The Elite Forum
The Big Three Plus One => GoldenEye 007 => Topic started by: sebastmarsh on April 04, 2020, 05:05:29 am
-
Hey elites,
I'm not a regular speedrunner — too busy with life/work — but I used to "sort of speedrun" games when I was younger before it became a more organized thing online. We played, like, MGS1 and N64 Goldeneye a lot, just with friends offline back in the day.
I found this scene online a couple years ago, and I think it's really cool. I watch videos from time to time for leisure and take them apart and do theory. I really like the optimization focus, learning, and collaboration. I'm the kind of guy that tracks my personal life time down to the 5 minute block typically and optimizes those type of numbers, and sometimes builds lessons from them into UI/UX stuff out in the world.
Anyway, I share that background cuz one thing I'm really interested is conversion rates between seemingly unrelated types of variables/resources.
In Golden 007, time is the main variable you're optimizing for. But there a lot of other variables. Some of them are true/false and make or break a run (guard opens a gate for you or not). Some of them are single independent variables (location of particular single NPC). And some of them need to combine to get what you want (multiple boosts on Dam Agent, hostage/guard timing on Frigate).
After watching a number of videos recently, I think you can probably come up with an explicit conversion rate between Health and Time. These two videos are probably enough for advanced analysts to start to map it out —
https://www.youtube.com/watch?v=hT0kXWd4jV0
https://www.youtube.com/watch?v=7db-dRm2KSA
Then, any level/difficulty that the current WR didn't end on the lowest amount of health — or that the boosts didn't "convert" to time savings at maximum efficiency — are candidates to be hunted out and optimized.
There's a number of untied WR's right now where someone ended with multiple health bars. Also, not all the boosts looked "maximally clean" - aligning a 1 or 2 damage boosts better, and getting an additional damage boost, likely shaves 1 second off a bunch of those.
It also might point at some other strats like wounding particular guards to get them to flinch/lock up from shooting until you run past them, instead of killing them — in a high-damage level, this increases variance and the chance you'll lose outright, but increases the best possible time. There might be also be some odd maneuvers like shooting up a computer terminal a little bit as you run towards it, and hoping a guard hits it that you've run past to get a correctly aligned perfect boost from it.
Basically, I think you can view your life bar as a resource that ideally always converts to the maximum possible time savings — which will vary based on weapons and explodables in that level — and which can be played around with in hunting for records.
Then, you might need to manipulate guard positioning via wounding them or doing unusual things in order to get "straight line best HTTC."
On some of the records, this looks like it's already maximized — but coming up with explicit conversion rates of best possible clean boosts per level given weapons/explodables, and playing around with guard positioning as well as one's own movement and shooting... seems like there's probably some simple notation you could write and theorize around, and go hunt out the theoretically best HTTC in each level.
I think there's a bunch of records possible with this. Thanks for all the entertainment and learning and camaraderie and cool stuff coming out of here, huge respect,
Sebastian
-
Hey and welcome Sebastian.
There aren't too many levels where you can feasibly lose all of your health while running with WR strats, so for most levels the kind of analysis you are talking about doesn't make much sense.
At a glance levels were you can use all of your health for boosts are probably..
Runway SA
S1 SA/00A
Silo 00A maybe?
Streets A .. dunno about 00A
Jungle 00a?
Cavs 00a
Honorable mentions:
Dam 00a, though mostly this is just survival rather than get boosts as I understand it.
Fac SA/00A, though they are mine boosts so if you do them well then health isn't an issue afaik
Even some of these are a bit perverse, there's a lot more effecting your time than just boost ratio. E.g. aztec 00a gets the BA just to stay alive.
The general attitude is just to completely forget health.. and not play the levels like Dam 00a where health is an issue! :nesquik:
-
I'm pretty sure the amount of health remaining, if relevant, is pretty optimized right now.
If it were possible to use health to gain boosts, so to speak, it would have been done. Sure there are some levels where you end with lots of health, but since it's not possible (or extremely unlikely) to convert that health into boosts, it's irrelevant.
