I continue to play Imperishable Night extra with Youmu. Compared to April, the route has not changed that much but some potential changes and actual changes were found and made.
Let's start chronologically in the stage:
It is possible to enter the midboss with about 3,000 graze more than what I currently do. I don't graze these bullets for cancel convenience, but doing so would be a fairly significant improvement in the route, as this is before even the first boss pattern in the stage. I did the math before, looking at how many bullets each pattern has and we're looking at 7m-8m score improvement, more or less. This is obviously not accurate as cancels will be different here and there since the amount of bullets when you cancel a pattern doesn't always have the same amount of bullets onscreen. The improvement in undeniable however. Doing it would increase the number of resets tremendously though, so I opted out of it for my current goal.
Next up we have Keine's first spell. In the last blog post I mentioned how I hadn't been able to find an improvement for it. This changed not long after. The improvement seems fairly small with 50-80 extra graze, but this is deceptive. This small graze improvement gains me 2m-3m score, due to the extra time gained keeping the spell bonus of the spell higher than it would otherwise have been. On a spell like Xu Fu or Honest Man's Death where you cap the spell bonus either way, this kind of gain isn't very large but for both Keine's first and her third spell, this doesn't apply. This new strat is also more consistent than the old one, even if it's harder in execution, because it removes the RNG element the spell had. You can find it here.
Next up is not a route improvement, but simply a vastly improved practice PB for Ichijou. Since 2015 (!) my practice PB for Keine's second spell was 2827 graze. Earlier this week however, I beat it and got 2892 graze on it. 65 graze doesn't sound like a lot but it's a significant improve at that level and also confirms that 2900 is definitely possible; something of which I wasn't certain before. This raises the theoretical maximum of the stage by roughly 2m score. I have uploaded it on Youtube as well. Additionally, I personally feel like my average went up its fair share, albeit my consistency suffers a bit in return. I'll have to practice this a bit more extensively in the near future so both things work out for me.
The last improvement I have to note since the last blog entry is that I managed to break 50m spell bonus for Imperishable Shooting. This is already some time back, and I have beaten that first 50m already, with my current highest spell bonus being 50,080,000. You can watch my first 50m replay here.
What follows is not actual improvements but rather some analysis I did on some things. I'm putting this here mostly for documentation purposes as I don't think this is particularly interesting to most people.
The first thing I want to talk about is xu fu. I've mentioned before that it is an incredibly difficult pattern to optimize. This obviously hasn't changed but for a long time I hadn't actually looked at the pattern more closely. It turned out that my practice PB (the only replay I have that is over 15,000 time) ends up making larger strides than I anticipated, also on parts I didn't expect. This replay is also 5 years old at this point and the break I took starting in 2017 certainly didn't help my memory. Armed with this new (old?) knowledge, I can try and aim for a much higher average, I believe. This may have felt obvious but knowing how much the grazing can be stretched per cycle is actually quite important and I had apparently forgotten.
The last thing I want to talk about is not in regards to a specific pattern but in regards to Youmu's time gain in attacks where Mokou's phoenix has a hitbox. Technically speaking, in those attacks, Mokou has three separate hitboxes, two of which do not have a collision box. These are the two wings and Mokou herself. However, what is important is how the damage is distributed and how the time gain work for these when shooting unfocused. I wanted to find out which way of shooting those three hitboxes would end up gaining the most time. The reason this might be problematic on Youmu is that her option deals half damage when hitting the same target as the main shot, which leads to lower time gain as that's tied to damage dealt to a target. My testing ended up suggesting that shooting the wings gains the most time, but that it also makes no difference whether both shot hit the same wing or whether I angle the option and hit the other wing with it. The time gained was identical. If my testing was accurate (which I believe it was), that means that the wings and Mokou are all linked in terms of damage/time gain, rather then being entirely disconnected hitboxes from each other. This is different from familiars, who have their own health, which in turn in linked to the boss HP. A familiar having their own health means that Youmu's option deals double damage to it, gaining more time. This not being the case with the wings is an incredibly important discovery because that means angling Youmu's shot if there are no familiars on the other wing is pointless and just makes it unnecessarily harder. Additionally, getting more time by shooting the wings rather than Mokou herself is proof that Mokou takes less damage from her wings, but the difference in time game is not nearly as large as it is for familiars. A familiar transmit 50% of the damage it takes to the boss, whereas the wings transmit I assume 75%.
This discovery means that shooting angled on attacks like Flying Phoenix or the final three non spells, which all spawn no familiars at all, gains no score compared to just hitting one side of the wing. This is obviously different for spells like Iwakasa or Phoenix Tail, which have familiar spawns. The pattern I tested it on was the final non spell, and instead of ~700 time, I gained just under 900 time. This means potentially ~600 more time from the final three non spells at a high level. This is significant at a level higher than what I play, and will play a role in the future if someone tries to get a score of 3.3b.
As it stands right now, the theory maximum of this stage lies somewhere between 3.33b and 3.35b. I'd have to do the math for something more precise, but this is not quite enough leverage to get to 3.3b itself, unfortunately; even at the high end of 3.35b. I will continue to play and hopefully raise the potential of this extra.
Thanks for reading.
Keine Kommentare:
Kommentar veröffentlichen