Haven’t seen the part yet? You can watch it on Youtube.
On Jan 31st of 2024, I got a message from MajorMilk, inviting me to a new collaboration that was described as the spiritual successor to NicoNico Skyscraper. We finished our Non Non Biyori part in late December, so I was not expecting our next rodeo to happen this soon. After some consideration, I ended up accepting.
I wish I could more accurately recount what exactly was happening in my life during that time, but my memory fails me, going from my own messages, I was apparently not in that much of a rush. I did have a few worries about how busy I could get in the future (with the place I used to intern at going bankrupt and stuff).
Because our experience in the NNB collab was still fresh, I factored it as much as I could throughout the lifecycle of the part. Making the audio for that collab wasn’t a bad experience per se, but it could have been more fun. I remember having to come up with a story that fit the Dragon Quest theme and skim through the whole thing many times to find fitting voicelines, it was pretty unpleasant. So I wanted to go a bit easier on the storytelling this time.
It’s with this mindset that I accepted the invitation and Joined the M2 server.
Chapters
A few words from MajorMilk
I always jump at the opportunity to collaborate with my good friend CH FR. So when I got an invitation from Tomoshige to participate in M2, a spiritual successor to the beloved Nico Skyscraper collab that got me interested in making MAD videos in the first place, I swiftly said yes. I convinced the organizers to accept CH as my audio partner and we got to work.
We had a long list of ideas we wanted to implement into the visuals. Unfortunately, real life work and my packed end of the year schedule kept us from implementing most of them. One of them included shoving every anime we liked (Yuyushiki, Hidamari Sketch, GA Art Class) and using rigged 3D models of the cast to do cool things. The other one was frame by frame animation like the ones I’ve seen for Blue Archive. Eventually we simplified the part and stuck to GA with hand drawn animations.
draft of a potential rig that was scrapped during production
M2: Our part
Without further ado, let’go straight to how this part came to be.
Song selection and initial storyline
After I joined, the first thing we had to do with was to agree on a song. That’s a step I really try not to mess up, so I listened to each song that MajorMilk suggested over and over as I imagine what the part would look like (and subsequently, what kind of audio I would make).
We agreed on using Ukigumo.
I didn’t want it to be too story oriented, my initial vision was to have an explosive multisource extravaganza with all of the stuff we like. Part of why we hit it off and became good friends was our similar tastes in anime, Milk is basically the reason I got into stuff like Hidamari Sketch, Kirara manga, PPD and SZS. The vision I had back then a part that would scream “this is the CH FR and Majormilk” part to anyone who knew us.
Here is part of my brainstorming, (please keep in mind that we didn’t even know if the song would make it in, nor what part of the song would make it in at that point).
(The section of the song in question)
With that vision in mind, we asked the organizers to get this specific part of the song, they were receptive and we got almost all of the requested time range, except for the buildup before the chorus, and a few extra measures at the end.
Over time, we kept talking about the part, eventually sticking to just GA. Out of all the source picks, GA was the one I wanted to work with the most (I already wrote about how much I like GA). The goal had become to make a good GA part that would be immortalized in this collab (Skyscraper had a pretty good one as well).
One of Milk’s idea that I ended up shooting down was to make the part a crossover of GA and Yggdra Union (a game that Satoko Kiyuzuki did the character designs for). I couldn’t find the will to finish the game, nor dig through its sound files.
Skip a few months, and at the end of July, the draft of the medley was done and we got our dedicated channel on the collab server.
It was finally time to start working on the audio.
Audio
That’s something I often say, so you may already be aware, but the transition between what’s in my mind and what ends up on my reaper project is always a pretty brutal one. It’s the stage where a lot of my ideas end up dying due to my lack of ability to wrangle the source material to tell the story I want.
My first draft was based on Ukigumo’s video PV. The first half being kisaragi running around, with giant art supplies falling from the sky based on the sentencing, and the second half being a quick succession of one-liners from the show’s supporting characters.
We were still figuring out the overall direction of the part, wherever it would be Yggdra-centered (I hadn’t fully rejected the idea yet), or Ukigumo-centered.
I later changed the second half to use Kisaragi’s monologue about cutting out a piece of the sky..
The timing was still a bit rough, but I felt more confident with this scene..
