Here's Why You Suck At Forza Horizon 5 (Pay2Win Physics)

Forgot to mention: The game speed is not slowed down alongside the timer. We looked at other runs from the “world record holder” mentioned in this video and some of them are actually faster than humanly possible. It is possible that some of the game physics are running faster than intended, giving him an even bigger advantage, but we are not 100% sure about this one and there are no reliable ways to test it. This is another thing for :_pggKEK: to investigate Correction At 4:28: That's 56 clean runs for 30fps

"he has a better gamong chair" is now officially working in fh5.

"The players, with less tools, doing developer's jobs for free." That hit hard

Interesting to have the physics refresh rate tied to your FPS. Major sim racing titles figured out ages ago that it was a bad idea to do so. Now they have physics updates detween 300 and 900 Hz for increased fidelity.

Undeterministic physics in competitive games is a crime.

Rivals mode's biggest sin is the inability to sort times by cars driven. That way you could pick whatever you want and tally your time with others who used the same car. Otherwise it's 1-2 META cars per track or GTFO. Apart from well known issues plaguing this mode, looks like there's even more, quite significant to anyone that cares or cared at this point...

At one point I know Forza (Motorsport, at least) ran a fixed 180hz physics refresh rate, and also most major racing sim titles run fixed physics refresh rates as well. I suspect this is caused here because at some point, Forza switched to tying the physics refresh to your framerate - or may do this in specifically Horizon titles. It still doesn’t explain the game timer issues, but it would at least explain the physics issues between 30 and 60 fps locks.

Im glad to see this about the physics and the timing. I honestly felt I was going insane and the only one thinking the physics were off. I’d enter a turn with all the same variables and it was almost always different everytime. Like my car will grip the earth like a champ, slide a little, or act like it was on ice. Frustrating to say the least.

"you might have tried dozens of tuning setups with little to no improvements" (shows ugrundy's tunes) this is kinda genius way to explain why there's no impovements lol

People won't believe this is happening in CoD, but you literally just proved it's going on in Forza. Lol

I actually really appreciate your effort in researching this. It’s weird because I had a gut feeling some things were off, like no matter how good I was or how perfect my timing when breaking on upcoming corners was the ghosts always were ahead of me. Seeing this kind of relaxes me and hopes this will eventually be fixed or acknowledged. I like how some one mentioned the ability to filter would be helpful.

Lego island physics.

I know you mention that it doesn't affect online races, but I can't even count the number of times I've clearly beaten someone to the finish line, only for the game to register them as finishing ahead of me. It was a big issue in FH4 as well. I have a sneaking suspicion that these issues are related.

I remember something similar happened to me in The Crew 1: Back then I had pretty old and slow hardware which limited my fps. This lead to the problem that I couldn't drive with/against my friends in multiplayer because there was some sort of correlation betweem FPS and the speed/distance a car drove. I could drive top speed with my LaFerrari and my friends with high fps could just drive past me with some default offroad car lol...

Reminds me of these good old DOS games that run their timers by the CPU fequency and therefore spazzed out on faster systems

I think this is because the delta time is a float of 32 bits and when it multiplies to get the values of the timer and physics it just loses a little bit of precision every time

Having worked with lots of physics simulation programming I think i could provide some insight into possible causes. Things like suspension, tire friction modelling, aerodynamics simulation are bound to be error prone when handled incorrectly. Time interval between the frames of simulation can change the outcome. In the case of the suspension - for example slower simulation rate may cause springs to get more compressed under changes of load on the car (like seen in the video with the cliiping on the loop) because there is more time between the cause (compression of the spring) and effect (spring force and damping acting on the car). Same can happen with tire friction but the time precision errors are usually more pronounced because the response in non-linear. You can search for tire friction curves if you want to see how they look. This can cause a tire to spin further before the forces are applied, therefore making it possible to spin faster than it should and getting less friction as a result. If the simulation results are dependent on the simulation framerate, an unstable simulation framerate will make them more random/non deterministic. Possible solutions to these problems are: running the simulation at a (edit: stable) rate independent of the graphics fps (commonly used, not idea why it's not used in fh5) and numerical integration or running critical parts of the simulation multiple times per frame internally to minimize the errors. It would be interesting to see you test the friction with different fps. Maybe some powerful cars with a lot with wheelspin? No idea about the possible cause of the timer thing, but someone definitely did a shitty job with that.

Have you ever noticed that sometimes when you downgrade the tire compound, the lateral g stat and acceleration stats improve? I'd love to have an explanation for this besides "don't trust the in game stats"

In "Beetle Adventure Racing" Physics also change with different frame rates. PAL N64 runs in 50 fps where american version N64 runs in 60 fps. The console calculates the cars position every frame. Means that if the car experiences any changes in (the track) directions, it calculates whether it gets closer to the ground or further away. Resulting in a change of the cars direction, or not. For example if your going up a ramp, your car wants go into the ground. The cars velocity and direction is always the result of the calculation from the last frame and the players input ofc. So going up a ramp means that the cars calculated direction and the direction of the street do not match, resulting in compression of the springs and a change in direction and velocitiy for the next frame. Long story short. In 60 fps it does that progress more smoothly than in the 50 fps version, resulting in the cars physics and behaviour beeing more volatile. It is more likely that strange things happen with lower fps. Maybe thats whats going on here

awesome research you clearly did a lot of editing and researching. Glad there are good youtubers out there