good spot to test it too would be the license test where you have a slalom around cones. If this is true, using different cameras will lead to different times. Ofc you'd need to do it hundreds of times to get some presentable data.

 

[baldgye]

Maybe a good spot to test it too would be the license test where you have a slalom around cones. If this is true, using different cameras will lead to different times. Ofc you'd need to do it hundreds of times to get some presentable data.

Couldn't you measure it mathematically using replay data?
Find a track with solid lines people could measure and then try turning at a set speed in all the views and seeing how the car moves in respect to said lines?

I'm dyslexic so I don't really understand maths, but couldn't you figure out the pivot point from the angle of the car relative to the speed and distance it travelled?

 

[MatskiMonk]

This thread is funny. Like threads about the Earth being flat and only 6000 years old.

I would actually give it a go, but because I'm hopeless with anything other than 'bumper' cam, all my lap times will prove is that I'm not adaptable to changing views.

For most players, the only clues you get as to what the car chassis is doing are visual ones, is it any wonder that moving the location and field of view will affect peoples ability to drive the car precisely the same each time? FOV affects how much of the surroundings are squeezed into a fixed width view, therefore the rate at which the scenery crosses your TV (i.e. how fast you perceive the car is turning, and therefore the perceive rate of change of rotation) will vary with every degree of FOV change inherent in the different views.

There might as such be different controller input smoothing going on to compensate... I could imagine that could be the case - but I think to suggest that the physics model changes depending on view is ridiculous. Lap times are not a scientific way of proving this, it's like claiming the Earth is flat because the horizon looks like it is. It's all about how you as a player react to pretty much the only visual cue you have.

If people are finding they're able to shave seconds off their laptimes using a view they normally wouldn't, I can only suggest saving the replays and watching them back in the view you do normally use - and looking for the differences in line and braking point.

 

[JacksHerer]

If you're not adaptable to changing views, then you may still believe the earth is flat and only 6000 years old

 

[baldgye]

This thread is funny. Like threads about the Earth being flat and only 6000 years old.

Not really that comparable... those two things are false and can and have been proven to be false. This thread seems to be proposing an idea and everyone else seems to put forward their opinions on it, rather than any really meaningful testing... unless I've missed something?

 

[GTWelsh]

Tried 1000 times.
In cockpit view, steering wheel turns faster, reacts faster, and car is more snappy than chase cam. (Im a chase cam user btw)

And no, it is not an illusion.

In cockpit cam you're sat right near the front wheels, so the input from the wheels has a large and immediate effect on the camera.

In chase cam you're miles away from the wheels, so the effect on the camera is very slight. This plus the game allows the car to move independent from the camera so that will reduce the sensation further.

It's like when you turn look-to-apex head-movement on in other games/sims. It makes the car feel like it does nothing but understeer all over the place. It's all just how it "looks" due to the camera movement and position in relation to the front wheels.

 

[Voodoovaj]

This theory is ridiculous. Why would PD go through the hassle of changing the physics based in the camera view when they can’t even push out more tracks or have nearly as many cars as other racing games?

Ok, this is going off track.

No one is saying that the PHYSICS are changing. The physics are NOT changing.

What I am saying is that the car is behaving differently and something is causing that. The behavior change is real, the question is, what is causing it? So, is it that I am merely giving more steering input and causing a behavior change because of the camera view OR is the car following the camera, meaning that when you change view, you move the target.

If you want to see the camera movement best, sit in the pitlane of Nurb 24 and cycle the cameras. You can see the huts in the pitlane are viewed from four distinctly different angles and that the cockpit view is further back than the (camera that no one wants to call) bumper cam.

In my video, you can clearly see that the front of the car is not following the front tires. The front is shuffling side to side. So, if the car is not following the front tires, what is it following and is it affecting the drive?

 

[GTWelsh]

Ok, this is going off track.

No one is saying that the PHYSICS are changing. The physics are NOT changing.

So, here's video showing the strange pivots. Notice that the car, in every camera location, pivots at the rear axle MORE OR LESS, which is ok.

First run is cockpit cam, second is roof cam, third is chase cam, last is front cam. On that last run, it could totally be that the conditions are not controlled, but I swear I see the rear end sliding back and forth more than in the others.

[video /]

Maybe I am reacting poorly to the front cam and giving too much input, but I am doing it every time and I can't seem to not do it.

As you can see in the video, that is not always the case in a video game. The front of the car is following something, and it isn't the front tires.

I tried to avoid going into this but I've just properly read this comment.

The car you see isn't the thing that's reacting to your inputs. The tyres aren't dictating where the car goes, it's all a lie, your inputs are observed by the physics engine, which then tells the car what to do.

The wheels have a bug where they don't react quick enough to user input on replays in some circumstances.

