Sweet Regret

Last edited: 18 November 2016, 11:07PM

Info

Sweet Regret was an originally produced vocaloid song made for osu!. The map was made by a group of individuals from osu! UCI as a summer project. The summer themed song was about GUMI overeating too many sweets over the course of day and, as one could guess, regretting her decision.

Background

"It was alright." - Royce Sato

After Silicon Heart ended, the team was pretty motivated to work on a new project. I was a little sad that Silicon Heart was never fully finished and polished, at least on the beatmapping side, but with the quarter ending, we wanted to move onto other things. We had quite a few ambitious ideas we wanted to pursue.

One of them was a vocaloid song. This was not something we could easily do during the quarter because fitting all the different pieces together seemed too difficult. Just imagining the process behind thinking of the song idea, writing lyrics, producing the music, adding in the vocaloid, drawing the background, making a skin, programming a storyboard, animating a video, mapping the song, among other things, and trying to fit all these parts cohesively together was too much to handle.

But school was ending and with summer coming our way, we thought the couple months would be enough time for us to commit to a vocaloid project. Why vocaloid, one might ask, out of all the other potential options. The short answer was that half of us were weebs. I listened to a ton of utaite and vocaloid music, and most of the others had their fair share of exposure as well. None of us had actual vocaloid experience, but we were willing to give it a shot.

Silicon Heart was a good song, don't get me wrong, but it was an instrumental. If we knew people gifted in singing, recording, or playing instruments, we would have loved to grab them on board to record something with us, but my only musical contact was Wes. Vocaloid was an attempt to bridge that gap and make a song more than just an instrumental without having to master an entire musical form overnight.

I also got in contact with Nick, who was a friend of mine from the Video Game Development Club at UCI. This club was also where I met Wes. From my occasional talks with Nick, I found out he had quit a bit of unusual experience with MikuMikuDance, a 3D animation program that has a distinct anime stylization.

I say unusual because having knowledge of the program was already rare, but his reasons for using it were a bit odd as well. He did not start using the program, if I recall, due to vocaloid fandom, where it's most commonly attributed for. Instead he picked it up as a way to get into animation, and he liked the cute style and models used in the program.

And to add more onto using vocaloid, Royce was Japanese and was studying the language, and he thought it'd be pretty neat to write a Japanese song. One of the things he adamantly fought for was using Japanese over English because English vocaloid was "unbearable" in his eyes. I'm pretty impartial to the whole Japanese vs English debate, but I can see where he's coming from.

Only doing Japanese definitely made the project more complicated since not all of us knew the language. The only other member with some experience with it was Justin. It didn't make everyone's job harder since mappers, for example, didn't need to worry about it at all. But for me, who was closely working with the lyrics for the storyboard, it made things several folds harder.

With a few guidelines in place above, I planned an in person meeting with as many of us as soon as possible. The first meeting took place right before finals so mostly everyone was available to make it. This would probably be the only time the group would meet together officially since after that summer would rip everyone apart from school. Here're our beautiful white board notes that apparently stayed there for over a year.

Our first meeting went extremely well. Justin, Royce, Wesley, Nick, and I were the ones that I believe showed up, and it was good that all of us with our different roles were present. Most crucially we were able to set the groundwork for what our song was going to be about, and from there we made a rough schedule to plan our future tasks.

The first task was to decide what the song was about. We spent a long time brainstorming ideas and trying to narrow down our interests to something we all liked. I think I actually learned a lot from this session about creating ideas. The big take away was that general and vague ideas don't help a lot. It's best to shoot in the dark for very specific though possibly unrelated or stupid ideas and then work from there.

We decided it was going to be summer themed due to the season, and we chose GUMI because she seemed the most appropriate vocaloid for the season. We wanted the song to be faster and more upbeat, which would make the song more playable and mappable in osu!. To go along with the tempo, we tried to think of a lighthearted story.

After throwing some ideas out, we eventually landed on the idea of sweets: GUMI overindulging in some desserts and eventually regretting her decision. We thought of some sweets people generally enjoy in the hot weather, namely candies, ice cream, shaved ice, and pudding. Each of the verses talked about eating one of these treats, and the chorus talks all about the regret.

At one point we added in Miku into the mix. She was the stark contrast to GUMI. Miku happily ate as many sweets as she liked without any problems and tempted GUMI to eat with her. Her role may not be that noticeable in the song due to lyric-fitting issues and not wanting to add in a second vocaloid, so we don't have any lines directly from her. Nevertheless, her presence was implied.

We wrote some rough lines and notes for the lyrics to get us started, and with our meeting dragging on quite long now, we decided to finish the lyrics another time. But as the last item of our agenda, I wanted to plot out and milestone our objectives going forward. There were a lot of different parts and roles to this project, and we need to get on the same page quickly so we can work in tandem with one another.

Music was going to be the most important. Like what we did in Silicon Heart, Wes would need to get us a rough version of the song as soon as possible so that the rest of the members could work. With the song, mappers could then start mapping the notes, and importantly for this project, Nick could then also get started with his animation.

For now, the mappers couldn't do anything, but we spent this down time looking for a few additional mappers. At the start, only Justin was interested in doing mapping, and we could use a few more to contribute. We were pretty much looking for anyone to help out without restrictions. Personally, I wasn't planning on getting this song officially ranked. This project for me was just something fun to do over summer, and we had a quite a few inexperienced members. If the group wanted to get the song ranked, that would mostly be up to the map makers.

