Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
Nice job, evilchen.
In total you managed to save 432 frames compared to my 100% run when entering Fire Eye's village.
This was of course far too much to be a realistic improvement to my run, so I went ahead and compared both runs more carefully, trying to find a reason for this. And I did:
I actually did a horrible mistake at this point by not triggering this dialogue to start appearing much faster by simply pressing the action button 2 (not 1 time like I did in that case) times before it. This first few dialogues in the beginning were the only case where I didn't use turbo/autofire in the entire TAS to fast-forward them, because I at first planned to do that manually by pressing the action button on the first possible frame (but later realized that it would be too much unnecessary work to do so, so I started to use turbo a few seconds later after that). This one and only dialogue-delay in my entire run costed 283 (+/- 5) frames alone compared to your run. :-(
The real time saver you managed to get were ~1 sec in the first jungle room by skipping the Root pickup at the beginning and not killing another Wimpy Flower, and around 1,25 sec in the Raptor fight by manipulating their position somewhat better when hitting them.
Also, you obviously saved a few frames by skipping the name entrance screen. I wouldn't do that (because it will look stupid in the entire run whenever your name gets mentioned), but everybody has his own opinion. I don't expect it to save much more than a few seconds in the entire run anyway.
Your run looks great so far, and I'm looking for more progress. Good luck!
Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
I didn't considered to use ZSNES for recording, but since the classic ZSNES v0.989c I used for all my old runs works perfect even on SOE, there is a chance that the newest, with rerecords and other features added, version would also be stable.
As for Snes9x 1.51, I have high doubts it would be stable on SOE, since it generally has even more desync problems compared to 1.43 on some games anyway.
As for the run, you should know that my 100% one didn't aim entirely for fastest completion time but rather show the complete storyline, every sidequest, trick, and all the major glitches combined with all boss fights. It's just my own criteria for a true 100% run in this game, to show off every little thing it has to offer. However it is possible to get 100% of all items without playing through all sidequests or fighting every boss like I did, so if somebody goes that way, then my run is obviously improvable due to the major skips of the mentioned game-story. Just something to keep in mind.
But aren't you interested to rather make a any% run with fastest possible completion time, no matter what instead? We are still missing this most common speedrun form for SOE, so I think instead of making another 100%, the first true any% TAS would be much more appreciated by most people here. It would also require much less work and planning, due to a straight route that doesn't require you to bother about all the additional items you have to get.
In any case, a new SOE run is definitely something I would look forward to see, so good luck if you plan to do one.
Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
Secret of Evermore is a game that will always desync during recording, no matter what emulator or settings you use. The only way I could circumvent it was to cancel the run after every new room, playback it until that point and make a new savestate there to record from it further, step by step. It's a pain in the ass, but it works slowly.
The best settings for it is to use WIP1 timing and Fake Mute desync workaround. I highly don't recommend to use "Volume Envelope Height Reading". Although it fixes the sound problems you mentioned, it causes even more desync problems than not using it. Maybe it would make a stable recording even impossible, but I didn't test that so far.
Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
Cpadolf:
Nice find. Not sure this will ever be of use in the original, but I will definitely keep this trick in mind for later TAS work.
Kriole wrote:
We ran out of things to compare to, so we are standing here completely paralyzed in shock.
A bit exaggerated, but generally well noticed.
Kriole wrote:
It was "good" progress since the improvements were pinpointed and already planned ahead, all we had to do was to apply them consequently. It was WIPed directly at Torizo since we wanted to see your humiliation when we yet again take something from your "perfect"-folder and rip it into shreds.
I see. It's well known that a few people here have nothing else to do than to desperately trying to humiliate me. I hope you enjoyed the moment. :-)
Too bad you guys don't seem to be interested to see your own pwnage by showing progress after Torizo though, but on the other hand it's probably for the better that way, since I wouldn't like to spend some of my already limited freetime just to point out missing frames in your runs compared to my old any% anyway. I'm already busy enough TASing SMR after all.
Kriole wrote:
with improvements in mind/planned out for every single one of them.
That's what I mean. Empty thoughts. Otherwise how about showing a at least 1-frame improvement demo for rooms like for example the landing one or my new OTES climb strategy with your subpixel fix already included instead, when you say that they were already planned out and truly achieved? Seems like it won't work out well after all, wouldn't it? ;-)
As to the diagonal room, good luck finding the optimal 1st shot position. That's definitely something for those who use memory watcher, as diagonal slopes are full of subpixel variations that change your projectile angles massively depending of your position on them. Probably the hardest room to optimize in the entire game because of that.
Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
Wow, this is a serious improvement. It shows some new strategies I have never seen before, like making entire laps without even turning around but just bouncing into a wall. And the highlights were definitely the first 2 Choco Island courses. Making 4! laps in one turn by just precisely bouncing into the double wall really made me go 0_0!
Entertainment level is same good in both runs, but this one beats every course in a faster time and generally has a slightly better optimizatioin level, so I vote YES. Well done cstrakm.
Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
Cpadolf:
Everybody knows that the only claim I did on something that is possible without showing a smv was my Ceres escape time before it got matched by Taco. Otherwise I have always provided a smv for them. And just to repeat what was said many times before, all my claims (no matter whether a smv is provided or not) are based on real achievements. I'm not that kind of guy who relys on simple thoughts, despite being right in 99% of cases in them anyway. It's just not enough.
And of course a smv doesn't need to be fully optimized. The only thing that counts is that it proves that something is possible to execute.
moozooh wrote:
Moonwalk would have helped in that particular case if landing in the doorway, though. It's just that landing there was wrong to begin with, due to a completely different reason: taking damage boost to a farther location.
Yeah, obviously that's what you need to avoid in order to
Saturn wrote:
...precisely landing on that platform in the previous room already...
It's actually quite surprising that you didn't manage to realize that yourself before, while even having Cpadolf's run as a demonstration for the optimal landing position after the bat boost there.
As for the current progress, Cpadolf summed it up pretty well. The most promising run we can expect in the near future would probably be Cpadolf's and hero's 100% run. You can also expect a new WIP from my Redesign TAS in the next 1-2 months, even though it won't be a complete run yet. Taco and Kriole also did good progress until Torizo (which is coincidentally the point where my TAS demo stopped to which they referred all the time), but since then we didn't hear any news on further progress for a while now.
And Fano, you should indeed finish a more simple run like SM-any% at first, to get some more TAS experience before tackling a hard project like SMR low% yet. However if you feel like having enough knowledge to do both runs in the same time, then go for it. I'm looking forward especially for your SMR run.
Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
As far as I know the de-equip trick only increases magic defense, not attacks. I did a quick test to be sure, de-equipping Crono's armor and helmet in a NewGame+, and Luminaire still did ~6050 HP damage, just like usual. Maybe inichi can clear us up further, since he indeed mentioned that it would make your magic attacks stronger. I can't see a reason why it should do so though, as armor has nothing to do with magic power at all.
As for the mom-skip using the pause trick, the problem is that once the dialogue starts, you can't enter the menu anymore and thus get stopped before the exit like usual. :-(
Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
moozooh wrote:
Well, dunno, I kinda like the NewGame+ run, and it seems it might be further shortened, thanks to the recent findings.
Again, there are no strategical improvements known for the NewGame+ run yet. However if you "invented" some new ones out of nothing, feel free to let inichi and everybody else know so that people can start working on the new run.
OmnipotentEntity wrote:
You can fight Lavos as soon as you reach the End of Time. However you need the Green Dream which is acquired by completing Lucca's Mom's Sidestory to use the glitch apparently.
Yes, it's needed according to inichi. And the earliest point to get it is after you visited Zeal.
inichi wrote:
Perhaps, as you said, it may already have enough quality to meet TASVideos' standards. However, I just can't motivate me to submit a run which is improvable by almost 30 seconds.
Typical inichi. :-)
30 sec improvement on a 3+ hour run. Thats of course way to much to meet the standard here compared to a ~1 hour improvable run. But I can understand you very well, I would probably feel same about that.
inichi wrote:
Maybe, 2:30 - 2:35.
Surprising to hear that from you. I really assumed that skipping so many bosses (which take alot of time each) and the leveling up scene would save much more than that. Let's see what will come out when the run is done. :-)
Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
As usual... People who argue with me here are mostly those who were never involved in serious SM-TASing. No wonder why they won't be able to understand a couple of things, no matter how much I will try to explain them. Oh well.
Fabian wrote:
You can use .smv's as proof that something is possible. You can't, however, use smv's as proof that something is impossible. For example, saying "it's impossible to improve this because I have this smv right here" won't work out very well at all.
Obviously true. But I was only referring to moozooh's countless claims (in which many were proved to be wrong) he did in the past months, while never showing a smv-proof to confirm them like Taco, hero, Cpadolf, Kriole and basicly everybody else did. A smv says more than 1000 words and avoids all kind of confusion, yet moozooh always prefers the latter choice. Now ask yourself why...
moozooh wrote:
I was going to show the .smv where I take a boost from a skree in the previous room as well as taking the refill from it without losing time, but upon reviewing it just now, found a 20-something frames faster method only applicable to low% run. Oh well.
