Forums  /  Left 4 Dead series  /  Left 4 Dead 2  /  Autosplitter that removes loading times
  bill_play3bill_play3

I'm currently working on an autosplitter for LiveSplit and I pretend to release it for everyone once it's done. It should work for every relevant category for this game. Considering that a few people complained about loading times in the past, I thought about implementing a feature on the autosplitter that would remove loading times, removing most of the hardware factor that could be different between runners.
If every one agreed in using this feature on the splitter, then i would make it part of the script. So first of all, does everyone agree with removing loading times? And if so, what could we do to the runs that used RTA timing? My suggestion would be retiming the top 3 runs for each category and leaving the rest for the runner to retime. Mods could add a comment to the not retimed runs or perhaps adding a new column for game time and keeping the current times on an RTA column. Sure, it would be a lot of work. But i guess that this sort of transition happened in every game that now uses timing scripts (like some other Source Engine runs).
So, what are your opinions?

p1xelfartp1xelfart, IvrinIvrin and 9 others like this. 
  bill_play3bill_play3

Also, I tested some things with the load time removal and discovered that the times without loads are put into the "Game Time" section of LiveSplit, so it times both RTA and Game Time. I guess i'll just leave the feature in and then we can decide whether we use RTA timing or Game Time timing

 
  MrFailzzMrFailzz

You've got my full support on this. Don't see why loading times are included since it only prevents people with bad hardware from being able to compete.

mariorules64mariorules64 likes this. 
  TheMasterTheMaster

Agreed that it is time to remove loads as requiring fast hardware to compete is not right. That is now the point the leaderboards are pretty much at, you are going to have a very hard time taking some of the records if you don't have an SSD, or even two in RAID0 like I do. Although this should have been in place a while ago, now that the leaderboard times are much more competitive it's more important now than ever. No Mercy might not even be possible anymore with slow loads, The Sacrifice is probably the same, Swamp Fever, Dark Carnival for certain. Even loading just a few seconds slower on some of those (which is absolutely the difference between HDD vs SSD) will make getting the record far more difficult, if not impossible.

Now that an effective autosplitter exists that can accurately remove loads there is even more reason to use it since it no longer requires knowledge of checking the frames, simply using the autosplitter and submitting the LiveSplit time will work.

As for the runs already up, well, the times can only be faster and worrying about how to fix the current leaderboards isn't a good reason to leave things as they are instead of future proofing right now as waiting longer will only make things more difficult. I would be happy to look at the frames of every submitted run of mine to adjust without loads accordingly and use the autosplitter from this point on. I'll even do one better and volunteer to pull the proper 2nd and 3rd place times as well assuming I can easily download them to open them up. The rest of the runs past that point are generally pretty slow so it probably isn't a big deal to leave the runs beyond third place alone and wait for new submissions to roll in using the new system.

Columns for RTA and no load times is alright but maybe not really necessary as long as the rules are clear about how to remove loads and instructs clearly on how to utilize the autosplitter. There would be no reason to use RTA timing in the future.

CanterlottCanterlott likes this. 
  bill_play3bill_play3

i guess we just need a mod approval here. so we can change the rules when the autosplitter is ready

DarkosszDarkossz and RawricaRawrica like this. 
  TheMasterTheMaster

Thank you, Darkossz!

I've been testing the autosplitter with bill and it is fully functional right now in recording times with and without loads and splitting accurately. The only problem is some jumpy lag while it's running, although it has been minimized significantly. To be honest, I may only notice it because I am on a high hertz monitor and accustomed to everything being very smooth, others running at a standard of 60 might not notice it as much. Regardless, loads can still be removed via checking the recording in the end if somebody didn't want to use the splitter anyway. Hopefully we'll find some way around the lag entirely, though.

I'll start working on building a list of the new times for existing records. Second and third place sounds good? I'm not sure how moderating these leaderboards works but I hope the times are easy to edit or add to new columns, however it's decided the loadless times should be displayed.

I wonder what runs will now be close to a sub-whatever time, I might find some campaigns I will want to improve now...

 
  DarkosszDarkossz

Well for the loadless time i'm working on it but i need to know when the timer stop in loading, and if you can send me the autosplitter that would be great

 
  bill_play3bill_play3

the autosplitter has been uploaded to the resources page. you can record your game running and teleport to saferooms, activate finales and load new campaigns and check frame by frame on the recording to see exactly when the splitter pauses the timer. remember to compare your time to Game Time on LiveSplit

 
  bill_play3bill_play3

and by the way, if you think that some info should be added to the splitter's readme, just post it here and i'll update it

 
  TheMasterTheMaster

Ah, you're on the times, alright, thank you again. Figured I'd offer initially just in case the work was reason to not update the leaderboards.

Oh, and while we're on the subject of updating things on the leaderboards, can you remove Freekill's 6:20 on The Sacrifice Solo? In the description of a run after that which is now obsolete he actually asked for its removal as he accidentally left the platform as PS3 so it's stuck with no placement now.

