Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
upthorn wrote:
So you're certain that none of the possible input combinations that you didn't test could have performed any helpful glitch or trick that you don't know about yet?
Of course not to 100% (this is impossible), but very likely to more than 95%, yes.
Oddity wrote:
You may even discover super great things, but once it's published, people will find improvements. Very unlikely? 100% likely. I think that's where people are getting confused by your speeches.
Well, it can be likely, but chances for that are very low at least.
Fabian wrote:
Most of his comments are just unreal to me.
Thats good. That way you will be even more surprised to see the unexpected when the day of submission comes.
Seriously though, do you really believe I would waste my time writing something that isn't true at all? What would be the point of that? I know it's hard to believe, but all my statements are real, and I will prove it to you sooner or later.
Though if like Fabian, you want to increase your surprise,
keep doubting guys! (wow what a rhyme) :-)
evilchen:
Please try it yourself. I really don't want to spoil anything about it yet, because I'm curious what time one can get by his own without influence (which is how I did it as well).
Cpadolf:
Around half of the time that is required to beat the game.
Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
upthorn wrote:
Fact: it is impossible that you would have tested all possible input combinations for the given number of frames, as doing so would take millions of years, even for a robot.
Fact: With the needed knowledge you can see immediately which combinations are promising and which are definitely slower, even without testing them. I only need to test very few strategies per room that are all similarly fast, to find out which one is still the slightly fastest in the end. It's hard to believe, but it's really like that for me.
Maximus wrote:
Why can't you just post a WIP so everyone can see what strategies you're using?
Because it would be a huge spoiler of my future run. I definitely think it will be for the better to keep it private until submission.
hero of the day wrote:
The only problem with your lame claims of perfection is that we will never see them. You started working on your run almost 2 years ago, and guess what.... you have to start over anyway! Thanks to a discovery by moozooh, and a discovery by myself. At this point in time it is just pointless for you to make a run. Cpadolf is using the exact same route as you, so almost all of your tricks have been spoiled anyway. You hope to beat his time by what... 10 seconds? The shock factor is gone at this point, and no one gives a shit about the escape timers other than hardcore in-game oriented speed runnners. The casual viewer here does not care.
1. You will see everything I claimed. It will just take some time to get the runs done.
2. Using the same route doesn't automatically mean that all my tricks get spoiled.
3. I think most people want to see how to get a fastest possible escape timer, not just hardcore speedrunners.
hero of the day wrote:
And I am 100% certain there are some new discoveries I have made that you do not know about.
If you wish to be proven wrong, we could make a blind test by comparing our frame times in the specific area without mentioning the strategies at all. But once again, it would be a risk for me to spoil something even by that, so I'm not sure if it would be worth.
Cpadolf wrote:
And what was your discovery, The red brinstar shaft shot?
Yes. The only true reason I cancelled my WIP (this was before Torizo skip, which of course made the cancel even more obvious). I didn't even restart it since then, as I worked on the Redesign TAS from that time on. This shot discovery made me test things in a even wider range than I did before, so it's now very unlikely I would miss such a similar thing again. I have to thank hero for that.
hero of the day wrote:
On top of that I also use a strategy that was 10 frames faster than his in the 2nd red brinstar shaft.
I realized that myself when I posted that demo of the 2nd shaft pass. The PB placement was too early. Your currently published run skips the important Super Missile refill though. But you probably know that even with the refill it's still possible to save this 10 frames there.
hero of the day wrote:
In case you were wondering 90% of my new techniques involve the CWJ.
Hah, this was so obvious! ;-)
Though respect about planning all of them ahead that exactly.
Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
hero of the day wrote:
There are strategies that I know which you and Saturn don't know about, there are strategies you know which Saturn and I don't know, and there are strategies that Saturn knows which you and I don't know about. Who ever submits their run last has a natural advantage. Comparing TASing abilities is just pointless at this time. I really see no difference in the quality of work among almost all of us.
