Death on Stairs = Run is Invalid? Discuss.

[aulos]

Okay so, first i wanted the split to be when the instruments appear because i thought the fade would not have been enough accurate. Though I just tested it and I changed my mind. The fade one seems to be good.

Just a little question to clarify the thing, the split would be when the screen is completely white (~when the music changes), or at the beginning of the fade ? I guess it's the first choice but I want to be sure instead of remaining in ignorance !

[giuocob]

It has to be at or past the point that Link and Dethyl unload. This may happen as soon as the screen is fully white, but we may have to do some testing to make sure.

[Flynn]

Well, unless I'm missing something with GT weighing in that covers all the 3 heart runners that this change would actually affect.

[mzxrules]

Eh, changing the end time point seems unnecessary. If you utilize a glitch to warp to the credits, but it had the possibility to softlock a few seconds into the cutscene, you wouldn't move the end time point back past the start of the credits.

[mabdulra]

Eh, changing the end time point seems unnecessary. If you utilize a glitch to warp to the credits, but it had the possibility to softlock a few seconds into the cutscene, you wouldn't move the end time point back past the start of the credits.

Actually, this happened in Oracle of Ages. If you were to do Veran Skip, the end time actually is later than if you were to kill Veran directly, since there is the possibility of softlocking the game if you do Veran Skip incorrectly. This was done in order to ensure that when timing ends you are guaranteed to reach credits.

It looks like these are our three possibilities:
1. Change nothing. Timing ends at stairs regardless of if you die or not after.
2. Slightly modify timing so that it ends when screen fades to white as you transition.
3. Greatly modify timing so that it ends when Link holds the instruments above his head.

Then there's a question of what to do for past runs:
1. No runs modified.
2.a. All old runs that died to Dethyl will be invalidated. Retiming will not occur.
2.b. All old runs that died to Dethyl will not be invalidated, but all future runs will be. Retiming will not occur.
3.a. All old runs that died to Dethyl will be invalidated. Retiming will occur, and runs that do not reach the Instruments screen will be invalidated.
3.b. All old runs that died to Dethyl will be invalidated. Retiming will occur, and an "average" time will be added to every run to emulate as if they reached the Instruments screen in the event that they didn't.

I am a fan of either 2.a. or 2.b. If everybody is in agreement for that, the question is whether or not old runs should be invalidated, or if we as a community will overlook that and only consider new runs to be affected by this ruleset change. If there are options I've missed, please list them and add them.

[ZorlaxSeven]

I guess I need to offer my two cents.

There is no way that dying on the stairs should still count as a "valid" run. The end-timing of "when Link touches the stairs" was a carryover from times where one could not die after that point. Adding an "average" amount of time is also fairly strange notion, as we can see by our precedent with OoB in Tal Tal Heights. We did add 5s to those runs that had it, rather we just left them the way they were.

While I would like the JRTA timing, it would allow an unfair advantage to those who were actually paying attention during the final cutscene. [I almost never do, since I usually am excited by that point and am ready to acknowledge the run.] Timing at the screen fading to white is probably the best way to go, since it would not force us to add an "average" amount of time, we can simply retime all of the runs (or add however many frames it takes).

For me, though, the biggest issue that we need to address, and have needed to address for a while, is the exact frame when timing ends. In my redacted 1:09:59.87 run, I had a lot of trouble getting a definitive answer on what frame exactly timing begins and ends. While that may seem like a simple notion (timing begins on file select, ends when Link touches the stairs), locating the exact frame was such a hassle, and I ended up making my own definition, which I thought was accurate. In other words, I'd like to see screenshots of exactly what frame is beginning and end (as well as a few frames before and after each section). Whatever we DO decide on, I think this is the most important part (especially for the Any% runners, where frames actually matter).

I hope we can come to a timely, definitive decision on this issue in the coming days. Thank you.

EDIT:

We also should have all future submissions reach the "The End" screen, incase any other timing issues come into play in the coming years.

[mabdulra]

I did some memory snooping.

There is a particular byte that indicates the "scene" you are in. This includes overworld, so not cutscene. If that value were to change to a particular byte, the final cutscene will trigger. This occurs whenever you change the byte, i.e. I can change it using memory editing in Marin's house and immediately trigger the final cutscene. To me, this is the point where the game acknowledges that you have beaten it. If everybody is okay with that, I will go ahead and set a breakpoint for when that value changes in the final fight. I will get a screenshot or even a video for when it happens so that everybody can see when timing ends. There will be no guesswork involved.

How does that sound to everybody?

[mabdulra]

Deln's account isn't activated, so he asked me to voice his opinion:

1. if you make it so all previous runs are invalid, better remove all of mine
2. timing on fade out
3. f*** any%

[mzxrules]

Deln's account isn't activated

Yes it is.

[mabdulra]

Yes it is.

Well, I guess Deln then really wanted me to post on his behalf.

[Rapid_]

I did some memory snooping.

There is a particular byte that indicates the "scene" you are in. This includes overworld, so not cutscene. If that value were to change to a particular byte, the final cutscene will trigger. This occurs whenever you change the byte, i.e. I can change it using memory editing in Marin's house and immediately trigger the final cutscene. To me, this is the point where the game acknowledges that you have beaten it. If everybody is okay with that, I will go ahead and set a breakpoint for when that value changes in the final fight. I will get a screenshot or even a video for when it happens so that everybody can see when timing ends. There will be no guesswork involved.

How does that sound to everybody?

I agree with this notion. However, may I ask what address this byte is at?

[Riddler]

I agree with mabs options 2a and 2b. I dont mind retiming or any of that and ive never died on the stairs once  but i think that time should stop on the frame where the screen is white.

[LoenP]

Like I said, I like on instruments the most in theory but for practical reasons I think on white is our best option here. I also feel retimings wouldn't be hard since it should remain a static figure, and even if retimings were done by hand, wouldn't take much time as it's only a 2-4 seconds difference.

Whether or not to invalidate prior runs is a tough issue for me to call, since I feel that for all categories that aren't Any% or AI, this is almost a non issue and the one or two runs I can recall that did have a dethl stairs death weren't exactly top times and I really don't feel that letting that particular runner keeping their PB wouldn't compromise the leaderboards in any way. However it feels weird and wrong to invalidate runs of one category and not another, so there's that.

Although I don't feel the end timing is (or has been) much of an issue other than any%, having an exact 'end point' has no disadvantages and you should fully go aheaad with that plan Mab

[giuocob]

Basing the timing off of a hidden flag that switches scenes isnt going to be practical - it needs to be a discrete visual cue. I'm in favor of the screen fading to white. It can't invalidate older runs due to people not mashing through the credits, and should be measurable to within a couple frames if necessary.

[mabdulra]

Basing the timing off of a hidden flag that switches scenes isnt going to be practical - it needs to be a discrete visual cue. I'm in favor of the screen fading to white. It can't invalidate older runs due to people not mashing through the credits, and should be measurable to within a couple frames if necessary.

Yeah, when I looked at the address $DB95 it seems that the fadeout is caused because the byte flips, which isn't really a good visual cue. I think I agree with timing to end right when screen fades to white, and I think all of us have come to that conclusion now.

So, let's recap, because there is going to be a weekly race tonight most likely:
* Timing does not end on stairs, but when screen is fully white after the fadeout into the final cutscene.
* Old runs will not be retimed because this is only a 1-2s time advantage, except in the case of the short Any%, which will be retimed.
* All old runs that died on stairs will be removed/invalidated due to the fact that they have not completed the game.

If there is a point of contention among any of these three, please speak your dues at your earliest convenience. Thanks!