Speaking of which, is it possible to change the ordering of things in the dropdown menu? The platform should probably default to PC since that is what the vast majority of runs are going to be so it's harder to slip up in the future. Maybe also default the version to Newest since that's probably what new runners would be playing on?

 
  DarkosszDarkossz

Sacrifice leaderboards is fix

Actually i can't change the platform order dunno why they don't make an option for changing the platform

Maybe we will just retire console because i don't see a lot of person running this game on console so

 
  TheMasterTheMaster

Yeah, back in the day there was a reason since it wasn't known yet if anybody would run it on console, but it seems pretty clear that will never happen, especially when you consider it isn't possible for consoles to compete anymore. Perhaps the platform column should just be removed entirely?

I've always been interested in seeing a difficulty column that reflects the lowest difficulty played during the run (so no Easy voting if you want to submit Expert), so if anybody wants to submit Expert runs you can sort by difficulty to essentially create Expert leaderboards without the clutter of entirely new categories. I was thinking this after the complaints of the game not being played on Expert in AGDQ, which made quite a mess of its YouTube comments. Maybe if platform is removed that will make room for this? I'll plop the idea down here and leave it up for discussion.

 
  TheMasterTheMaster

I got a deathless Main Campaigns Solo!

Working on its loadless times before I submit it as I wasn't using the autosplitter for the previously mentioned slight lagginess. The method I am using is to measure by the first frame any element of the HUD appears to the last frame that doesn't have the loading symbol in the corner, since the first frame it shows up is the start of the load. For each level beginning I start at any HUD element I see appear, sometimes it isn't the entire HUD at once, usually the weapon/crosshair/item bar will appear a frame before the character portraits and stuff, so in that case I go for the weapon and so on appearing. Seems consistent. For campaign transitions I end the finale on the final frame before the next poster appears, and start the next campaign on the first frame where it is a black screen after the poster disappears.

I may do the rest of my times as well as I realized I was having a desync issue when I was encoding them before upload (my local source is 30k bitrate so I encode to 12k!) meaning the YouTube versions may be slightly off in time. Twitch versions would be fine, though. Regardless, I have the unencoded originals locally that I can work from to make it easier. This new run doesn't seem to have a desync issue and I've since updated GeForce Experience and the NVENC encoder with it so it must have been some framerate issue that was fixed? I don't have to pull out the audio to fix its duration to keep sync anymore so that's nice.

 
  TheMasterTheMaster

Working on my own times right now, I will have them all done tonight and ready. Can you add the load and loadless columns to the rest of the IL's so I can edit the runs to insert everything?

Also, No Mercy seems to be an odd exception where the HUD elements appear before you actually gain control. Simply using the first frame where there is first-person view and a gun for that one.

 
  DarkosszDarkossz

IL Leaderboards is fix you have now the load and loadless columns

Just have a question for all of you what timer should be by default time with load or loadless?

 
  bill_play3bill_play3

i think the loadless should be the default one, because that time is the determinant one, right? unless i got the "default" concept wrong

 
  TheMasterTheMaster

Yes, loadless should be default since it is the new, more accurate standard. If it is default, what will happen to runs that don't have the loadless column filled in? Does it fall back on the time with loads to sort those instead? That's the only hitch I could think of, if something strange like slow loadless times get sorted ahead of old, faster runs that just don't have loadless filled in.

Also, I forgot about co-op rules, which should be the same thing with the load time removed for the fastest-loading player (which would be the host, like in my case). If it were removed until the slowest loader is in that would allow faster players to make some distance for free so the fastest load is the only way to go here. This also will require the fastest loader be sure they have a video of their perspective to upload which shouldn't be a problem.

Edit: Duh. Took me a second to realize if a blank loadless time was a problem the load time could just be mirrored there to fix it well enough for the slow runs. Can't believe it took that long to occur to me.

 
  TheMasterTheMaster

Alrighty, all of the first-place solo and co-op IL's are done and resubmitted. Still to do are the extra co-op runs with other players (like the Dead Center repeats), Main Campaigns Co-op, and both All Campaigns before I'm done with my own times. Those may roll in a bit slower, particularly All Campaigns since it's a lot to do.

 
  TheMasterTheMaster

Almost done with mine, just the All Campaigns Co-op to go. There are also a couple I need to tell deagle to update (Swamp Fever 10:00, Death Toll 11:29), and those were interesting... I don't have my perspective recorded anywhere for those anymore and it's impossible to tell from his view how long my load is. Luckily, both of the runs were recorded on the same respective days/sessions that the WR's were so what I did was take my total load time from the WR's and mirrored it to those two runs. It's not going to be pinpoint accurate, but it should be extremely close since it was done in the same sessions.

Also, I feel that, assuming it doesn't mess up the sorting on the leaderboards, the times with loads of every run should be placed into the time with loads category with the time without loads left blank since it is implying those are all the loadless times, making the times look slower than they really are.