But just because the wheels havn't turned quick enough doesn't mean the car can't turn. The physics engine knows what it expects and moves the car accordingly. The front tyres (the ones you see, at least) have no real input on anything.

At least that's my understanding of physics engine vs graphics.

 

[Voodoovaj]

I tried to avoid going into this but I've just properly read this comment.

The car you see isn't the thing that's reacting to your inputs. The tyres aren't dictating where the car goes, it's all a lie, your inputs are observed by the physics engine, which then tells the car what to do.

The wheels have a bug where they don't react quick enough to user input on replays in some circumstances.

But just because the wheels havn't turned quick enough doesn't mean the car can't turn. The physics engine knows what it expects and moves the car accordingly. The front tyres (the ones you see, at least) have no real input on anything.

At least that's my understanding of physics engine vs graphics.

THIS. Yes. The graphic car and the code/physical car are not the same thing. I was hoping the graphical car would show something but it really appears to just pivot on the center of the rear axle.

I was driving the RSR around Nurb GP seeing if I could beat my bumper time. In the hairpin with the cockpit cam, the RSR is loose. In the hairpin with the bumper cam, it isn't. That got me thinking about the RS01 bug.

Now, this may be related to THIS thread about the RS01 because I tried to recreate this bug and I CANNOT for the life of me make it happen. These guys were going into the hairpin and just cruising around it with no gas or brake and the RS01 was spinning. I was dead set that it was user error. Why was that my opinion? Because in my game, it is IMPOSSIBLE for me to recreate this bug. I couldn't do it back then and last night it dawned on me that maybe it was related.

I have tried all night and all this morning because I was sure this would show what I am seeing, but after what has to be 100+ attempts, I cannot spin the RS01 in that corner the way the guys in the video do it. It simply does not happen for me. Not even close. I want to apologize for being so adamant that their bug wasn't happening because, apparently, We are not playing with the same equal experience.

We are all assuming that everyone has the same things happening, but the grip bug showed us that some people have an extreme advantage. What if my game is different? It's definitely different than the game of the people with the RS01 bug. What if, the reason I feel such an undeniable difference is that mine is, legit, not the same?

@Ryan Dungey said he was doing 1:35's with the Huracan with TC at 2. Was that Huracan stock? I can usually get to 1-2 seconds of the top ten, but I can't for the life of me figure out where I am losing 4 seconds at Dragon Trail. What is a normal time? I am flat through most of that track and my fastest is 1:39.2.

 

[MatskiMonk]

The wheels have a bug where they don't react quick enough to user input on replays in some circumstances.

Is that a bug? The physics model wheels will always react slower than controller inputs because a controller can go from lock to lock in a fraction of a second, a steering wheel, steering rack, wheels and tyres don't, therefore nearly all controller inputs are only dished out from your thumb to the physics engine at a certain rate. This rate is/was also speed sensitive, and that was easy to prove in GT6 using the motec data export. I'd say it was a bug if the graphical representation of the wheels didn't match the physics model of the wheels (in terms of steering angle anyway), but I doubt it's a bug if they don't match thumb inputs.

 

[Ryan Dungey]

THIS. Yes. The graphic car and the code/physical car are not the same thing. I was hoping the graphical car would show something but it really appears to just pivot on the center of the rear axle.

I was driving the RSR around Nurb GP seeing if I could beat my bumper time. In the hairpin with the cockpit cam, the RSR is loose. In the hairpin with the bumper cam, it isn't. That got me thinking about the RS01 bug.

Now, this may be related to THIS thread about the RS01 because I tried to recreate this bug and I CANNOT for the life of me make it happen. These guys were going into the hairpin and just cruising around it with no gas or brake and the RS01 was spinning. I was dead set that it was user error. Why was that my opinion? Because in my game, it is IMPOSSIBLE for me to recreate this bug. I couldn't do it back then and last night it dawned on me that maybe it was related.

I have tried all night and all this morning because I was sure this would show what I am seeing, but after what has to be 100+ attempts, I cannot spin the RS01 in that corner the way the guys in the video do it. It simply does not happen for me. Not even close. I want to apologize for being so adamant that their bug wasn't happening because, apparently, We are not playing with the same equal experience.

We are all assuming that everyone has the same things happening, but the grip bug showed us that some people have an extreme advantage. What if my game is different? It's definitely different than the game of the people with the RS01 bug. What if, the reason I feel such an undeniable difference is that mine is, legit, not the same?

@Ryan Dungey said he was doing 1:35's with the Huracan with TC at 2. Was that Huracan stock? I can usually get to 1-2 seconds of the top ten, but I can't for the life of me figure out where I am losing 4 seconds at Dragon Trail. What is a normal time? I am flat through most of that track and my fastest is 1:39.2.

A completely stock huracan GT3. Only changes I madebwas brake bias at 2 on the front and TC at 2