I really liked the idea of “clouds clearing up”, as it drew a direct parallel between the video PV and Kisaragi’s desire to paint beautiful blue skies. This idea unfortunately didn’t materialize.
Once I had the second half figured out, I went and redid the first half entirely.
Honestly, Tomokane is my least favorite girl of the group, but the scene of her showoff against Awara (in a dream) was very easy to twist into something resembling a battle, so this is what I ended up doing.
It’s important to know that, during that time, MajorMilk was also trying to do as much work as possible in advance, as he would be going to Japan by the end of 2024. However, I could only give lukewarm answers, as my only way to see if an idea would stay would be to test it first. As a result, I had to tell him to commit as little work as possible, I was afraid of having to change everything and pull the rug under him if I failed to pull off some of my ideas on audio.
For this part, I decided to be a bit more strict on the ideas that I would allow, so that making the audio could be more fun than for the Non Non Biyori part, but while it indeed was more fun, it also meant that my visual partner would be completely blind for as long as my own vision wasn’t concrete. I’m sorry and thankful to MajorMilk for bearing with this uncertain workflow during a time where he also had a lot on his plate, such as working on other videos and preparing for Comiket.
But finally, things started to take shape, I had the start and the end of the part pinned down, now to do the middle one.
I combed through Ukigumo Otomads in order to find some inspiration, but none of them leaned into sentence-mixing that much. That’s when I stumbled upon a video on NicoNico from いび, titled I sung Ukigumo on a whim, its description is as follows:
Nice to meet you. sm4483132 left a deep impression on me, so I tried to sing lyrics for it on a whim.
Other users joined along and made their own covers, using the same lyrics, いび even uploaded a cover of the full song a few months later.
I found this whole trend very cute, and the idea of those becoming the “unofficial Ukigumo lyrics” stood for everything I love about the internet and derivative works. So I decided to pay tribute to this video for the middle-third of the part.
As for the execution, it was the same as always, grab syllables from character songs, pitch, VocalShifter, and we’re done.
Just like I did for the Non Non Biyori part, the “singing in unison” is not done by taking multiple voices, but by using the ending song of episode 12 where the girls all sing together, and pitching that.
Covering the other instruments was business as usual, I won’t dive too into it because I’m not really qualified for it (there’s a reason why I always start with sentence-mixing). I guess the only noteworthy thing I did was to use the drum patterns from the original song during the monologue section.
Because the sentence mixing was about Tomokane and Awara fighting, I wanted to add sfx. At the time of making this audio, NicoNico had only recently come back on its own feet after the hack, so I figured I could try to browse NicoNico commons for the first time. The site turned out to be a treasure trove of useful sound effects that made the audio feel much more alive.
Here are all the sfx I ended up using:
nc137535_【SE】落雷【ピシャーン】
nc278648_【効果音ラボ】可愛く輝く1（魔法の音にも使える）
nc278649_【効果音ラボ】可愛く輝く2（アイドルがポーズを取る）
nc278951_【効果音ラボ】ひらめく2（電球がピカーン）
nc278965_【効果音ラボ】武器をクルクル回す（風切り音）
nc283144_聖なるオーラ・回復キラキラ
nc293152_音素材No.09（ショック・ドーン）
nc314546_ビリっと、バリっとショートする電撃音（短め）
nc355977_効果音_キラーン
nc363898_カーテンの開閉
nc367386_ブシュッ！！_（連打）ソードで斬る連続打撃音_０２
nc374907_ビーム・ライフル（ジェガン）
nc382671_【効果音】ガラスが割れる音_その1
nc383635_【効果音】ビーム砲その2・SF
nc383636_【効果音】ビーム砲その3・SF
nc390020_【効果音】剣を振る・その4
nc390021_【効果音】剣を振る・その5
nc85019_【効果音ラボ】落雷_その２
Yada yada mixing mastering blah blah blah here’s the final audio.
No BGM
With BGM
As an aside, R.M. did an awesome job in adding the little sub-melody (the plucks you can hear in the background) to this part of the medley, it really helped in making the part sound more emotional! Here it is overlaid with the midi.
Video
Now you might be wondering, did I put all those SFX blindly with just a vague idea in mind. Truth is I didn’t, there is one recording that happened much earlier during audio production: the initial storyboard of my half of the visuals.
