Donnerstag, 28. Januar 2016

World record adventures

Earlier today my culminated effort for Imperishable Night Extra bore fruit and I broke the overall world record for it by 55 million points. I have played this category more seriously starting October 2014 and played it for much of 2015 until today. To start out with this blog entry, I will go through the replay step by step more or less; I won't go into detail for every grazed bullet and will kind of skip over simple attacks that went as they should but you get the idea. I may refer to ASL's run for comparison in terms of time and maybe graze (but probably not graze). The run is available here.

Ignoring the opening furballs, the stage starts with a frame perfect cancel which I miss as I was 2 frames early. The desired time after it is 750, but I got 658 but anything over 600 is alright to me at this point. Following this is killing the fairy that spawns together with the one I just killed. My method for it is slightly different from everyone else for it: I shoot the fairy directly at the right side of the big fairy that spawns as I go for the kill which cancels bullets from the big fairy. This allows the familiars to spread out more as opposed to killing the fairy by shooting the familiars it spawns. My way is slightly more optimal but only marginally so.

The three large fairies that follow are very simple. I delay killing all of them a bit (for the first and third I misdirecting my option whereas I stop shooting for the second). This allows for bigger cancels on all of them due to more bullets having been spawned. For the fourth and last large fairy, I once again stop shooting after dealing appropriate damage to allow the furballs to shoot their pellets and cancel them for a cancel twice as large as if I had just killed the large fairy on its own. After this section my preferred time count is 6,500 (ASL reaches 6,150), which I reach comfortably.

The following cancel has several parts to it. The first part is position myself properly (horizontally is more important than vertically, but too high or low can cause some time loss) as well as deal some damage to the middle fairy and the right fairy that spawn without killing them. Part two of this cancel is as follows: they shoot their familiars, I move over to the left side of the screen and kill the leftmost fairy. I killed it a few frames (two or three at most) early so the familiars didn't cancel out as many bullets as they should have, which cost some time. The damage dealt to the middle and right fairy wasn't perfect either so I destroyed familiars on both of them before actually killing them. The third part of this cancel simply entails killing the two fairies that spawned later. In total this can be done for 2,300 time in my experience, but this time I got just below 2,000, leaving me with about 8,500 time thus far (8,250 for ASL). Preferred time here would be 8700 but I decided that 8,500 was enough to continue.

The next cancel is very similar to the previous one but tighter with the timing as all five fairies spawn at once. The angle for my option is different this time and the angle to directing the familiars has to be steeper but the procedure in killing the fairies is quite close to what I did before: After they shoot their familiars at me I move over to the left side and kill the two leftmost fairies. Just like before I now have to kill the middle and right most fairy, leaving the fairy second to the right alive. I kill it afterwards. In total, this cancel can give as much as 2,700 if everything was perfect (positioning, damage control, timing and RNG for the bullet density) but in my case I get about 2,300 which brings me up to 10,850 time (ASL has 10,600 here). The time I want to have at this point is 11,200 but again, I wasn't so low that it warranted a reset.

The last cancel before Keine is the hardest in the stage and a complete mess to be honest. This part spawns a total of eight fairies, all of which spawn the first six familiars, whereas only five of them spawn the second set of familiars and this cancel aims to kill six of them with familiars spawned. Unlike the previous cancels, I start at somewhere around the middle on the screen in height and aim my option towards where the rightmost fairy is going to spawn. As soon as they start shooting I tap down to have my option sweep toward the third fairy on the left. I now move over immediately to kill the three rightmost fairies. After they died I move back to the left side of the screen to kill fairy number one, three and five from the left with the second set of familiars spawned. Unfortunately my damage control was bad so the fifth fairy dies as I go back, which it shouldn't. Moreover, my failed damage control also led to some familiars being destroyed where they should not have. If this cancel is done perfectly, as much as 3,500 time is possible but I only got 2,850 in this run, which is pretty bad. However, my total time was over 13,800 which is the minimum I want to have when I enter the midboss (ASL has 13,400). So in that sense, the run had the worst possible stage portion before Keine that I would have accepted. My goal to reach for this is normally 14,000 time because it's a round number but what I really would like to have here is 14,200-14,300 as that was as much as I had in my previous high score.

