The LSD of Amiibo Training: Amiibo Brain Transplants

by Doc – Owner, Founder, Didn’t Get the Title of Amiibo Doctor for Nothing

This topic is really hard to explain: if you have questions, ask them in the comments or send me an email at amiibodoctor@gmail.com so I can clear up any confusion.

One of the last training methods that an amiibo trainer learns about is the amiibo “brain transplant”. This is by design: it’s typically pretty hard for someone to grasp the standard “mirror match and do what they should do” training style that every high-level amiibo trainer recommends, but once a trainer understands that, they’ve got most of the necessary info to train an amiibo. As a matter of practicality, one’s “amiibo education” often ends there.

However, the hallmark of a good trainer isn’t training an amiibo the same way as everyone else – good trainers are defined by their ability to take something previously considered optimal or worthless and to improve it. Anyone can take a high-tier amiibo and beat their opponents, but the best trainers win with underdogs.

That ain’t always easy – the higher-tier amiibo are the most optimized amiibo in the meta, and potential improvements become more scarce every year. Over time, trainers begin to agree on an identical training method to the point where training guides can sometimes be summarized in only a few sentences because their optimal training method is so orthodox it may as well never change.

But this is Smash Ultimate, and in Smash Ultimate any character can beat any other character given the correct playstyle and set of inputs. To that end, new options and playstyles should always be discovered and experimented with – that’s where the amiibo brain transplant comes in.

A demonstration and explanation of amiibo brain transplants

The Amiibo Brain Transplant

The amiibo brain transplant is a technique that I invented back in the days of Smash 4, much to the consternation of the amiibo scene at the time. It’s very simple to execute, but its uses are difficult to understand. The actual technique simply involves writing a fully-trained amiibo of one character onto another character, and using a workaround that allows amiibo to write to different characters. The most common method for this is Tagmo, but bin-file-savvy people sometimes use MiDe’s Brain Transplant tool as well if they know what they’re doing.

When brain transplanting an amiibo, everything gets carried over to the second amiibo except the character. This image illustrates it:

In the above image, I’m taking my Mega Man and transplanting him into the Joker. My Mega Man’s spirit loadout, level, name and training data all carry over – the only thing that changes inside the amiibo is the fact that my Mega Man now lives inside a Joker. You could say that it’s almost like I’m transplanting their brain.

After this transplant, my Protoman will now play as Joker, and use the Joker amiibo’s AI, but he’ll do so using the behavior data that he gained while he was a Mega Man.

The Drawbacks

Amiibo brain transplants are, basically, a complete reimagining of what’s possible with an amiibo. Problem is, they typically don’t create a more successful amiibo. The original amiibo has to be fairly well-trained in order to produce a transplant that will also hold its own, and even then it’s not usually a great fit. This means that, nine times out of ten, the resulting amiibo will be a pile of stinky hash browns. This is because different amiibo characters have different AI, so they’ll respond differently to the same behavior data. This image demonstrates it:

But one time out of ten, it’ll be a reimagining of an amiibo that takes advantage of untapped potential.

Fixing the Drawback

While the brain transplant method does have its problems, they aren’t so deep – I’ll give you an example.

I was training my Sephiroth amiibo in preparation for the upcoming Sephiroth amiibo guide video, and wasn’t able to train one I was really proud of. Because Sephiroth’s AI focuses on a few specific moves and behaviors, I wasn’t able to get one to play the way I theorized that it should play. I decided to throw caution to the wind and brain transplant my Mega Man amiibo, Protoman, into the Sephiroth. Protoman’s playstyle is very aggressive and fast, and I figured it would be a good fit for Sephiroth.

The result was as you’d expect: he mostly used Up Tilt and Up Air, some Up Smash, and Back Air for edgeguarding. These were all the moves that Protoman-Mega Man was taught to prioritize, and Protoman-Sephiroth had the same behavior data. Despite this, I wasn’t especially proud of the results.

But I saw that Protoman-Sephiroth used his movement significantly better than the previous Sephiroths, so I sprinkled in some Forward Tilt and Forward Smash, two hallmarks of Sephiroth, and the end result was significantly better than the original Sephiroth amiibo or the Proto-man Sephiroth He started taking games off of amiibo of all tiers, and even entire sets with a little more rejiggering. I wouldn’t have been able to replicate Protoman-Mega Man’s movement behavior on my own, but now I can thanks to the brain transplant.

The Usefulness

In a practical sense, amiibo brain transplants are typically not a great option on their own. They require an amiibo that’s already well-trained, and there’s no guarantee that the transplant will succeed where your previous attempts failed (but most of these problems can be solved by turning Learning back on and retraining for a few short games).

However, brain transplants function as a very useful tool in two different ways:

  1. They make comparisons of different amiibo AI very easy
  2. They can bypass learning restrictions that certain amiibo have

Comparing AI

As you saw in the above picture, Protoman was trained to use Up Tilt and Up Air – when he was originally trained as a Mega Man, this made sense. The Mega Man AI is programmed to be very well-rounded and use most of its moveset regardless of its training, due to Mega Man’s nature as a complicated character. You could say that the AI is weighted in favor of using several moves in its moveset.

Joker, however, is programmed a different way. Joker’s AI is also heavily weighted, but unlike Mega Man, he’s weighted in favor of using Up Air and Up Tilt, and not the rest of his moveset. The Protoman-Joker transplant would produce a Joker who relies on Up Air and Up Tilt significantly more than Protoman-Mega Man ever did, as a result of Joker’s AI being weighted differently.

In other words, we can compare the Joker and Mega Man AI because we now have a controlled and identical set of behavior data to benchmark the amiibo against.

(Let’s also not forget that it’s just super fun to make a pretty new amiibo out of nothing.)

Bypassing Learning Restrictions

When I was training Pyra and Mythra for the Pythra amiibo guide video, I was having a real issue – I had trained three amiibo by that point, and while Pyra would use Blazing End, Mythra would never use her Side Special. I’m still not sure why that’s the case, but other trainers proved to me that it was possible for Mythra to use that move.

In order to test this theory, one of the other trainers took an optimally-trained Incineroar (who revolves entirely around Alolan Whip and only uses their Side Special) and transplanted that Incineroar into their Pyra/Mythra amiibo. The resulting Mythra amiibo did indeed use Side Special, disproving my theory.

It seems to me that, for some reason, Mythra has a harder time learning to use her Side Special, but the brain transplant proved that it was possible given a high enough focus on using it. While training a Mythra to exclusively use her Side Special in training would probably succeed in getting her to use it, it wouldn’t produce a very successful amiibo because ensuring that the Mythra uses it would require it to focus too much on Side Special.

So in this case, we’ve proved that Mythra could use Side Special, but (in theory) one would need to start a Mythra amiibo off with a brain transplant in order to have a Mythra that uses Side Special appropriately.

In Conclusion,

Amiibo brain transplants are an untapped corner of the amiibo meta – while they won’t be taking the place of traditional amiibo training methods, they do enable anyone to take the outside corner when it comes to producing the most capable and well-researched amiibo possible.

1 Comment

  1. OH! I never knew this tool existed. I was slumming it with that older Python based program. I can finally get to using my old Spirit AI files.

    Like

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s