Bug with replays

Moderator: English Moderator

Post Reply
Posts: 234
Joined: 15 Jun 2010, 11:05

Bug with replays

Post by JumperJack » 25 Jun 2020, 20:28


This was initially meant to report a cheater in the TMOne Speed campaign, but appeared to involve a larger underlying problem with Maniaplanet (that might have two or three possible causes, as far as I can see).

The problem arose when I noticed that someone beat my time at A01 of the new TMOne Speed campaign. But while being 0.008 s ahead of my time, his ghost was actually finishing later than mine. And that his ghost stopped moving about 0.01 s before the finish, instead of right after like most other ghosts.

But when I started looking at different replays of myself in the replay editor, I noticed that the same happened at some of my own runs. Some replays finished all the way through, a few others stopped right before the finish.

However, at the same time I also noticed a clear discrepancy between some replays with (nearly) equal times. I'll illustrate:


On the above screenshot you see five cars with the following finish times:
– The yellow car has a finish time of 16.638
– The green cars have finish times of 16.674 and 16.678
– The silver cars both have finish times of 16.706

At this moment, all cars are still moving in the replay editor. The gap between the two green cars seems quite big already for being so close together in finish times, but the order of cars is still correct.


If you go forward in time a bit, the yellow car (.63) suddenly stops moving in front of the finish. This is what also happened with the ghost of the person I was initially planning to report as a cheater, but apparently this sometimes happens. Even though this is weird, what is quite clear is that the car was in front the moment the car stopped. But for how much? I do not have any replays in between .63 and .67, but the moment I drove my time, I was driving against my .67 ghost, and was very surprised to see that I drove that much faster. It felt like I should've had a high .65 based on the distance between me and my ghost.


Go forward even more, and the second thing that is noticeable is the distance between the green cars. While the grey cars have an equal finishing time (and it looks that way), the green cars should only have a 0.004 s difference. But they're nearly a car length apart! On this particular map I have the experience that a car length at the finish is about 0.020–0.03 s difference.

What this illustrates seems to be an underlying problem that might only involve the Speed car and/or environment, or all ported TMUF cars. I figure this issue does not exist for native TM2 environments, because we would've already known by now. This issue could, as far as I can see, have two or three possible causes:

1. The least worrying, but annoying cause would be a problem with how replays are saved. Maybe there's some problem with the synchronisation of them, causing slight delays or accelerations, and the visual difference between (nearly) equal times. That could also explain some replays cutting off before the finish. Practically, it would explain that when driving my .63 time against my .67 ghost, it would seem like a small difference, while it was in fact bigger if the ghost of my .67 would be better synchronised.

2. Much more worrying would be that for some reason, times driven with these cars are not synchronised properly in real time. Thus that the replays are in fact correct, but there is a small difference in how the game interprets the position of the car in relation to the map (and the other ghosts). I'm not sure how this would work exactly.

3. Another possible cause, but less likely, might be that it is just the moment that the replay gets cut off is the root of the problem. That would explain my .63 time being that much faster than the others (while not showing for it), but it would not explain the difference between the two green cars (the .67 times) being that far apart, as they both continue to play until after the finish line.

If you want to see for yourself, here's the combined replay of the five runs: https://www.dropbox.com/s/a4fpubz7bluqq ... y.Gbx?dl=1

Edit: Another replay combination to check whether CP times have discrepancies too. But that's not the case. CP times seem to be very accurate.


Only at the finish you see the discrepancy. Here you can download the combined replay: https://www.dropbox.com/s/0dgg7d9k6adbc ... y.Gbx?dl=1

Here are the CP times of the four cars:

Yellow car: 3.019 / 5.122 / 8.652 / 10.951 / 12.554 / 16.638
Bronze car: 3.069 / 5.158 / 8.678 / 10.955 / 12.566 / 16.674
Green car: 3.034 / 5.118 / 8.628 / 10.905 / 12.524 / 16.678
Silver car: 3.039 / 5.132 / 8.602 / 10.943 / 12.554 / 16.704

Edit: Another screenshot showing myself (16.638), Lanz, (16.664) and Xtreme555 (16.666). As you can see it looks like Lanz and me are very close together, while the it should be Lanz and Xtreme555 when looking at the finish-times.


Also, I don't know why the mods moved this post to the TMOne topic. Florenzius is aware of the problem, and adviced me to report it as an MP bug.
CMC JumperJackツ

Post Reply


Who is online

Users browsing this forum: No registered users and 0 guests