Previously, Keine's first spell was a speed kill where the only thing worth paying attention to was the cancel but recently I looked into the attack and found a profitable way to graze it. The graze nets me slightly over 1,000 graze and a total of 3,700 time due to the fact that I start the spell at -50% but can quickly go to 30% on the gauge as well as the slightly larger cancel as opposed to 2,200 time (and like 100 graze) compared to what was done before. Comparing the strategies, grazing the spell is this manner is worth over 12 million points but there doesn't seem to be much room for optimization. Small cute detail: I go from 50% to -50% with the time gained by the familiar cancel to collect the items and then switch bad to 50% with the time gained from capturing the spell so I can immediately start grazing the next spell. After this attack I had 17,500 time (ASL has 15,600).

Keine's second spell went rather well in this run. My goal for Ichijo is to get 8,000 time if at all possible (ASL gets 7,400) and in this particular run I managed to get as much as 8,200 which is pretty good even by my absurdly high standards. The highest I have gotten is 8,500 time so this run didn't have bad grazing here at all. Upon finishing the spell I have 25,700 time (ASL has 23,000 roughly).

Neither ASL nor me actually use the best strategy for grazing the third spell. But I have a good excuse: I didn't know about it. It was only after I had gotten this replay that I saw a new replay uploaded by the player t1100 who improved on this spell by an additional 300-400 time compared to me. But be that as it may be, my way is a slightly tweaked approach to ASL which isn't a huge improvement but it translates into 350 more time if done properly. My gain on this spell is 4,200 in this run (ASL gets 3,850). In addition to the time and graze, my spell bonus is 1.3 million points higher, which makes this improvement somewhat sizable overall. I finish the mid boss with 29,900 time (ASL has 26,850 at this point).

Following Keine is a cancel identical to the very first cancel of the stage with the same idea behind it, too. The only difference is that I now have full power - so of course I mess it up. I actually never took note of how much I gain from those two fairies in total but it was 1,100 in this run; so about the same as it is in the beginning of the stage if done badly. And just like in the beginning of the stage, what follow are four big fairies. This time I do not stall for extra bullets to spawn because that would kill me. Each one of them is 1,200 time if no familiars are destroyed and after this section I have 35,600 time (ASL has 32,500). I personally would like to have 36,000 here but that doesn't really add up with my wanted numbers everywhere else so I'm satisfied with anything over 35,500.

Now comes the only part in the stage where every player loses items. How many differs depending on how many fairies escape. Taking ASL run as a basis (who enters Mokou with 668 items), I let three fairies escape. The highest item count realistically should be 671 or 672 but only if you really get everything realistically possible which doesn't happen very much if you're me. This PoC section is pretty annoying because of that because I sometimes had great run dampened by a really low item count. My average is 666 items entering Mokou. I'm a mischievous child.

The cancel before the final big fairy is identical to the cancel where five fairies spawn at once, except they have less health now for some reason, which makes this cancel easier than the other one if you have found a working approach. It goes well in this run too; the large fairy before Mokou also always gives the same amount of time. I enter Mokou with the following:

Score: 776,264,850
Graze: 8,575
Items: 665
Time: 40,541

Score: 766,936,560
Graze: 8,780
Items: 668
Time: 37,904)

The reason why I have less graze is because before the cancel before Keine there are some furballs which shoots tons of pellets like in the opening and ASL grazes those unlike me, who instead sets up for the cancel.

Overall this run is very average if I put together everything individually. The cancels in the first half of the stage weren't very good but I made up for it with a good Keine fight and the second half wasn't particularly bad or good. As far as time entering Mokou goes, I could get as high as 42,000; maybe more. But that's also only if everything is really good. I didn't have any gripes with the run and had no qualms continuing like I would have on a bad run. So I did.