Nothing special here, now that I helped you to realize the optimal landing position after the damage boost. That's irrelevant though. The point was that moonwalk won't save you any time here (and generally not enough in the entire game to make a activation of it worth in any kind of run). Since you don't show a clear smv-proof, I assume you managed to realize that by now (in that case congratulations on the progress).
PitKidIcarus:
Ignore the useful stuff I contributed and keep backtracking to all the immemorial quotes. Yes, I was wrong on some of them, but at least I have learned from my mistakes.
Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
You're right mmbossman, I was terribly wrong there. But I learned my lesson well by now and am very careful when making any claims. Most of the time I do so, I already have a actual smv-proof to confirm them if needed (unlike moozooh). I think the past few months showed that clearly.
Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
moozooh wrote:
Saturn wrote:
Giving a example would never hurt, as we could surely help you find out whether it really could be of benefit outside of Ceres, unless you plan it to be a surprise for your run of course.
Here. Haven't tested it yet, but I'm anticipating at least 1-2 frames of gain.
That just proves it again. You land too far into the room in that smv, which is already somewhat slower than precisely landing on that platform in the previous room already (like Cpadolf did in his any%v2) if skipping the Spazer. The moonwalk will only be slower here, because you waste a frame stepping back for nothing.
I think everybody agrees that making false claims like this doesn't really help out except of yet again causing unnecessary confusion, so try to make a smv-proof next time in which you can clearly see the gained frames of the moonwalk compared to not using it.
moozooh wrote:
I talked to him about it personally. You clearly didn't. It doesn't make it any different from all the other situations that you've failed on, and others have succeeded.
Just to repeat the obvious again, we all got the same time at that door without moonwalk (hint: the previous post clearly mentioned that already). Whether you personally talked to him or not doesn't change anything as well.
Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
moozooh wrote:
It will easily save more, and I'm pretty sure I know where. The only question is if it'll cost anything somewhere else due to direction lock.
Giving a example would never hurt, as we could surely help you find out whether it really could be of benefit outside of Ceres, unless you plan it to be a surprise for your run of course.
moozooh wrote:
For the record, I already tried that door almost a year ago, and came only about a pixel short from the door opening trigger without testing all possible positions (it's boring), so I'm sure it's possible.
I, Taco, and many more too, and we all got the same result. And if even subpixel master Taco didn't manage to save a frame without moonwalk there (considering how much time and work he spent on Ceres alone, I'm sure he tried every possible combination), then I think this clearly speaks for itself. I realize that making realistic estimations isn't your cup of tea though. Hopefully you manage to improve this ability a bit more someday, as it can save you alot of unnecessary work in the end.
Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
Nice find Taco. It was definitely unexpected to see that moonwalk could ever be of use in a speedrun. Too bad you will only benefit of this in Ceres due to it's stupid, unique doors, and only save 1 frame max on a few of them with some luck, so it's unfortunately still not worth to use in a realtime run because activating moonwalk will easily cost you more than a sec. Even in a in-game oriented run, I wouldn't bother to activate it just to save 1-5 frames in the entire game.
moozooh wrote:
I've yet to see a conclusive proof it's impossible without it.
Nothing holds you back to try it yourself to get your conclusive proof. It's kind of strange that you always show interest in finding improvements, yet don't make any effort at all to fulfill this interests. Keep waiting. That's the best choice a dependent TASer can have after all.
As for the LN speedball tech, nice find hero. Although it looks very impressive and unexpected, it is unfortunately more than 150 frames slower (yes I know, it sounds unbelievable) than using it the usual way. Just compared it. The reason is that you lose alot of time while bouncing towards the bomb blocks since it's extremely slow compared to a usual non-bouncing morphball move.
And yes, you need the X-Ray in any case for that trick because it's the only way to turn around without losing blue suit, just like it was showed in the "Temorary Blue Suit" demo as well.
Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
I actually support your decision to use the skip trick, inichi. The test run already showed the limit for a legit playthrough without any boss skips while using very optimized strategies on them. Skipping the entire Lavos fight with the instant kill glitch but fighting every other boss would be inconsistent after all. And since you don't even have to level up for Lavos anymore, bosses like Nizbel II would be unnecessarily delayed as well due to the much lower level and inferior techs.
