Hello everyone. There are some big changes coming to Resident Evil HD Remaster. The first big change is we are going to be moving from IGT to LRT. This change is happening so all players are on an even playing field regardless of their PC specs. The way the LRT works is the timer will pause during the door loads and the cutscenes. This is where high end cpus give the runner their advantages. IGT will be a variable available on the board. We want to thank TheDementedSalad for creating this LRT for the community.
Now here is the second big change. Due to how the LRT works, vanilla runs aren't losing 6.5 seconds per door which makes it comparable to doorskip runs. This effectively removes the need to separate vanilla and DS. Analog skating does not hold enough weight to have separate categories that will be identical in terms of the route. If any runner wants to use doors, they can now analog skate. Runners that wish to run the game "pure" can still use doors and mash if they want. Runners who want to use doors WILL NOT be at a disadvantage. Runners who want to mash WILL be at a disadvantage. Showing your inputs will be required for any submission.
Every run will be retimed to fit the new LRT. TheDementedSalad created an ASL that only tracked the loads on the doors. Several runners did tests so that we could find the medians required to retime these runs. The results can be found on a chart below.
https://docs.google.com/spreadsheets/d/1jauLUHuimHfA0a0U4bCsbPnnmq4-0oUaaO0IDzWqhO8/edit?pli=1#gid=0
I should mention that the medians found for runs with the old doorskip were just off of my machine. No active runner uses the old doorskip, so finding people to test would have been difficult. Additionally, Salad made another ASL that paused during the cutscenes as well as the doors so we were able to find out exactly how much time we must subtract when it comes to cutscenes. The amount of doors shown below might seem a little off, but the game sees several rooms as one big room, therefore it doesnt split when going through said doors. The LRT still pauses on those doors so that had to be put in the equation.
Runners that have submissions in both vanilla and DS, we will take the fastest time between the two and put it on the board.
We know that this system is not 100% flaw free, but we do believe that competitive integrity is on the line with these changes.
You can find all of the LRT calculations under this post. Feedback is much appreciated. We will begin work on the boards this week.
For new DS, we gathered 1,840 numbers across 13 different machines and the median was 0.53.
Jill Any%: 248 x 0.53 = 131 seconds. 131 = 2 minutes and 11 seconds. 24 seconds of cutscenes. Subtract 2 minutes and 35 seconds for LRT.
Jill Best Ending: 262 x 0.53 = 138 seconds = 2 minutes and 18 seconds. 29 seconds of cutscenes. Subtract 2 minutes and 47 seconds for LRT.
Jill RS Any%: 250 x 0.53 = 132 seconds = 2 minutes and 12 seconds. 25 seconds of cutscenes. Subtract 2 minutes and 37 seconds for LRT.
Jill RS Best Ending: 264 x 0.53 = 139 seconds = 2 minutes and 19 seconds. 31 seconds of cutscenes. Subtract 2 minutes and 50 seconds for LRT.
Jill Knife Only: 260 x 0.53 = 137 seconds = 2 minutes and 17 seconds. 22 seconds of cutscenes. Subtract 2 minutes and 39 seconds for LRT.
Jill NG+: 244 x 0.53 = 129 seconds = 2 minutes and 9 seconds. 24 seconds of cutscenes. Subtract 2 minutes and 33 seconds for LRT.
Chris Any%: 264 x 0.53 = 139 seconds = 2 minutes and 19 seconds. 17 seconds of cutscenes. Subtract 2 minutes and 36 seconds for LRT.
Chris Best Ending: 278 x 0.53 = 147 seconds = 2 minutes and 27 seconds. 23 seconds of cutscenes. Subtract 2 minutes and 50 seconds for LRT.
Chris RS Any%: 271 x 0.53 = 143 seconds = 2 minutes and 23 seconds. 17 seconds of cutscenes. Subtract 2 minutes and 40 seconds for LRT.
Chris RS Best Ending: 298 x 0.53 = 157 seconds = 2 minutes and 37 seconds. 23 seconds of cutscenes. Subtract 3 minutes for LRT.
Chris Knife Only: 264 x 0.53 = 139 seconds = 2 minutes and 19 seconds. 17 seconds of cutscenes. Subtract 2 minutes and 36 seconds for LRT.
Chris NG+: 264 x 0.53 = 139 seconds = 2 minutes and 19 seconds. 17 seconds of cutscenes. Subtract 2 minutes and 36 seconds for LRT.
For Vanilla, we gathered 1,416 numbers across 11 different machines and the median was 6.51.
Jill Any%: 248 x 6.51 = 1614 = 26 minutes and 54 seconds. 24 seconds of cutscenes. Subtract 27 minutes and 18 seconds for LRT.
Jill Best Ending: 262 x 6.51 = 1705 = 28 minutes and 35 seconds. 29 seconds of cutscenes. Subtract 29 minutes and 4 seconds for LRT.
Jill RS Any%: 248 x 6.51 = 1614 = 26 minutes and 54 seconds. 25 seconds of cutscenes. Subtract 27 minutes and 19 seconds for LRT.