I wouldn't bet on that. Knowledge is a very important factor here. Although I will sound bad again, I now passed every situation in such a run, and I know all the fastest possible strategies in each room. Sure there might be strats you use which I won't, but then there will most likely be slower. Too bad I can't show you my any% WIP so that you can see yourself what fast strategies are, and learn a ton of new ones as well, but it's just not finished, and a unfinished run won't be submitted, obviously.
hero of the day wrote:
Also, obsoletion based on more precision is a bad idea. It would be impossible to know which run is more optimized, since they would be collecting different items and using different routes.
It's pretty easy, actually. The in the end faster run is more optimized. Which route you take, or what items you refill is your problem, and it's irrelevant for judging the final completion time. Also you can see a perfect example of true precision very clearly in the Ceres and Zebes escapes from the timer. This should also be considered when judging a run, since it's a time every watcher can see, unlike the unnoticeable 1-2 realtime frames between door transitions.
Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
hero of the day wrote:
I know your run will have good flow because you did not use excessive pause menus, but a future runner may enter the pause menu a thousand times just to beat your in-game time. With a solid goal of realtime there is no guess work, it is plain and simple "as fast as possible, no matter what".
Good point. Though I doubt that a 50 in-game frames faster, but 2 minutes slower realtime run would obsolete the non-menu screen abused in-game time run. It would suck badly in entertainment, and it wouldn't change anything on the in-game clock at all, so I don't see this as a problem to worry about.
Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
The in-game oriented any% run has some nice advantages to the realtime based one:
It has more variety because it shows more new techs Samus gets from the items, and thus the run gets more entertaining to watch. 2 bosses get significantly faster to kill (less boring standing and shooting at one point), and instead you show more cool techs when passing additional rooms. Thinking that it also saves another minute on the game clock, are 2 major enough reasons for me to aim for that kind of run, not to mention that it also allows one to completely skip the menu entrance to switch off some items, making it a 100% uninterrupted run.
And hero, murder beam is no longer a problem in the in-game oriented run, because we skip Spazer anyway, which saves more than enough time to avoid it, and instead makes a much better and shorter in realtime fight to watch. Also it's obvious that it's not worth to enter the menu to save 10 frames by sacrificing 400, for like gravity jumps and other minor stuff.
On the other hand I also admit that it's cool to see one passing all the difficult areas with a only minorly equipped Samus. This is very difficult, and thus the entertainment factor surely is high there because of this, also because of a overall shorter route with less backtracking and the fastest real-time.
In the end I think it's really the best to not compare this 2 runs at all, as they both have their fair amount of advantages. It will always depend on a persons opinion which run it will prefer.
Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
Cpadolf wrote:
On an unrelated note I found a silly thing today, eye doors can't be killed by to high speed projectiles, they blow up inside them... Although i guess that this is the same thing that allows the green gate glitch.
Correct. It skips the damage block (just like it does when passing the gate block), and crashes into the solid one behind it. This can happen when shooting at normal blue doors as well, resulting in not opening them.
Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
I just wanted to inform moozooh to not expect too much of his attempt and thus not get disappointed when he won't be able to beat (note that I didn't say match) my times (which he surely expected). I didn't mean he will be necessarily slower, it was just a assumption. It wasn't meant bad at all, really.
ZeXr0 wrote:
Reading this topic reminds me of a certain person who thought everything was perfect.
I didn't say "everything". I said my Ceres part (other than the very first room) is perfect. It's a fact.
Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
evilchen wrote:
i doubt he will finish any run (ergo no smvs from him :P) *smiles at saturn :D :D :D *cough rbo cough*
You are wrong. I will finish EVERY run I started, trust me.
It will just take some time.
Tonski wrote:
is it 110% sure that 49"15' can't be beaten? I mean by delaying frames in the opening in more affective places or stuff like that.
