I saw chip's post about this: https://www.speedrun.com/sexy_hiking/thread/5mdx5 To summarize he says he wants to retime runs to ensure consistency, and then goes ahead and does it. He posts a spreadsheet of what he has done and the job is done. But I have a problem with some of this work. He says he adds .33s if someone started their timers while the hiker is standing on the ground as opposed to naturally loading into the level. This is a problem.
The time it takes for the hiker to fall down to the ground and settle varies from level to level. Every level except 3 the hiker takes longer than .33s. I've recorded myself loading into each level 20 times and analyzed the recording frame by frame and these were the most common times for each level:
- .37
- .48
- .33
- .42
Is it vital to add .04 or .15 to a couple runs? No. But much like chip I want the leaderboards to be consistent and future-proof. I have some free time this week to do it and I wanted to discuss it before I take action.
I will look at full game runs tomorrow. edit: I will only retime EYESOCK's Level 1 run, the other retimes are within margin of error and we can look at them later if someone ties them so we can decide a winner.
Only a handful of the runs are properly recorded with a timer(thank you) so I just looked at all of them because why not. Even the any% wr is choppy on the timer which causes it to look 0.08s slower than it actually is xd "good" means my timing was within ±0.05 the current time of the speedrun so I won't bother.
Upon further inspection spener's level 2 run is a 5.52. I realized this after how close carotina's recent run got to it. Decided to check if his timer was accurate and it makes his run look slower lol. I double checked.
edit: After thinking about this, I should revert this change and similar changes where the runner has a timer on screen but it's off by a little miliseconds. I ran my own timer and recorded every run, The problem with this is if the recording of the run I'm verifying has dropped frames it can make a run look faster than it is. I will list the changes I make here
edit2: I ended up leaving a note under each run that was retimed.
I will admit I don't love bumping this thread but it has to be done :)
We have an in-game timer now! The way it works is it counts game frames and converts them into seconds. Sexy Hiking runs at 50 frames per second. So each frame is 0.02s long. The in-game timer is great because it makes recording and timing your runs a lot easier when you don't have to deal with an external program like livesplit.
However there's a couple small issues about this method which make the runs that use the in-game timer look 0.04s slower for each level. So a full run would look 0.16s slower than it actually is. The good news is this situation is consistent. The bad news is the timer isn't easy to fix.
Our top priority is to make sure what a runner sees on their screen compares correctly with the times on the leaderboard. It would be confusing and mentally taxing to force the runner to do unnecessary math each time.
So the solution is adding +0.16s to each full run and adding +0.04s to each individual level run. It sucks to make runs appear slower but it's for a good cause.
Why is frame counting causing the runs to look slower?
The way we manually timed runs equates to whichever frame a level ends on minus whichever frame it started at and adding up all 4 levels. So for example in first level that would look like Frame 251 - Frame 1 = 250 frames. Our example run is 5 seconds long. However the in-game timer would show 5,02. And then on the second level, it doesn't start from 5,02 again, it starts from 5,04. just like level 1 started at 0,02 instead of 0. it doesn't make sense as 0.02s haven't passed. but it's a new frame, the 252nd frame. so it shows 5,04.
Ok, but that only adds up to 0.08s for 4 levels. Where does the other 0.08s come from?
Turns out when you touch the angry spinny red face to end a level, the game skips a frame! Going frame-by-frame on recordings, when the runner ends a level, one can see that the timer jumps 0.04s instead of 0.02s like it should. It's also visible if you pay attention to how fast the red face spins, its animation skips a frame. This can be seen in versions without the timer as well. So that's where the other 0.08s comes from.
i will post another message when i'm done adding to non-igt runs.
Thanks for clarifying what the cause of the discrepancy is. Now I see my mistake with submitting the time for MqcCheeze's level 2 run (I should've used the final level 1 time as the start time, not the initial level 2 time, otherwise the IGT ends up being +0.02s compared to the frame count time instead of +0.04s).
You didn't make a mistake chip it's just, confusing xd also i updated every leaderboard except any% leaderboard, i'll start now and complete it this evening. Takes a bit longer than i thought with all the obsolete runs
because of how nice the frame counting in-game timer is, we needed to change how we time runs and that's ok :D