Jill RS Best Ending: 251 x 6.51 = 1634 = 27 minutes and 14 seconds. 29 seconds of cutscenes. Subtract 27 minutes and 43 seconds for LRT.
Jill Knife Only: 260 x 6.51 = 1692 = 28 minutes and 12 seconds. 22 seconds of cutscenes. Subtract 28 minutes and 34 seconds for LRT.
Jill NG+: 244 x 6.51 = 1588 = 26 minutes and 28 seconds. 24 seconds of cutscenes. Subtract 26 minutes and 52 seconds for LRT.
Chris Any%: 264 x 6.51 = 1718 = 28 minutes and 38 seconds. 17 seconds of cutscenes. Subtract 28 minutes and 55 seconds for LRT.
Chris Best Ending: 278 x 6.51 = 1809 = 30 minutes and 9 seconds. 23 seconds of cutscenes. Subtract 30 minutes and 32 seconds for LRT.
Chris RS Any%: 271 x 6.51 = 1764 = 29 minutes and 24 seconds. 17 seconds of cutscenes. Subtract 29 minutes and 41 seconds for LRT.
Chris RS Best Ending: 298 x 6.51 = 1939 = 32 minutes and 19 seconds. 23 seconds of cutscenes. Subtract 32 minutes and 42 seconds for LRT.
Chris Knife Only: 264 x 6.51 = 1718 = 28 minutes and 38 seconds. 17 seconds of cutscenes. Subtract 28 minutes and 55 seconds for LRT.
Chris NG+: 264 x 6.51 = 1718 = 28 minutes and 38 seconds. 17 seconds of cutscenes. Subtract 28 minutes and 55 seconds for LRT.
For old DS, we gathered 150 numbers from 1 machine and the median was 1.06
Jill Any%: 248 x 1.06 = 262 = 4 minutes and 22 seconds. 24 seconds of cutscenes. Subtract 4 minutes and 46 seconds for LRT.
Jill Best Ending: 262 x 1.06 = 277 = 4 minutes and 37 seconds. 29 seconds of cutscenes. Subtract 5 minutes and 6 seconds for LRT.
Jill RS Any%: 250 x 1.06 = 265 = 4 minutes and 25 seconds. 25 seconds of cutscenes. Subtract 4 minutes and 50 seconds for LRT.
Jill RS Best Ending: 264 x 1.06 = 279 = 4 minutes and 39 seconds. 31 seconds of cutscenes. Subtract 5 minutes and 10 seconds for LRT.
Jill Knife Only: 260 x 1.06 = 275 = 4 minutes and 35 seconds. 22 seconds of cutscenes. Subtract 4 minutes and 57 seconds for LRT.
Jill NG+: 244 x 1.06 = 258 = 4 minutes and 18 seconds. 24 seconds of cutscenes. Subtract 4 minutes and 42 seconds for LRT.
Chris Any%: 264 x 1.06 = 279 = 4 minutes and 39 seconds. 17 seconds of cutscenes. Subtract 4 minutes and 56 seconds for LRT.
Chris Best Ending: 278 x 1.06 = 294 = 4 minutes and 54 seconds. 23 seconds of cutscenes. Subtract 5 minutes and 17 seconds for LRT.
Chris RS Any%: 271 x 1.06 = 287 = 4 minutes and 47 seconds. 17 seconds of cutscenes. Subtract 5 minutes and 4 seconds for LRT.
Chris RS Best Ending: 298 x 1.06 = 315 = 5 minutes and 15 seconds. 23 seconds of cutscenes. Subtract 5 minutes and 38 seconds for LRT.
Chris Knife Only: 264 x 1.06 = 279 = 4 minutes and 39 seconds. 17 seconds of cutscenes. Subtract 4 minutes and 56 seconds for LRT.
Chris NG+: 264 x 1.06 = 279 = 4 minutes and 39 seconds. 17 seconds of cutscenes. Subtract 4 minutes and 56 seconds for LRT.
Thanks to everyone who participated in the testing. We really appreciate your time.
When the leaderboard conversions were done, were the differences subtracted from IGT or RTA?
I'm sorry to stir the pot, but I think this warrants a discussion.
is LRT's base time from RTA or IGT?
As far as I can tell, it's RTA? When I reset the game, sometimes the counter continues to go up, and the timer literally keeps counting up when I kill the REmake PC executable. There were several occasions where the LRT autosplit script kept running when I would reset the game as it didn't read the memory value correctly. The timer on livesplit didn't reset to 0, so that means the LRT does not draw time from the in-game timer.
If that's the case, then it doesn't make sense to convert the old times from IGT as IGT never had parity with RTA to begin with.
If IGT for HD remaster on PC does have mysterious gained/lost seconds (and sometimes minutes) to begin with, then doesn't that make conversion from IGT to LRT wholly erroneous?