Remember back in January when I tried to post one drawing every day on Twitter? Well there were a few days in a row where I posted something along the lines of “I did draw today, trust me bro!”
And remember February when I kept doing that challenge, but focused on backgrounds and had another two blank days?
Well this is what I was working on:
Now I can finally show the proof that I didn’t break my drawing streak!
About that… I stopped drawing daily mid-March because I started drawing using Blender’s grease pencil, and wouldn’t have been able to learn it in-depth while continuing this challenge. But grease-pencil shenanigans are a story for another time.
Anyways, after drawing all the characters (the backgrounds only came in later), I started with something rough, placing all the characters across the timeline before doing multiple passes of polish.
I would like to say that working with Autograph was fun and smooth sailing, but that would be a lie.
While the core functionalities allowed me to accomplish pretty much everything I had in mind, I once again encountered a plethora of crashes and suboptimal design choices that made working on the part a lot more painful.
I dutifully reported all of the bugs I found, and got some mouth-watering messages telling me how much better things would be with the new update that was just around the corner.
But the deadline was in one month, I didn’t have time to wait and hope that LeftAngle would deliver by then.
Even when it comes to bug reports, it’s a major pain to have to stop what you’re doing, create a minimum-reproducible-example, zip-up the project, and email it to the team. But I still did it because the bugs I encountered were breaking all of my momentum either way.
The worst thing that happened to me was when, at one point, the program crashed (I’ll spare you the details, had to do with adjustment layers freaking out and getting a huge buffer size when paused and hovered), except the project autosaved exactly at the point where it crashed. Preventing me from even opening it.
I thought that it was a Linux issue when MajorMilk told me that it worked on his machine, launching the project on windows seemed to confirm that. But as I looked further into it, it was not an issue with my OS, the reason windows fixed it was because it unlinked all of my imported media (with the paths being different and shit), and my images were one of the conditions for the crash. This allowed me to fix my project file and then remove the crash condition after I identified it, and finally relink every drawing I made.
All of that to say, can you imagine your project just refusing to open when the deadline is close? I’m all for Autograph becoming a big player in the motion design industry, but good god does it need a LOT more polish before its user-base grows to a large size!
Anyways, more stuff happened and I finished my half.
Bridging the gap
As you’ve seen, I’ve worked with MajorMilk differently this time around, during the Non Non Biyori collab, he did pretty much all of the visuals, while I did the storyboarding and helped with menial coloring task.
This time around, we did one half each, Milk was largely responsible for ideation in his section, coming up with the Kisaragi Falling and angel scenes. I chimed in when it came to bridging the gap between our halves, Tomokane, Noda and Awara were turned into Kisaragi’s small familiars. I threw around ideas such as that earlier feud being them just play fighting, but I wish I were able to link those sections with a more cohesive narrative.
Looking back
I like this part, I really do! I think that MajorMilk and I once again did a good job together and I hope that despite it all, he had as much fun as I did.
But as 2024 went on, and we kept working on the part, I guess I forgot all of the compromises we had to make along the way. I am being sincere, up until a few hours ago, I completely forgot about this message that you saw at the beginning of the article:
Reading it all over again gave me a weird feeling. I really AM proud of the part that we made, we poured our skills and our passion into it like always, and yet, there’s a part of me that also wonders what that original storyboard would have looked like. As I write this article, going back over the whole process, I am seeing a story of “scoping down”. Because Milk and I had our own circumstances irl, we had to put a lid on our ambitions, cutting corners here and there in order to deliver something by the deadline.
I think we made a good collab part, but when I read the storyboard, I saw extinguished ambitions of making a part that celebrated our friendship.
I’m sorry. I know that in the future, I’ll be able to look back more positively on this part, but right now, the feeling of pride and accomplishment is accompanied by a bittersweet aftertaste. I didn’t plan for that when I started writing this. Ending it there would have been a bit morose, but thankfully, there’s still one thing that this article has yet to touch on.
P.S. Went to bed right after writing this, I’m feeling better now!
