I did a WR run today that is clocked at 10:22:200 (it's awaiting approval) and started looking at the level beat time differences between my and Sleev's WR runs. After summing up the differences it seems that overall I beat all the levels around 6.02 seconds faster than Sleev. So that would mean that I lost around 6 seconds in level select.

6 seconds over 11 level selects means over half a second lost for each level select. I don't think I'm that slow at menuing to warrant such a difference so I started looking at time differences between my and Sleev's runs in the moments that we can't control, which are 2 things:

1. Time spent between clicking Start in level select and first frame in which movement inside a level happens.

2. Time spent between completing a level, and the first frame that level select screen shows itself.

All those events are easy to observe and pinpoint, thanks to visual and sound cues.

I measured my times on event 1 on 4 different occasions, and it turned out to be 2.98 seconds on average, event 2 turned out to be around 1.55 seconds on average. There are variations within those times, but they didn't exceed 0.05 seconds. Event 1 happens 12 times in a full run, and event 2 happens 11 times, which gives us 35.76 s and 17.05 s respectively cummulated for both, which sums up to around 52.81 seconds.

I did the same for Sleev's run. Event 1 measured at around 2.15 seconds on average, and event 2 at 1.20 seconds on average. Cummulatively, they're both 25.8 s and 13.2 s, which sums up to 39 seconds.

So there seems to be an almost 13 second difference from that, although I don't think myself it is accurate because that would mean that Sleev actually lost 7 seconds in menuing to me, and it just doesn't seem right. But the purpose of this thread isn't to calculate the actual difference between my and Sleev's run, but to point out that there are discrepancies in loading times between PC and Quest versions, which can be easily explained by VR PC's usually having faster RAM, CPUs, and, in the case of SSDs, storage devices than Quest 2.

I think this warrants Quest version being its' own subcategory, seeing as every Quest has the same specs which should make those load times consistent in each device, which isn't the case with PCs.

One possibility would be to just take the in-game times for each level and add them together instead of timing the entire run. This would technically allow for resting between levels which feels like it goes against what speedrunning should be, but it would have several benefits including standardizing times across platforms and making measurement more consistent.

I've also noticed that Oculus Quest recording audio drifts out of sync during long recordings. If you listen to the ending sound effect on Easy1 vs Hard4, the Hard4 one is noticeably delayed. So either the audio or the video is either too fast or too slow. That could be affecting timing differences. Not ~15 seconds of difference, though, so what you observed is definitely a more important issue to resolve first.

Crap, you're right about the recording delay. I noticed in my first run that audio ends sooner than the video so what I did is stretched the audio to the video's length and it sounded to be on point. However with the second one it was actually delayed, so I played with it a little until it was synced, and retimed it and it turned out to be 10:19:900, so sub 10:20! So, not only is it desynced, it is also inconsistent so you have to manually stretch the audio section to fit with the video section, assuming video is the source of truth, but it seems to be that way, as when synced with the video, the time between the timer start and end sounds correspond with the timer at the end of the level. It's annoying to say the least.

About your idea of counting only the level times, it would lead to the Minecraft situation, but there could also be additional rules, like not spending more than say 10 seconds between end of one level timer and start of another. Then we could count the accumulated level times as in-game time, and overall time as run time. It wouldn't be that hard to verify, since they only take 10-12 minutes, and it would solve the discrepancy problem, so it is a compromise that I think would work, at least going by each video and reading the level end times wouldn't take super long, and people wouldn't have to sync Quest recordings.