TrackMania

[Friker]

Btw - just few months ago it was discovered that the replay system of TM is based both on position of a car and keystrokes during the ride. Karl Jobst says it is unusual thing in racing games - only position of a car is usually stored in replays. And - because of analyzing inputs (frequency of tapping) they discovered pretty big cheating scandal. If you have time, you can watch:

• Wirtual's analysis - he is a very very good rider and he is very active in community - https://www.youtube.com/watch?v=yDUdGvgmKIw
• Karl's video - he is doing mostly speedrunning content. Lately he is also creating content about cheating while speedruning - https://www.youtube.com/watch?v=Ww7x5elophk

[Daniel3D]

The replay system is "exactly" the same principle as stunts.
There was a track designed contest based on replay data. The car inputs were the same but by using power ups or track elements totally different tracks were raced.

[Cas]

This is very interesting. So after watching the videos, I quickly made this little program. I can compile it for GNU/Linux and Windows, but I want to make some changes to it, make it better, so to quickly post it, I only compiled it for DOS so far, which is enough, as it's super simple. The program is called "FTAPS"

You call FTAPS with a replay file name and it reads through it and it calculates the number of fast actions in the replay. There are two types of it being counted. Any of the two will, in a different way, give you an idea of the likelihood of a replay having been either tampered with or recorded at low speed. Normal replays will sometimes yield zero, but more commonly result in some positive value, but not too big. Honestly, I don't think this will be super useful for Stunts as it is for TrackMania because Stunts has a precision of 20 frames per second, which isn't much and TrackMania probably has some 60 frames per second or similar... but... who knows?