The LRT is just RTA minus door loads/cutscenes. Unfortunately retiming old runs by deducting door loads from IGT was very inaccurate. The IGT behaves strangely the higher the framerate and actually starts to count slower throughout the run, mostly when playing at 120 fps, making longer vanilla runs have even less parity when compared to new runs using the LRT, so I'd say older retimed runs (especially vanilla) have an advantage on the boards over new runs using LRT.
Right, that's what it looks like to me.
Since I stopped modding forever and a half ago, it seems like people figured out IGT is unstable as a result of variable framerate but nobody really picked up on that until recently because everyone (including me) assumed it was all based on framerate, even with VFR.
Probably half-true, if not for the game speed fluctuating/speeding up slightly while the IGT is calculated weirdly.
Truthfully I don't even know if IGT locked to 60 is based on framecounting like with older PC RE titles.
Everyone started running on VFR when it was found that running at VFR yielded faster times. But IGT also probably fluctuates with framerate somehow because the game speed doesn't slow down much when framerate drops.
But, are the leaderboard times from Old DS + New DS to LRT all from IGT or RTA? Nyx says IGT. So if all of those conversions weren't done from RTA, then doesn't that mean that all the times on the leaderboard that use LRT conversions are inaccurate by a large margin? Especially since they're subtracted from IGT which is found to be inaccurate to begin with? If it were me, they should all be retimed with new submissions and a cutoff date where IGT should be gradually phased out and the other runs discarded.
That would probably make a lot of people unhappy though so I understand why the conversions were made the way they were, but I think a different more thorough method of retiming should be employed.
Sidenote, when you take into consideration Twitch discarded a ton of VOD highlights, a lot of the runs on the leaderboard don't even have a video anymore so they need to either be linked from YT or be discarded completely.
When I did alot of testing last year, comparing the LRT with the IGT, it seemed like the IGT performed almost properly at 30 fps, over a 1hr 20m period it only counted slower than the LRT by 3 secs (probably because the games original version was locked to 30 fps and it wasnt adjusted accordingly for PC) but as the fps cap was increased, that's where large differences started to occur, at 60fps it turned into a double digit difference, then almost doubled again with the game unlocked at 120 fps. I posted these findings in discord. My Jill Any% pb for example, ran last year using the LRT and most recent door skip, if I had done that run a few months earlier with just IGT and it had been retimed, it would be 10 seconds faster on the boards. With retimed Vanilla runs they could be inaccurate by 15-30 seconds. Retiming runs using RTA would have been way more accurate.
Yeah so to that extent, all of these runs need to be retimed, do they not?
No doubt the current mod team is lacking the manpower, but I think it needs to be done. I think we gotta go a little deeper to figure out the mechanics of the LRT script and the relative timesave compared to Old DS, New DS, etc (ignoring cutscene skips probably). Retime only the runs that are in the top 10 percentile in the most competitive categories since the ones that can easily be improved by whole minutes aren't worth consideration... Unless they want their own run retimed. Nobody wants their runs deleted, but for some of these runs there's no earthly way they'd be that fast post-conversion.
The door load calculations in the figures above seemed pretty accurate for old & new door skip and also Vanilla, the testing was done using an LRT that ran during the loads and cutscenes so it would have been very accurate, although I'm not sure cutscenes should have been part of the LRT as skipping them seemed more user based which would create a bit of inaccuracy for retimed runs. The problem is the IGT just not counting in real time, I don't think this was realised when the LRT was first brought in for testing. If the LRT is RTA then older runs need to be retimed using RTA also, it's the only way that makes sense.
When we were doing these the intention was to keep peoples times on the boards, and if runners wanted a more accurate time they could submit another run using the LRT. When we were doing these I'll admit I pretty much only knew what LRT was and not the specifics of it for this game, nor did I know how inaccurate it would be to use IGT for the conversions. Just never crossed my mind at the time these changes were being made. With that being said, I do agree that these runs need to be retimed. I can't promise anything concrete like a timeframe when these would be done but I'll do what I can. Willing to take feedback and suggestions for how to most efficiently tackle this. Some of the suggestions already made seem like a solid start but if there's anything else I'm willing to hear it.
Well Jill Any% for example, the calculation was 2m 35s for loads +7-8s because the IGT starts 7s sooner, so 2m 43s was deducted from IGT to retime the runs I think? Then If you take the RTA of a run using vod timstamps, starting from when you get control of Jill just like the LRT, it would be that time minus 2m 35s. When I've tried using this method on runs on the board that used the LRT, the retime is way closer, within 3 secs or so. Basically the figures Pessi posted above are pretty accurate and can be used with the RTA time on runs from every category, meaning there wouldn't need to be any extra testing or anything.
Apologies for the wait. Retiming all these runs and getting them updated on the board has taken quite a while. With that being said, as of today every run has now been updated with a more accurate time. This still isn't 100% accurate but its as close as I can get without runners doing new runs with the current LRT. There inevitably will be shifts in some runners spots on the board. But, for the most part, most runs should be in similar placement as they were before. I do apologize for any inconvenience this has caused.