-
Jungle SA/00 could be 0:50/0:51 respectively with perfect runs/boosts but I bet you it lowers to 0:51/0:52 and stays there a long time. All you are suggesting is we grind for luckier runs. Most jungle wr pace runs that survive somehow do so with a boost or two or even three of health remaining to take. It's just the way it works. Yes we can grind a perfect run.... But we haven't even got 1s off max yet so the point is moot. Getting 3 decent guard boosts and no backs is great luck. Getting 5 literal perfect guard boosts, doing a self boost losing 0 health, no backboosts and an otherwise perfect run is one in a billion on that level. It's the same as saying the hostages can complete faster on frigate SA/00 than current wr, and suggesting we just go faster and play till it does complete, or saying why don't we just buy lottery tickets because we could win millions of dollars if we win lol. Anyway no harm no foul I'm just pointing out why this isn't relevant. We know of the boosts and we know of the general odds. It's moreso the stacking of odds of perfect runs plus perfect boosts. The game is only getting more optimized, you'll see more records broken in the coming years from doing just as you suggested, but the same would have happened had you not suggested it either. There's only a few ways to gain time in these speedruns, better movement, better boosts, better strats, once all the movement and strats are completely optimized all that will be left is boosts. Something I'm sure the elite has been aware of since 2000 at least.
-
^ Fully agree with above 2 posts. No disrespect to the newcomer as he's just looking at this intelligently trying to help but yeah it's just not feasible.
You would basically be turning every level into Dam A 0:52 except much harder because you have to worry about dieing, more technically difficult levels and so on.
These runs can always happen by a fluke of course, but to actually try for them in a 100-500+ hour grind is pretty insane. You would pretty much be trying to be a human TAS as far as manipulating every last bit of luck....not fun.
-
Fair enough that a lot of those records would be grinding-heavy and might be unpleasant to get — nevertheless, they're possible and putting down some explicit notation on the equations and optimizing where exactly to take damage for max speed gains might be a useful tool.
But I think there's some strats that maybe haven't been explored around this that might be viable.
Take Surface 1 Agent for instance —
https://rankings.the-elite.net/goldeneye/stage/surface1
Current world record is 0 damage.
Not sure how close the current 1:02 is to 1:01 but I wonder if there's a way to interact with that first guard to not kill him and then take 2-3 damage boosts en route to the first objective, and then potentially take the guard out on the way the building for the KF7.
Or — not sure if he drops a grenade on agent — but if so, possibly get the grenade drop and juggle the single grenade for a damage boost.
There might be a possible routing like — shoot that guard x1 targeting a certain wounding animation while running past him, get shot from behind for 1-3 damage boosts en route to Objective 1, shoot him on the way out of the building for the KF7 and a grenade, juggle the grenade x1 en route to the exit.
Maybe it's not viable, for any variety of reasons.
But that type of analysis seems like it might uncover some possibilities.
-
This can also be contrasted against Surface 1 Secret and 00 Agent records that do effectively convert most of their health into speed.
And yeah, this is RNG-heavier and of course levels or strats with more RNG are less pleasant/fun than skill-based levels/strats. That's just... true, of course. RNG is one of the less fun things about speedrunning.
-
Unfortunately there's a lot of catching up you gotta do before your strats could even have a chance making it to being tested on console. We already know that you can't get a grenade on s1 agent in a speedrun. And 1:01 is already close enough to where you won't need a boost to achieve it or any of the possibilities you've mentioned, Calle is really close with current meta already. I could say to try testing your strats yourself first but unfortunately when playing unoptimally you can get boosts and grenades you wouldn't get on perfect wr paced runs so it doesn't always work. I welcome your ideas but you just have to realize everyone thinks of this stuff, and after playing the levels for hundreds of hours we gather the info of where the best boosts are etc. It's just about grinding everything else about the level plus the perfect boosts. Dam 52 is a great example said above. You can in theory say what time the best boosts would give on a run, but grinding any complicated level is hell enough, and plus it's not like we are asking the guards to not boost us ;)
-
Yes, I was going to chime in and say that basically, on Streets Agent, Surface 1 SA/00A, and in some regard, Facility; do use your health as a resource.