The first non spell was good. I got 4,000 graze; 3,800 time because I have to start it at -50%, which is as much as ASL also gets. Depending on Mokou's vertical movement, it's possible to lose a bit of graze but it fortunately wasn't the case here. Iwakasa went by without a hitch.
For the second non spell, I actually included spinning around Mokou into my route and have to start the attack above her but I was somewhat absentminded and forgot to position myself. This lose me a fair bit of graze worth up to 1.5 million points, albeit slightly alleviated by the fact that in return I got to shoot Mokou with her armor for some extra time instead. It doesn't quite make up for it but makes the loss less painful. I still milk the attack for the whole duration for graze. Flying Phoenix, much like Iwakasa, went by without trouble or problems.
Much like the second non spell, I milk the third non spell just for graze. Unlike the second non spell, I did not forget to position myself for this because there are no spins for this attack. And again, the spell following this non spell, in this case Flaw of Forgiving Shrine, was all okay. Following these three non spell and spell cards, my time is at 51,259 and my graze at 14,559 (ASL is at 48,720 and 13,503). As far as time gain is concerned, there isn't any big difference between us, considering the fight so far has been quite uneventful for this point.

The fourth non spell sort of heralds the start of the great milks for time. Of course, neither the fourth nor the fifth non spell give all that much graze but the spells in between do. For me, xu-fu is the hardest spell to match ASL on. It has quite difficult optimization and it took me a long time to be able to actually match him. I did well this run as far as grazing is concerned but my mind blanked out somewhere in the middle. This is problematic because at one point I just "woke up" and heard the counter and just finished the spell in a hurry, which was one swing too early. This missing swing lost me about 100 graze. I do believe the cancel was worse than it would have normally been but I'm not convinced about that part. In the end, my time gain on Xu-fu was 13,000 (as opposed to ASL's 13,600). My personal goal for Xu-fu is always to at least match ASL. This was not good and one of the more annoying mistakes to me because it was a brain fart and nothing more.

The fifth non spell is just a slightly harder version of the fourth so I'll leave it at that. Honest Man's Death is my biggest improvement in this run and is also the biggest improvement I had in theory. The reason is very simple and obvious: I'm much higher on the screen than ASL and graze a lot more bullets. My practice result indicate that as much as 20,000 time is possible on the spell (ASL gets 16,200). Of course as this is a run I don't go balls deep and settled for a reasonable height. Obviously as everyone knows this is the most difficult spell in the stage so whether something is actually a "run" to me depends on whether or not I capture this. Much potential has been lost for me at this spell but this time it went well. I finished this spell with an extra 18,900 time (and anything over 18,700 is fine to me). After Xu-fu and Honest Man's Death I have 84,679 time, 25,339 graze and my score is at 1.547 billion (ASL is at 79,869 time, 23,583 graze and 1.513 billion score). Already my score advantage is very large.

Starting with the sixth non spell, I don't milk then anymore for graze but just finish them as quickly as possible. As I always make known: I can't dodge so this part of the run is very scary for me. Until after Fujiyama Volcano, every attack has dodging of some kind and it's freaking me out every time I get this far. I was fortunate that the RNG didn't force me to do anything wacky on this non spell.
I was pleasantly surprised by this Woo, to be honest. It rarely goes so cleanly. I don't know if this is RNG or me becoming better at the spell but I have died to the parts I refer to as RNG in it less often than I used to which is nice.

Seventh non spell was very nice to me just like the sixth, which I was very happy at. Phoenix Tail on the other hand wasn't. When I started the spell I contemplated whether I should just shoot Mokou's wings by angling my option and all. This is actually surprisingly profitable and nets up to 500 more time but makes the spell last notably longer so I didn't do it. But the RNG in this spell forced me to dodge a lot in quick succession which ended up forcing me above -30%. This was actually a quite large loss as I lost around 3 million from the low spell bonus alone. I don't know how much time it was that I lose but it was rather substantial I assume. Even after watching the replay I don't really see how it could have been avoided other than looking into the future and knowing I would be forced to dodge that way. The only alternative I can think of would be to just dash to the side unfocused but there is a high chance I would have hit something on the way. An unfortunate incident.