M2: The collab
Now let’s talk a bit about the collab itself, and the way it was organized. M2 was semi-private in that participants could also invite their own friends (which is how MajorMilk got me in). Despite that, we were encouraged to keep its existence a secret until the very end, and considering how many participants there were, I am impressed that everyone was (mostly) able to keep it under wraps up until the official announcement!
The first thing we had to do as participants was to give a long list of BGMs that we wanted to work on, as well as the source we wanted to use. The organizers made a point of not allowing too many repeating materials and avoid doing too much multi-source, this resulted in a collab that manages to stay fresh for an entire hour (sure there were quite a few train parts, but not as much as you’d expect for something this long). MajorMilk and I submitted many songs, but at the end of the day, we personally went to the organizers and made it clear that Ukigumo was our first choice by a large margin, thanks again for accommodating us!
When R.M. started doing announcements about drafting the medley, I was shocked to learn that he was the sole medley-maker for the entire project. I looked nervously at the never-ending participant list, feeling second-hand dizziness at the insane amount of work he had ahead of him.
You may have noticed that some of my previews used the medley while others had the original songs, this is because the organizers were very forward-thinking with drafting the medley, and actively encouraged the audio-makers to start work on their part as soon as their section of the draft was set in stone. We were told that things such as melodies and sentence-mixing were OK, but to avoid covering elements that could change in the final medley, such as drum patterns and chords.
The draft was completed at the end of July.
TLDR:
You can start planning and working on your audio, but some aspects of the medley can still change.
Watch out for voice timing, drums, and YTPMV covers as those are the most sensitive to changes made on the medley.
Medley schedule: First version with temporary MIDI by September, then tweaks and fixes, and completed Medley+MIDI by november
Special variations for those who requested them (for example, a part without drum patterns) will be delivered after the medley is complete
After that, R.M. got to work, there would rarely be a day where I checked the server and did not see him toiling away in one of the voice channels. I rarely ever joined the server’s voice chats (for obvious linguistic reasons), but would still spot him there more often than not.
As this went on, the server got regular updates, with R.M. digging at it little by little, and notifying participants after every work session.
Things ran a little bit late, but thanks to the regular updates, more than 84% of the participants had their part covered by the medley by mid-October.
our section was delivered on 2024-10-13, I was staying at a hotel for a few days before I could do the final visit at my new place. I remember taking a walk around as I listened to part 120 to 135 of the medley.
Lo-and-behold, R.M. had actually done it, on November 21st, the medley was fully completed and polished (save for #161-Ultra-trailer, which he would be working out with the host Namacream).
With that, we had a medley.
Most of the organization was done using discord, with links to external resources and spreadsheets.
You would think that a 161-parts-long medley would be absolute hell to navigate, well… maybe it was for the organizers, but as participants, things went surprisingly smoothly, each part had one specific channel, whom only the designated participant could see, overall, this server actually had less visible channels than many of the casual servers I’m in.
We can get a glimpse of the higher-order-organization:
Channels were batched in groups of 15, and it seemed like each group had specific managers that they could report to.
We had specific roles that we needed to ping for notifying of our audio/visuals submissions. I don’t know how severe the progress checks were, or if there were even any, because the organizers or the managers didn’t bug us about any of it. I try to not be a burden on collab hosts, by making my progress known and frequently sending previews whenever I’m doing progress, so I assume that’s why there was very little friction with the organizers in that aspect. As of today, our designated channel has 640 messages in total.
There were also audio compile and video compile channels, to ensure that participants were all OK with how their part would sound and look like in the final product.
The final part
The collaboration starts and ends with Ultra Trailer, namacream wanted to do the same kind of finale that he did for Skyscraper, so the channel for his part was made public, and we were all asked to provide some audio-sources and an image that we would want to show up on screen for it (I guess that’s why multi-source was discouraged).
The completion of this went just like the completion of the medley. namacream crunched just as hard, videosharing on voice chat almost every day, and sending very frequent progress updates. We were able to hear the audio get fuller and fuller each time, and final result was just as good as I expected.
Hyping things up
As previously mentioned, participants were told to keep this collab a secret until the very end.
The plan was to sneakily show the medley in a NicoNico livestream, and only later would it be announced that the medley is tied to a collaboration.
There was also a private IRL screening that was organized for the participants that were able to show up (too bad I couldn’t just take a few days off and book a flight there, but we got to see a recording of the whole thing).
