Should we not just use the in game timer on the save file for our official timing selection? I'm new to the game but this seems like an easier route to normalize times across systems. Looking to get a run going tonight!

I've only set the RTA as of now because I've noticed that the IGT probably isn't very reliable as the IGT on the run I have on the leaderboard was a lot higher than my RTA. It was probably the time of the run + cutscene/credits while it normally only counts the run time.

I'd also be for using the IGT to allow easier and more fair cross-platform timing if it's reliable but if this kind of problem can happen it's going to be a bigger problem.

But we can still add the IGT as the secondary timer and see how it behaves across people to see if it's reliable enough to be used as the primary timer or not.

Yeah it's hard to know whether or not the timer runs with loads or not. It would vastly change timing for my time without loads for the run I submitted yesterday. I wonder how we could test this? Maybe reach out to the devs on twitter and ask if they have specifics about how the IGT works?

It's interesting because my IGT was 23 seconds faster than your time w/o loads, but my RTA was still 9 seconds slower overall. I'm considering getting on the PC as well and playing with that to give me a time w/o loads to compare against.

For the last cutscene/credit, it appears that the playtime stops at the very end of the cutscene just before the credits which were probably the reason why my IGT was higher than my RTA so I think that solved my first problem about the IGT.

I already know exactly how the IGT works since I've made the load remover to work very closely with the IGT (without the issues). Basically the playtime is updated each time the game saves, mainly during scene transitions and some events. At each saving, it adds the time since the last save minus the time of loadings, which is pretty great. BUT the thing that makes the IGT very bad is that it only saves the seconds as a whole (without the decimals) which means that the added time will always be rounded down and then added to the playtime.

For example, if you start and do 2 transitions as of 12.7s and 10.9s your time will be 12+10 -> 22s of playtime. If someone else does 12.1 and 10.2 it will also be 22s while you clearly took almost 2s less.

It's only 2 transitions here but if there are roughly 100 transitions for the Any%, between the best and the worst, the difference could be 100s which makes the IGT very inaccurate.

However, since consoles don't really have the choice to either use IGT or real-time, I don't know if an inaccurate IGT will still be better than using a real-time.

If IGT is confirmed to be inaccurate, it shouldn't be used. Consoles losing to PC because of loading isn't great but false times are worse.

I think it could depend. If for example there are xbox and switch runs and that xbox loads are a lot faster than switch, the IGT will put them on the same level while it would have been a no match with RTA.

But yeah imo, the only fact that the time used would be inaccurate, which would mess all the fair comparisons of any sort regardless of the support is a really bigger problem.

Okay thoughts on starting timing on the first action after the cutscene? Or maybe when we gain control of Ember rather than file select?

You can skip the opening cutscene, which IMO counts as a first action. You can also skip the dialogue before getting control. Is the time between file select and control inconsistent?

It is for switch at least, yeah. I can get the opening cutscene skip at 10 seconds or 30... And nothing I can think of influences this.

My thought would be to start when you gain control of Ember to take a step. This would require retiming for the current runs on the leaderboard and my new WR that is pending, but should be easy enough to do.

I personally consider the cutscene skip and especially the text skips as a part of the run and starting at ember's wakeup would mean ignoring these inputs which would cut off things that can impact the run.

Still, we could agree to move the start to the wakeup input if it can help people to reduce the inconsistent start time since the inputs are very minor but the thing is that if there are inconsistent times for the first cutscene/scenes to load, there are really high chances that it is also going to be the case all over the game.

Okay, after trying some stuff with the intro, here's what I think:

You don't have to be precise when skipping the intro cutscene, you can pre-hold A during the loading screen. Given that the loading screens before and after are inconsistent, maybe timing can start on clearing the first dialogue bubble.