Streets Agent is the most prominent of these for sure. I remember for a while, Illu was going for 1:11 (?) on PAL because it was easier to survive all the boosts on PAL for whatever reason.
btw, regarding Surface 1 Agent, you should watch my video about it: https://youtu.be/KMGrAFHiRvY
It goes over all the reasons why you can't get a nade, why getting a boost is rare, and basically everything you seem to be thinking about on the stage. Watch it and then chime in with any novel thoughts about a pathway for 1:01. I'm curious to see how this changes your perception of things :cheesy:
-
Great video RWG. Didn't think about potential locks through trees — fascinating.
It also answers the grenade question as probably impossible, unless you could do some craziness to get the key guard out.
But I keep coming back to this idea —
https://youtu.be/dH5RQUO-g9k?t=29
Look at how straight the movement is relative to the guard after taking him out.
Imagine shooting him x1 in the leg or arm while passing and leaving him alive.
It just looks to me like there might be a couple boosts there. You'd have to either get the KF7 later or do locks with sniper of course, but it just looks to me that there might be two good boosts worth .4 to .6 seconds in there.
Of course I could be wrong -- just trying to theorize a bit. Incidentally, it was your video breakdown of Dam and Karl's of WAR! in Perfect Dark that led to this type of thinking. Before Rayan, people were neglecting a little extra trading off of life for speed, which made the difference in the record.
-
In general I think it's safe to say that all 60 stages on GE have been fully scouted out for possible boosts at this point (without breaking the optimal lines enough to negate the time save), and any remaining health in runs is not something desired but just a byproduct of how actual runs play out. Your idea makes sense, but in reality it's already implemented as much as possible on every stage; the ones where it isn't is simply because the missing boosts (which are already known to be possible) just don't happen often enough on WR pace when combined with everything else, not because of a lack of trying for all remotely possible boosts. Adiditionally, it seems to me that you're overestimating the time save from boosts in comparison to the time save possible from simply better lines. It is rarely ever the case that going out of your way or really doing anything at all other than firing (on the line you already optimally take) to lure a boost is worth it in GE. I'd have to go through to check, but I can't even think of any cases where such a thing is done as-is. A tiny deviation from an optimal line almost always costs .1-.2 and a deviation significant enough to get a boost that would otherwise not be possible usually costs more than that, fully negating the time save from a boost and then some. That's also not taking into account the fact that active guards significantly reduce frame rate in many cases, which can cost tenths here and there. On some stages, the guards stay active for quite a while causing even worse lag (Train in particular, where a higher frame rate results in much faster locks).
Guards are also really, really inaccurate. The thought of one guard boosting you twice on S1 Agent is completely out of the question since they can barely even complete a firing animation and then start another before you are way too far away to get hit.
TL;DR: Yes, you have roughly the right idea, but people already do this and anywhere where they don't isn't due to a lack of thinking about this but mostly due to WR path/line routing and other factors such as lag.
-
But I keep coming back to this idea —
https://youtu.be/dH5RQUO-g9k?t=29
Look at how straight the movement is relative to the guard after taking him out.
Imagine shooting him x1 in the leg or arm while passing and leaving him alive.
It just looks to me like there might be a couple boosts there. You'd have to either get the KF7 later or do locks with sniper of course, but it just looks to me that there might be two good boosts worth .4 to .6 seconds in there.
Echoing what Kev said about the guards on agent being exceptionally innacurate - also, where else would you get a KF7 without losing a bunch of time?
GE has been rather extensively picked over at this point
-
Boost lures are used in caverns agent, and it's very possible there are more levels and places where we could benefit from them yet. Research would be worth it if it results in even only 1 more lure. Perhaps these very lures will become commonly used in SA/00 in the future.