Re-timing runs as a moderator
4 years ago
Aquitaine, France

Hi everyone

I'm a moderator for the mobile game Rider, a game that requires precise timing to the millisecond.

When I need to re-time a run, weather the runners asks it from me or not, I like to do it frame-by-frame (using Sony Vegas), for the most accurate timing. Sometimes though, runs look accurate enough, so I just check with LiveSplit and verify them if I get around the same time.

However this got me thinking about timing in general in speedrunning. Humans can't get a timing right to the milliisecond by just splitting with LiveSplit or using a regular clock, but yet 90% of the time when a runner submits a run using LiveSplit, it gets verified as it is, trusting the runner's start and stop of the timer (this is often the case for longer runs that actually need LiveSplit, which isn't the case for Rider, except for one category).

--> So what should I do? I can't re-time some runs frame-by-frame and verify some others by roughly using LiveSplit. What are you guys's experience with moderation and timing? Do you have any advice?

Thanks! :)

Modificato da l'autore 4 years ago
Laika_the_Spacedog piace questo
Israel

As a moderator for some games, and a runner of some other games in the past few months, I can gladly share my experience.

In terms of timing, you can split games into 3 categories:

  • In-game timer. This case is the easiest one, the game does all the timing work for you. This timer also usually determine if you will have milliseconds in your time or not.

  • RTA time without milliseconds - Usually for longer games (5 minutes and up). Timing is lax enough to get into a certain range of a second (xx.00 up to xx.99 are all regarded as xx.00). In those cases, Livesplit time might be enough for many cases, but it's not accurate - the runner might start the timer too early or too late, and so for the final split of the run. If you can see that the runner started and ended the timer very closely to the relevant start/end frames of the run, and the timer ends up on the range of about xx.25 - xx.75, you might say it's good enough to be regarded as a final time of xx without any deeper checks.

  • RTA time with milliseconds - Usually for short games (5 minutes and down), IL runs, or some games like Super Mario Bros which are heavily contested and optimized. In those cases (like your game Rider), Livesplit is just not good enough. Every frame matters here, and so a frame-by-frame timing (like you do on Sony Vegas) is a must. It only takes for the runner to start the Livesplit timer 1 frame early or 1 frame late, and bam, the final time is WRONG. For example, take a look at this leaderboard: https://www.speedrun.com/blym (IL run of world 1) At first the mods there used the Livesplit times of the runner for final timing, but as this category became more popular (and very close for top runs), we changed it to exact frame-by-frame timing.

There are exceptions to those, of course. For example, in this case: https://www.speedrun.com/rwk the game is very short but we don't have milliseconds. If two runners get the same time in seconds, they are at the same place, and we are ok with that.

So, for my own experience: In the games which I moderate, at first I configured them with milliseconds, and used my own (inaccurate) Livesplit time for my submissions. Gradually over the months, as I became more experienced with configuring games leaderboards and rules, and after running some other games, I came to the conclusion that milliseconds aren't really interesting for the most part (this is my personal opinion), and removed them from most of my games, and fixed my times accordingly, and changed the rules. However, any IL runs in my games still have those milliseconds.

About Livesplit - As I said, at first I verified runs (for others and myself) with Livesplit, but now I ignore Livesplit completely (when checking other people runs, and ALSO when submitting my own runs). When you think about it, Livesplit is a useful tool but mostly for the runners themselves, to know if they are doing good or bad enough, where they can improve, and some other stuff. Also, remember that Livesplit can be "retimed" by the runner - I used to do that myself, in case of a run where I did the splits in many "off" times, so I improved them by recording the recorded video of the run again, with new overlay of Livesplit where the split times are predetermined by a script. Why do I say this? Imagine a case where I run the 50 points category of your game, and get a real time of 21.450 (50 milliseconds, or 3 frames, behind the current first place). I can retime my livesplit in such a case where the final Livesplit time in my video will show 21.383, and so I "magically" got to the first place.

In conclusion - as a moderator, I feel like we should always set exact rules for frame-by-frame timing which will be easy for all runners to understand and follow, and determine the time according to those rules, and those rules alone. Especially for your case, where you have RTA timing with milliseconds.

Modificato da l'autore 4 years ago
Gaming_64, soru e 3 Altri ti piace questo
Aquitaine, France

Okay all that makes a lot of sense. I think I'll add to the rules that each time has to be calculated frame by frame (fbf) for IL, but that it's not necessary for the main categories. I also think I'll get rid of milliseconds for the 100 points and All Levels categories, because they don't have many runs and milliseconds are quite useless. So basically from now on I'll double check every IL run and 50 Points run fbf, I think it's for the best, even though some people will probably struggle with fbf timing, but it's better for the game I think, especially with some ILs having very close times.

Thanks for your comment it helped a lot!

Anyone else feel free to share if there's anything else you'd like to mention of course

Modificato da l'autore 4 years ago
Norway

In my opinion the runner should add a timer in their video editing software before publishing the run. Start a timer on the first frame and hold the timer after the last frame. Splits are only an approximation and guideline to the runner and stream watchers to how well a run is going, it's not a conclusion.

Example:

Modificato da l'autore 4 years ago
United States

@Catna No, putting a timer in in post is meaningless. It doesn't give any real frame of reference and only tells you information you already knew from the video itself. It also doesn't prevent cheating by editing the video speed, since the timer is in post and runs in real-time where the video might already be sped-up. Best to either have the timer be part of your layout to begin with or not to bother with it at all.

I use the video timer as a reference, but not what I explicitly go off. I find when the run starts (to the frame), and note what the timer's offset is at that point (either how early it started or how many frames it starts late). I find when the run ends (to the frame), and note what the timer is at (assuming it's not stopped early). From here I can calculate easily what the timer is. If the timer stops early I instead calculate how many frames the timer stopped early and add on that time before calculating what the run time was (via the initial offset I determined). This isn't 100% precise as the livesplit capture isn't always perfect but for runs where an exact framecount isn't necessary it is close enough that you will get the correct second almost all the time. If it's very close I might use what I do below anyway.

If the runner only shows seconds (or even only tenths of seconds sometimes), the above method doesn't work too well. In that case I find the second the run starts, calculate how many frames into that second the run starts, and note that down. At the end of the run, I find the second the run ends and do the same calculations. This gives me a run time accurate to +/- 1 frame of the actual time, while still being a quick method for calculating it. The only real way to get more accurate is to actually do a framecount, which takes far longer.

Dawned, Quivico e 3 Altri ti piace questo
Norway

Perhaps it's not helpful for the moderator, but it's not meaningless as a whole. It helps people viewing the run, those that assume the speedrun timer is correct/accurate when it might not be. People can only work with what is presented to them. The more accurate information the better. If you only present a speedrun timer with an offset from the actual time, that is the info most people will receive.

Modificato da l'autore 4 years ago
72pinkush piace questo
Aquitaine, France

Thanks everyone for your answers! I really have a clearer view of what a moderator is supposed to do now, and I needed it, because exactly as @Liv said, I might have been lazy on some runs in the past and just verified them without checking the exact time... Won't do that again for sure.

Now the only thing I'm wondering is weather I should re-time all the runs or just say "from now on, I'll check every run time frame-by-frame so please try to be accurate", because in one way or another, it could be a handicap for some runners...