Going for the absolutely fastest completion time no matter what would on the other hand be a completely new run never seen before, so it's a good decision in any case. We can still have the option to publish your test run (which can easily be considered a pretty optimal one that totally surpasses every TAS standard for this game already) alongside the final glitched one, if needed.
Good luck with all the testing, and don't pressure yourself to finish everything faster. What counts is to make a optimal run in the end, no matter how much time you would require for it.
PS:
As for the final time with all boss skips possible, I would guess something between 1:50 - 2:20 to be possible.
Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
Thats right, GreenDream doesn't carry over, so no improvement to the NewGame+ run as long as it's required for the instant glitch trick (which seems very likely to be so).
Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
NO! This is too much already.
If this allows to completely skip any leveling up, but therefore just getting the GreenDream, then it would easily save 20 more minutes I guess. Who would ever imagine a game like CT could be broken so badly.
Unbelievable stuff you find over and over again, inichi. You proved without a doubt to be one of the most creative TASer on this site already. Much respect man.
Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
I don't get what problems some poeple have when one states his opinion. So let me try to explain a few things:
PitKidIcarus wrote:
Also your comment about
"My previous statement was actually meant for all the other rooms" is complete bullshit. Its just another reason why you cant accept things so you make stupid excuses like this one to try to cover up the fact that you got owned. So you refuse to face the fact that someone knew something you didnt.
No, I mentioned it at least 3 times before that I never tested the new "aim-down" falling tech people started to use in various rooms only recently, so every room that involves this tech can from now on be considered improvable by a not yet known amount of frames, since it still isn't fully tested. The last 4-frame improvement Taco showed is just another optimization to this falling tech, and I don't exclude further possibilities that there are even more improvements possible in that mainstreet fall room.
But since every other room until pre-Torizo doesn't require you to fall somewhere, there are very likely no further improvements possible (except of the mentioned 1-2 possible subpixel optimizations, which are still only theory), since everything was double-optimized to the max already. The 10-frame improvement in the morphball room alone that Zurreco mentioned is just far from being realistic, which I just wanted to make clear.
symbolic X wrote:
Keeping a secret?! Saturn, are you nuts? Yes, there are a few secrets in a video game like this one the more you play it. BUT, calling your progress a secret makes me believe that you're too selfish to share your input on this game.
Has nothing to do with being selfish or anything. As Drewseph already said, there is just something really cool to show off for the next WIP (which should be done in the next 1-2 months, since I started to work on it again after a longer pause), so it's definitely better to avoid progress reports on it for a reason you will clearly understand when the next WIP is out.
AKA wrote:
Its not clever to claim perfection on a game like this, but equally its not right to claim perfection can never be reached.
Very wise statement, which finally somebody realizes. Although it's admittedly very hard, or in SM's case pretty much impossible over the entire run due to the random item drops after every boss fight, perfection can be reached when you eliminate all improvement-possibilities, at least on games that don't require any luck manipulation at all.
And to the few idiots unaware of anything who call me a troll, I recommend you to check this thread more thoroughly so you can eventually realize how much useful stuff I already contributed to this game alone, not to mention a couple of other games. I wish to thank AKA for giving a good example of what a troll really is, since a few people appartently don't seem to get it.
I hope people chill out a bit. There is no need to bitch about everything somebody says.
Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
Taco wrote:
Saturn wrote:
Maybe there are 1, at most 2, frame improvements possible until the pre-Torizo segment (and even that is highly doubtful), but definitely not more. You will see.
Well done, Taco. I realize that the new aim-down falling tech has much potential for new time saver, which however I didn't bother to test as mentioned before, so I wasn't aware of them yet.
My previous statement was actually meant for all the other rooms, without counting the mainstreet fall, since it still lacks the needed tests to be sure about the limit there so far. If not counting that falling room, my statement still stands.
Tub wrote:
Little bald boy: Do not try and improve the frames. That's impossible. Instead... only try to realize the truth.
Taco: What truth?
Little bald boy: There are no frames.
Taco: There are no frames?
Little bald boy: Then you'll see, that it is not the frames that improve, it is only yourself.
Matrix ftw. ;-)
Nice improvisation, which I admit has indeed something truthful in it.
symbolic X wrote:
How far have you guys gotten in the regular and redesign projects?! Have you guys at least reached Ridley?
My Redesign progress should be a secret, so that the next WIP can blow everything off when it's released. ;-)
P.JBoy wrote:
You will still have to brute force it (to a lesser extent) without RAM watch due to lag frames, keep that in mind
Yeah, to a lesser extent, which is already enough to ease the work alot for me.
Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
mmbossman wrote:
Saturn, I'm curious how you're so confident about everything being subpixel perfect when you don't use any sort of RAM watcher?
It's just knowledge I guess. I generally don't need to see any subpixel numbers to realize whether something is improvable or not. It's a routine for me. There are only very very rare cases where I might benefit from the RAM-watcher, like at the jump towards morphball as seen in Powerrangers recent WIP. Other than that I would rather blame my lazyness to overlook subpixel optimizations, not the lack of the RAM-watcher.
DK64_MASTER wrote:
Taco wrote:
Saturn, please stop saying that everything to Torizo is perfect.
http://dehacked.2y.net/microstorage.php/info/184967545/zebes%20improvement.smv
Another frame is gained in that smv by not turning around before entering the OTES (the first visit), which allows me to jump one frame earlier. Of course this messes with the elevator, so I shoot in the OMBR to add lag to the door transition (real-time is what affects the elevators, not in-game time).
I'm also sure that there are still more frames to be gained.
Btw, here's a good timesaver in Red Brinstar: http://dehacked.2y.net/microstorage.php/info/1956540303/fall.smv
Tacowned.
Not really. I just didn't test that Zebes frame improvement because it is only possible due to the new falling tech that obsoletes the moozooh-jump (which I didn't analyze at all yet). It's actually pretty obvious to not use a turn around there, since it's possible to just fall through the door without hitting it. However in the older falling techs you would always hit the door in the last frame before it's fully open, so the only purpose of that turn around was to circumvent the falling cancel of it, nothing else.
Maybe there are 1, at most 2, frame improvements possible until the pre-Torizo segment (and even that is highly doubtful), but definitely not more. You will see.
As for the Red Brinstar PB Fall, I used the morphball variant Taco showed in my any%, but without wasting the Missile or taking damage. The only person who actually pwned me and Taco there is Hero. His falling tech (which yet again involves the, to the time I worked on my run, completely overlooked aim-down-fall) is even slightly faster in the end, since it allows to shoot the door much earlier without any sacrifices.
And yeah, the jumping height starts to vary massively at different speeds, once you get Speed Booster. I realized it pretty soon after I collected it in my RBO run. Abusing it allows to gain some serious time especially when you don't have HJ-Boots, as we could see in some places in Hero's any% v2 run. Much thanks to P.JBoy for posting the complete value list of them. It will surely be more handy than to bruteforce every jump like I had to do in my previous runs.
Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
In no particular order (since it's hard to compare the skill between different games):
- Super Street Fighter 2 (SNES) - Except of very few crazy SSF2-masters on the internet, I have never faced a superior opponent in this game, till now
- Nintendo World Cup (NES) - No superior opponents ever played, even on the internet
- Super Metroid (SNES)
- Super Mario Kart (SNES)
- Secret of Evermore (SNES)
- Terranigma (SNES)
- Chrono Trigger (SNES)
- Seiken Densetsu 3 (SNES)
- Sim City (SNES)
- Super Mario Bros 3 (NES)
Experienced Forum User, Published Author, Player
(209)
Joined: 2/18/2005
Posts: 1451
I see, you have no idea for any improvements, yet believe there are still any. I can understand that. Until you convince yourself in every single room, you just won't be sure about it. I redid that part 3 times myself over more than 2 years, doing it fully optimized in the final attempt except a few missed subpixel optimizations, which Taco and Kriole fixed by now. There is nothing more left after that because you just hit the limit at one point, as unbelieveable as it may sound at first.
I actually did a horrible mistake at this point by not triggering this dialogue to start appearing much faster by simply pressing the action button 2 (not 1 time like I did in that case) times before it. This first few dialogues in the beginning were the only case where I didn't use turbo/autofire in the entire TAS to fast-forward them, because I at first planned to do that manually by pressing the action button on the first possible frame (but later realized that it would be too much unnecessary work to do so, so I started to use turbo a few seconds later after that). This one and only dialogue-delay in my entire run costed 283 (+/- 5) frames alone compared to your run. :-(
The real time saver you managed to get were ~1 sec in the first jungle room by skipping the Root pickup at the beginning and not killing another Wimpy Flower, and around 1,25 sec in the Raptor fight by manipulating their position somewhat better when hitting them.
Also, you obviously saved a few frames by skipping the name entrance screen. I wouldn't do that (because it will look stupid in the entire run whenever your name gets mentioned), but everybody has his own opinion. I don't expect it to save much more than a few seconds in the entire run anyway.
Your run looks great so far, and I'm looking for more progress. Good luck!