Yep, 120% if you want. I'm very positive on that. ;-)
evilchen wrote:
breaking news with that smv i was able to open the last door 4 frames earlier then jxq - without optimizing it very well (yet) maybe that helps to archive a very good time now!
Wow, that looks very promising. Great job you do so far, evilchen, respect!
EDIT:
And to everybody who still misunderstands me, I never said I'm superior to anybody. I just said that I got a perfect time. Speaking the truth is hopefully still allowed, or isn't it?
Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
Awesome job, inichi!
I'm highly impressed how good you managed to manipulate the criticals and the enemies position to always be in reach of the Fire Whirl tech.
I very much agree with hero, you should really try to do the any% TAS, since you are the person with the most speedrun knowledge in this game. It will also be a very interesting challenge to see if you can beat your SDA speedrun even by using the US version (which according to hero should be extremely hard if not impossible).
Regarding the Mom skip, this is extremely close. I really hope somebody can manage to do it, as it would save much time in any run. I too failed in it after trying, so let's see if hero can manage it.
Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
Cpadolf wrote:
Also you have made your fair share of "that's impossible" statements.
If you mean the Maridia Gate, thats true. Whats wrong with it?
Cpadolf wrote:
I don't know how you meant to say this either but to me it clearly sounds like you are saying that moozooh will be slower.
Just assuming, because chances are not very low for that. I would be glad to be proven otherwise of course.
I see there is a slight mistake in that statement, so let me fix it:
Saturn wrote:
And don't have too much hope on the escape either. You won't beat my times there, because they are perfect as well. I'm actually glad you work on it by now, so hopefully we will see your (probably slower but without seeing my version still impressive) result soon. :-)
Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
Cpadolf wrote:
And how can you know for a fact that moozooh will not get as good times as you?
I didn't say moozooh will not get as good times as me. If he does it perfect, he sure will (though chances are not very high for that). I just meant he won't be able to beat them since a perfect time can't be beaten. Thats all.
Cpadolf wrote:
Also I should say for your own best to avoid major dislike from many that if you have information that you and only you have but are not willing to tell then for god's sake don't say you have such information, it makes you sound like "I'm better cus I know stuff you don't"
This is only my opinion though.
Hero actually asked me for this information, so I gave him because I'm a nice guy. I didn't mean to cause troubles with it.
Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
moozooh wrote:
Hey, I haven't even tried the escape yet. So far I'm only in the second room. And you forgot those 3 frames you have lost in the first pre-ridley room. ;)
I actually meant the first rooms until Ridley.
And you are terribly wrong about my frame loss. I thought I explained it to you on IRC, but here you go again: I was 3 frames slower, because I missed the elevator hop trick at the "very" beginning in the first room that saves this 3 frames. All other, later rooms were frame-perfect and thus same fast. As you know, you all got 801 until Ridleys room, I got 804 because of this.
moozooh wrote:
Also, it feels like you're comparing apples and oranges, since we weren't able to monitor randomness via RAM watch, either.
Nope, you monitor pixels and subpixels to optimize every door, where I did it just by routine. It sure gives you big advantages to save some frames at them.
And don't have too much hope on the escape either. You won't beat my times there, because they are perfect as well. I'm actually glad you work on it by now, so hopefully we will see your (slower but without seeing my version still impressive) result soon. :-)
Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
Upthorn, I actually use RAM watch, but only through Snes9x, since gochas memory watcher tool doesn't work for me. Because of that I only use the most important adresses to not spam the screen unnecessarily, such as in-game timer (this is always on, and it allows me to indicate lag perfectly by just comparing it to real-time), and when I'm at bosses of course boss-HP. I don't feel limited at all that way, in fact having too many adresses to follow would even be disturbing when optimizing a run because it would distract me from concentrating on the most important things.
I'm pretty sure I still optimized my Ceres rooms to the max, because when even JXQ, hero and especially moozooh (who is known for extremely crazy frame optimization) didn't manage to beat them "with" using the memory watcher for all possible stuff, then this says much, I think.
Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
moozooh wrote:
Hmm? Weren't you asking me how to optimize the subpixel positions (and what did they mean at all) a couple months back?
Yeah, I did out of curiousity, because I don't need to use the memory watcher for pixels and subpixels in order to make a perfect run. I just did it the conventional way, by comparing to the in-game time between doors only. I obviously optimized every subpixel to the max, but just without watching it in the memory watcher.
About the countdown-timer changes, Cpadolf is right. Although it can vary up till 0,02 sec, it's not manipulable other than that. It will always go by 1 sec after every 60 frames, and this is good, otherwise you could just manipulate and cheat the timer up without any true frame gains.
And no, moozooh, I didn't determine any patterns about the timer. It's random.
hero of the day wrote:
Clearly Saturn found a key to obtaining perfect steam with a very small delay, but without knowing the 2 different spots he delayed (the exact amount of frames per spot, is what I mean), it is impossible to duplicate.
Yeah, TASing isn't just about dublicating a existing run. Show your own work. :-)
Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
hero of the day wrote:
Mind telling me in detail how you improved JXQ's near perfect ceres escape by 7-9 frames? I don't expect you to answer, but I think that being open with the community will be more beneficial to everyone including yourself.
Nope, it wouldn't be of any benefit for me at least. But anyway, here is how I did it in detail: I optimized the subpixel position for every door, and I manipulated luck for perfect steam by delaying some frames in the intro scene and at Ridley. The combination of this was the key. There is nothing more about it.
Regarding RBO route, I'm pretty sure getting Grapple Beam will be slower in the end. Also you get the gauntlet tank without the speedbooster, which will lose many frames there. Otherwise it looks good.
Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
Hero, you seem to have misseed this:
Saturn wrote:
Well, I think your last smv was pretty optimized already. I honestly can't see where you want to save 12 more frames from it. Are you sure you didn't confuse something?
Mind telling a bit more detailed how you want to save the mentioned 12 frames from a near perfect attempt already? I wonder because it's very unlikely in my view.
Regarding Cpadolfs situation, it's a hard decision. Pipe bugs cost some time as well. Maybe a 2nd PB combo would indeed be slightly faster in the end, but it's just a too high risk to go for it, since it will be very difficult, if not impossible, to fully hit him with the 2nd combo, as said. I personally wouldn't go for it, to be on the safer side.
Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
Well, I think your last smv was pretty optimized already. I honestly can't see where you want to save 12 more frames from it. Are you sure you didn't confuse something?
Regarding Botwoon, no, another PB combo will definitely lose more than 70 frames compared to Super Missile spamming. Also because of Botwoons very small hitbox (90% of his body is invulnerable), it's very unlikely you can get all 4 hits of the combo in without more delays when he is already fully out of his hole and moving around. The beginning of the fight is a exception, because of obvious reasons.
Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
Hero, your steam boost is not faster than a usual jump. See yourself:
http://dehacked.2y.net/microstorage.php/info/769593669/cerestest_withoutDMGboost.smv
The door starts to open at 27'13 which is exactly the same time where it did with the steam boost variant. I used your smv, so the conditions are 100% same. Hope it's obvious now.
Regarding alternate HJ-tech, yes I see now that it's 15 frames faster than JXQs (it's still not 30 ;-) ). My version was 10 frames slower because I just didn't optimize it to the fullest. I couldn't allow to skip the Super Missile refill in my conditions, so I just did it sloppy to roughly see how much it could potentially save. Out of curiousity.
Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
Regarding the steam boost, nice find, but I just tested it, and with perfect jumping like we usually did, it's exactly same fast, not faster. Just try it without the boost and compare the time from the last jump till the door to see.
hero of the day wrote:
hey Cpadolf, maybe this will be of some use to you. I just developed it :)
here you go
Oddly enough this technique only saves about 30 frames, after taking into account the shudder walking part. Also I think taking the bottom route might be slower than the top route after collecting the boots, but at least now you have a choice in what route to take.
