A theory on the Dynamics of Hurricane/Rapid fire with DR in Pink Bean

If you have done PB before as a corsair/BM, you know that it is very difficult to survive DR when you are using hurricane/rapid fire. The meta strat for BMs have been to switch to strafe when DR is off cooldown, trading off DPM for consistent survival. However the DPM loss of such a strategy is not negligible, the following is a comparison of BM DPM using pure hurricane and hurricane + strafe on DR.

Spoiler: pure hurricane

Spoiler: hurricane + strafe

Because the dpm loss is so significant, there have been pushes to find a consistent strategy to survive DR using only hurricane. Some people have claimed that you can react to the DR animation as long as you look closely. Indeed there is a fair amount of delay between the start of PB's DR animation and the beginning of DR doing damage. However, the reason why reacting the animation is inconsistent is because of situations like this,


The DR animation is actually already starting, but obviously you cannot see the bean at all given the clutter, let alone the DR animation.

Some authors have posed the idea of reacting to an audio cue instead of an visual cue. PB does in fact play a particular sound effect before casting his DR animation, it sounds like sounds like a low humming noise.

Spoiler: DR sound

https://voca.ro/1du9qxJxZznV

However in midst of a 30man run, this sound is often drowned out by the much louder noises of attacks hitting PB. It has been noted by akashsky that if you stand closer to PB, the DR sound is louder than usual, which may make it easier. However, after listening to many recordings of PB runs and trying to listen to the sound standing as close as possible, it seems to me that at least with my presumably average hearing abilities this does not lead to a consistent strategy.

These ideas have been thrown around, however it has yet to be demonstrated that PB can be consistently survived using pure hurricane. There are several people who still attempt to pure hurricane PB, and while they can react to most of the DRs on a good day, they still end up dying 1-2 times per run on average.

With the introduction to our problem out of the way, I would like to propose a new method of surviving DR's with hurricane,

Only looking for 1's

Formerly, it has been thought that reacting to the 1's is not viable, as it seems like by the time you see the 1's it is already too late. However after doing some more quantitative analysis, I will show that this is not the case. My method is as follows, let us look at recordings of recent PB runs and count how many frames there are between frame 0, the first frame where a 1 is visible, and where the DR damage starts. The videos I analyzed are courtesy of akashsky and are at 30 fps. Here is an example,

Spoiler: frame 0

Spoiler: frame 21

You can see that frame 21 is the first instance where someone took damage from DR. Of course there are 2 problems with this naive analysis, there is no guarantee that someone will hit the bean on the earliest frame that a 1 can be produced, and there is no guarantee that someone will hit the bean on the earliest frame where DR damage starts. What is invariant however, is the first frame in which a 1 can possibly appear (that is first frame where if you hit PB you get a 1), and the first frame in which DR starts (that is when you hit DR you take damage).

We can go through some more scenarios to try to figure out the minimal gap betweent he first visible 1 and the start of DR. I went through this type of frame counting for every DR I could find from akash's videos. The smallest gap I could find between the first visible 1 and DR damage was 10 frames.

Spoiler: frame 0

Spoiler: frame 10

Now the human reaction time is well within 10 frames as 30 fps, but if the gap really was 10 frames it would be very hard to do consistently. If this was the end of the story, I would have to conclude that it is either extremely hard or impossible to consistently survive DR with hurricane.

However there is one more thing to keep in mind. The above scenario represents the minimal gap, it means that that the first visible 1 happened as late as possible. We can however, make the first visible 1 consistently appear consistently within some window of the earliest possible 1, within 3.6 frames actually. This can be done by using hurricane/rapid fire on the bean before DR happens, since the interval between each hurricane/rapid fire arrow is 3.6 frames, the first visible 1 can appear at most 3.6 frames after the earliest possible 1. See the following scenario where akash was using rapid fire through a DR.

Spoiler: frame0

Spoiler: frame 19

On frame 19 akash cancels his rapid fire and takes no damage from DR. So let us assume the worst case scenario, suppose akash got his first visible 1 the earliest possible. Which means that at worst, the first visible 1 can appear at frame 3.6 at most. Let us also assume that akash cancelled his rapid fire as the last second possible, that is if his last bullet was just before the start of DR. Taking into account that rapid fire bullets have a period of 3.6 frames, the earliest his last bullet could have been shot is frame 15.4. This means that in the worst case scenario, there are 11.8 frames between the first visible 1 and the start of DR. Furthermore, if you have over 15k hp, you can afford to take 1 hit of DR, which extends this minimal window by 3.6 frames, in which case you would have at worst a total of 15.4 frames to react to the earliest 1.

Doing a quick reaction speed test on training pack for ssbm,

Spoiler: reaction test at 60 fps

I get around 16-18 frames at 60 fps, which would be 8-9 frames at 30 fps. My reaction time is not the greatest either, for example top ssbm players can consistently react within 13-14 frames at 60 fps.

This means that even in the worst case scenario, if you commit to hurricane/rapid fire on PB you have more than enough time to react to the first visible 1, in fact you probably have enough time to react to the 4th or 5th earliest 1 given that we are considering the worst case scenario.

I went through some old PB recordings and tried to press pause the moment I saw a 1, and in almost all of the situations I was able to react within 10-11 frames of the first visible 1. The only times I would have died were in cases where the first visible 1 was very late, like the 10 frame gap I showed up. But in those situations there was not a BM or corsair hurricaning/rapid firing PB, so it is to be expected. However, I have yet to test this in an actual PB run, so I cannot claim this method to be conclusively viable.

In summary, you have at worst 15.4 frames, or about half a second to react to the earliest 1 to stop your hurricane/rapid fire. I welcome any brave BMs out there to test out my idea and completely ignore the animation/sound, reacting to the 1's instead. If I die 4 times again next I blame akashsky

Spoiler: The most important part(melee hurricane)

I was thinking about this as I was going on my walk today, and I realized that I failed to take into account the time it takes for the arrow to reach pb. The amount of damage that an arrow/bullet does is calculated the instant the arrow leaves the bow, however the damage does not show up until the arrow hits the bean. Therefore to optimize the earliness of the first visible 1, we must also stand as close to possible to PB to minimize arrow travel distance to the bean.