Following Phoenix Tail is the hardest part as far as dodging is concerned and I'm always scared on the last non spell that I just die to something I didn't expect to die to. Fortunately the RNG was forgiving in this run so I was able to capture it. Incidentally, this is my first capture on the attack in a run so that's a quite funny coincidence.
Fujiyama Volcano has two viable ways to cancel it. I've talked about this before. My cancel is a big improvement for something so short, and if done on one of the two best frames it may be a 5 million improvement with my graze. There is a risk though: If you cancel too early you lose your items (45 items at a bit over 1.5m value each). If you're a bit late then it doesn't matter much. The score increase is just going to be a little smaller when compared to the optimal cancel. Now, I don't care about the risk but I need Mokou to cooperate a little bit with her movement in order to deal enough damage so I can comfortably choose the cancel timing. She unfortunately didn't do it this time which left me with no choice but to go for the scrubby cancel everyone else uses. Unfortunate. After I captured Fujiyama Volcano, my time was at 96,431, my graze at 26,289 and I had a score of 1.951 billion (ASL was at 91,903, 24,697 and 1.916 billion). My score advantage increased a bit more but I lost a lot of points due to the mishaps in Phoenix Tail and Fujiyama Volcano.

At this point in the run, I was rather nervous. I had visions of failing the streaming in Possessed by Phoenix but then I remembered the spell is free. It was an okay capture too.
Hourai Doll is the spell that has killed most potential new high scores and I was understandably worried it might happen again. That actually led to me avoiding graze in the beginning. But as I saw the bullets fly past me, I remembered the horrible grazing on my previous high score and recollected myself a little. The grazing still wasn't good in the end but let's use the word acceptable here. Some euphemisms are necessary. my desired amount of graze for Hourai Doll is 3,000 but I ended up with 2,850 this time (ASL does get almost 3,000. He's at 2,980 or something).

Imperishable Shooting is the final spell and at this point I knew I was going to get a new high score but to have a chance at a satisfying score I needed to capture it with not horrible grazing. I think I managed that well. I finished the spell with a spell bonus of over 38 million. 38 isn't good for me if it's in spell practice. I usually aim for over 40+ then. But I was a nervous wreck and couldn't really go for it somehow. Bullets scared me. However, this wasn't spell practice and there was actually something to lose this time if I failed the spell so I think 38 is adequate. The final numbers of my run are as you can see in the screenshot. An overall 55 million improvement over what was a very long standing world record.

That's the run in a more of less detailed manner. I have difficulties saying how I felt when I got it. Predominantly, I was quite happy. This is my best score by far and saying I'm not proud of it to some extent would definitely be a lie but the same goes for the mistakes I made. I can't say they don't annoy me at least to some degree. But still, I think this is the kind of score one could be satisfied with in the long run and just leave the category behind. I have spent many fun hours in this stage, learning new things and improving upon them. Understanding the nuances of certain cancels better than before and improving upon the grazing. Those were very fun aspects. Also trying out new grazes was fun. I didn't find very many new things on my own (The only thing I found by myself are Keine's first spell graze and the wacky Fujiyama cancel), which is somewhat unfortunate but I'm not very creative I suppose. Overall I believe scoring this stage as seriously as I did has helped me improve as a player on some levels (precision and understanding mainly) but I can't really be sure about that until I've tried other things. And that's what I plan to do now. My next main project is going to be Phantasm with ReimuB. Along with it, I will continue to play EoSD lunatic as a secondary project. I do not know how far I want to push either but I hope I can enjoy them enough to get somewhere kind of relevant. However, this does not mean this category is dead to me. I will come back to this more likely than not and then try to improve this score further. More people should play this stage for score in my opinion. The step to a really high level score is steep but starting it is quite pleasant in my opinion (compared to like SA, let's say). Looking at how new things are still being found (by other people mostly), I think 3.18+b would be nice as a final score.

So please wait warmly. Girls are preparing.

Keine Kommentare:

Kommentar veröffentlichen