Old stuff. I did test this when I was at this place in my any% WIP, and it doesn't even save 30 frames (I wonder where you got this value). If you fully take the shudder walk into account (that is, from the very first frame you start doing it), you lose so much time there that in the end it's only barely faster. I compared this alternative tech back then, and it was less than 5 frames faster than the usual method I and JXQ do. We get a very important Super Missile that way though, which will definitely save more than 5 frames because of less refills later. Alternatively you could do it the way I did in my 1st RBO demo, and this would most likely save the last 5 frames, but as said, the missing Super Missile refill wouldn't be worth it.
Oh and I confirm, the bottom exit route is even slightly slower than the top one, so it doesn't matter as well.
Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
I did some progress, and am to about 50% done with the section for WIP5. I can't tell where I am though, since the new route will be that surprising to see, that I better should keep it secret for now.
I'm currently very busy with work though, so the progress is very slow. I still hope to get WIP5 done in 1-2 months, and I can promise that it will be the best WIP so far. :-)
Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
Wow, that was a nice shock with the shaking rumor at first.
But no, I NEVER experienced any block position differences in both, horizontal and vertical direction during the screen shaking from the Super Missile (I had plenty of such situations during TASing when opening red or green doors while needing to make a jump onto a platform immediately after it (ex. Spazer room), so it won't help here as well.
Here is my take on heros test, this time with 100% same shot height in both conditions, which proves that shaking does not affect anything regarding hitboxes:
Without shakeWith shake
Kej, you said that such gates would require one to get at least the horizontal speed of when Samus starts getting the blue suit echos (without any speed lowering tricks like the shortcharge), which would require you at least 3 or 4 screens of acceleration room in order to reach. And this speed would even require a perfect shoot-position to the gate in order to work (we talk about the last possible tech - the diagonal shot through the gate). Unfortunately I can't see the 4 screens of acceleration room before this gate...
Hence, the impossibility.
Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
All respects to hero for still trying this, but honestly guys, don't expect too much about the gate. It's a technical and physical impossibility to open it from the wrong side at this conditions. I say it just so that you won't get disappointed that this breakthrough will never happen.
I admit that I didn't test it that much as hero already does, but I just don't feel like wasting my time on something that is impossible anyway (which I realized pretty early during the tests).
Also I have some more news:
moozooh once wrote:
I once again repeat that it isn't about ingame time or loading or anything else like that, it's about differences in timing and behaviour that are clearly observable. If you don't believe me, open my 1.51 Draygon battle and watch the turrets shooting about TWICE as fast as on 1.43. Sounds like not-really-100% correct emulation from one of them, huh? What about Ridley with his rain of fireballs whose degree of intensity on 1.43 isn't even close to that on 1.51?
I just tested it one more time, and as already thought, the differences moozooh mentioned depend entirely on the input. I'm 100% sure about this now. He was just unlucky to trigger them. I played through both bosses in realtime on both versions, testing different inputs to trigger different patterns, and for example I managed to not let Ridley shoot even one fireball during the entire fight on v1.51, and the opposite on v1.43. Same with the turrets at Draygon (whose behavior I actually know very good from SM-Impossible, since you can't kill them there).
So just to not let people stay confused about this matter, all this boss acting differences are just luck based. The only differences between this emulator-versions are the loading and door transition delays. The in-room emulation, including Samus' moves, and all possible enemies in them (bosses are not different than any other enemies in this game, just larger and with more complex move patterns) is 100% same in both versions.
Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
Hmmm, that sounds much better and more realistic, hero. But since the speed and acceleration of a Super Missile shot can't be compared to anything else (it's still much faster than a shinespark), there is no way to get it offscreen, but then catch it up again when it passed the 2 blocks.
If it would be possible, then I'm sure you could open the gate, but unfortunately...