For those who couldn’t attend, there was also a watchalong on discord, I happened to be working remotely that day, so I was able to just grab my laptop and took a little break to join the voice chat (don’t tell my employer). I got to see the completed thing for the first time, and it felt like less than an hour had passed once we reached the final part.
Things were truly made “official” when namacream_ tweeted a trailer. Like many, I was surprised at the lengths they went to to hype people up, I wasn’t aware that each participant would get their own stylized name tag, so that was a nice little surprise.
How to make a 1-hour-long collab
5 years ago, as I participated in the F-side of ytpmv, I thought that 20 minutes was pushing it a little in terms of length. But as time went on, we started to see more of those long-form collabs, and then VERY long form collabs such as Kyoro’s Toshinose series, the final iteration of which has a total runtime of two whole hours! But even then, the free-form nature of that collaboration could explain the runtime, as anyone was just allowed to let their part drag out forever, do a relay on the same damn Pokémon BGM for 7 whole minutes, or even organize sub-collabs within the collab.
What I hadn’t seen before though, was a classic 1-hour-long collaboration like M2. The most insane thing to me was that, in the end, it released on time. Hopefully, everything that I showed in this article can show that this was not just dumb luck, the medley crunch, the meticulous organization of the server, the clear and frequent communication, all of these played a part in ensuring that such a huge project could be delivered on time.
I am at awe that this collaboration exist, and that it went as smoothly as it did.
“Collab prestige”
People in the community often tense up at all of the implications behind the ideas of “big collabs” but I feel like in M2’s case, the fact that it was a celebration of the 200th Otomad Sarashi, and that it was the successor of Skyscraper played in its favor. I know that this part of Otomad culture can have many pitfalls, but there, it helped create one crucial element: drive.
When I look at the collab, I really get the feeling that everyone wanted to make the best part they could, the parts were all very diverse and unique, and this didn’t come at the expense of humor either. Even though I know that people won’t watch a hour-long-collab that many times, and that my part is just a grain of sand in the final product, knowing that I was working for “the M2 collab, successor of Skyscraper” was motivating enough to look past those things, and still put my all into it. And also seeing key players like R.M. and namacream put so much effort in their work was, in itself, motivating. At least that’s how I felt.
A few words from Namacream
ご共有してくださりありがとうございます。 添付していただいたpdfを拝見しました。
文章から二人の熱意が身に沁みて私は感動しています。 私の最後のパートは、CHFRさん,MajorMilkさんのパート含め 全ての方々に支えられて完成したものだと思っているので、 皆様には頭が上がりません。
感無量です。
改めて参加してくださり、 本当にありがとうございました。
Thanks for sharing. [I] looked at the pdf you gave me.
[I’m] touched by the enthusiasm of both of you as it sinks in from the text. My last part was completed with the support of all, including CHFR, MajorMilk’s part. I can’t stop thinking about you all so moved.
Once again, thank you very much for your participation.
Conclusion
All in all, I enjoyed being in this collaboration, it allowed me to work with MajorMilk again, and while we had to cut back a lot, we still made a result that I believe is within the quality standards of such a collaboration.
I was also able to witness what it takes to organize and pull-off something at such a huge case, and am very grateful for the experience as a participant.
I’m thankful to everyone involved in this project, and will see you in the next article.
Still though, one hour, I can’t imagine there being other collaborations on this medley. No matter how you slice it, that would be very difficult to even gather the amount of participants needed to pull this off…
Let’s check Twitter, oh hi Theas
I’m watching the premiere right now. I can’t believe it.
So as it turns out, that whole saga where Theas uploaded for more than 100 days in a row was… in preparation for this!?
That’s absolutely crazy, I could have noticed that he was preparing something like this if I had paid more attention to the medley, but I actually avoided listening to the full thing so that it wouldn’t spoil my experience. I could have never prepared myself for a stunt like this, I have no word for what just happened lol.
The premiere just ended, that was incredible. Man, Theas is awesome! Even as a participant, I would have never guessed that he was picking the songs for his personal uploads and other collab parts based on the medley. I don’t know if any other participant was able to figure it out before the fact, but if they did, then they’ll have one hell of a story to tell.
Anyways, article over, for real this time.