And to touch on the ranking some more, the process as far as I know takes quite a bit of work. It requires going out and actively looking for people to mod your maps and then hoping/asking for nominations from beatmap nominators. I've never fully tried ranking a map before so I can't say too much about it. I'm neither a good mapper nor an active one, so I'm not involved with the process.

From what I have glimpsed of though, I can imagine that it's a lengthy and consuming process that requires a significantly more amount of effort than just making something and leaving it. It takes time to gather feedback, make changes, and find approval. This wasn't something I particularly wanted to get into, and, as this project would end up, not something we'd have time to get into.

For animation, Nick would be scouting out models, props, and settings for us while he waited for the song to finish. He was somewhat worried about the length of the song since up until now he only did shorter animations. His concerns would come to light later in the project, but for now we said that it would be fine for him to think about copying animation sections like the chorus. Worst comes to worst, we could cut the song and remove some repetitive sections in favor of better animation.

The biggest question was what to do with vocaloid. For a while, we were desperately looking for someone to fill the role. I asked around to see if anyone knew how to use the program, but the mere one or two people I knew with some experience were busy or uninterested. The others weren't helpful and didn't have any vocaloid contacts either.

In the end, we gave ourselves a two or three week final ultimatum to find someone for the role. This little buffer room was for the most part pretty safe because we couldn't work with the vocaloid until the music was finished in the first place. Our ultimatum involved several parts. One part was self research. Both Justin and Royce, who had some Japanese knowledge, would look into the program and try using it for themselves. We would also look online to try and find someone to help us out.

Things worked for us. As they experimented with it, it turned out that using the program wasn't as difficult as we thought. Granted, they probably wouldn't be able to churn out a master, master piece, but we could quite easily get some semblance of vocaloid into our song without too much difficulty.

We decided to let Justin lay down the base notes for the vocaloid and then have Royce do the tuning later. After working with the program for a bit, however, Justin resigned his post and passed his job onto Royce, citing that Royce would probably do a better job since he knew the language better. Although apprehensive at first, Royce took up the mantle and completed the task.

Originally, we didn't want to overload Royce with work because he still had lyric responsibilities left to do. For the first couple weeks, our weekly meetings were spent writing out the lyrics in English on a private Google doc. Royce then went in and translated all the lines to Japanese. Additionally, because we all suck at Japanese, Royce had to annotate and write notes on the lines to indicate how many syllables each line had. This was to mainly help music so that Wes could arrange it properly.

The way music was going to work now was that Wes was going to add in a distinct sounding synth instrument to indicate where the vocaloid should speak. This would make placing the vocaloid vocals easier since the placer only need to directly follow the vocaloid melodies rather than planning the notes by scratch.

We did run into a few issues with music though. Our criteria for the song was not something up Wes' alley, and he expressed some troubles and concerns with that. Usually, Wes wrote some slower songs at around 135 BPM, but this slower speed was not the norm in osu!, at least on the higher ranked end. Faster songs were more difficult to play and easier to map.

We wanted him to shoot for a song around the 180 BPM range. He struggled for a while to give us something, in the end, he came up with roughly what you hear now. If you listen to it, however, it feels more like 90 BPM than 180. I guess the lesson here is that you shouldn't pressure people too much to push outside their comfort zones, especially if everyone's on a time deadline. At this point, it was difficult to adjust his song since we were running close to deadlines.

With the rough song complete, it was time to task more roles out. Lyrics were finished at this point, so Royce could start working on vocaloid. Nick got started with animation with the rough copy, and he would add in lip sync later. The mappers now had something to work with, and they could get started now.

During the down time waiting for the music, I was able to pick up a three mappers who were interested in helping out. They weren't extremely experienced, but I wasn't picky. We met most of them through the osu! North America Facebook group. I made a post looking for interested mappers, and we found two there.

Langston did a standard map set. He hadn't previously gotten a map ranked yet, but he worked pretty fast and was pretty active in our discussions. It turned out he was also going to be joining osu! UCI next fall quarter as an incoming freshman, so it was good to see him show up early for this project.

Josh, oddly enough, was a mania mapper. He was a little more difficult to get in touch with, and he did somewhat fall off towards the end of the project. Despite that he still finished two maps for us, and I'm thankful for that. I tried to get him on board to help do storyboarding, but he didn't get a chance to help out amidst other commitments and interests.

And lastly was Andy, who wasn't from the osu! North America group but was a friend of a friend in osu! UCI. His background was a little weird since his actual osu! account was banned due to some stupid things going on. I actually somehow had dinner with him in person when he visited Irvine. He's kind of a quiet, funny guy.

I can't say too much about him since he actually didn't end up making us a map. He told us early on that he would work on one, but by the summer's end he didn't have anything to show. It's summer, so things like this I expected to happen here and there. By the end, we still had a half dozen or so maps, so one being lost to the elements wasn't worth harping on.

I guess it's time to go into my roles. I played several hats in the group, one of which, as should be seen above, was the main organizer and leader of the project. I arranged all our meetings and was present at pretty much all them. We used this time as a work session, as well as a way for me to check in with members and evaluate progress. I updated all our Drive and Facebook pages to keep everyone up to date on what's going on and what we needed in the future.

There was a lot of pipe-lining and parallel work going on, so planning things properly was essential. If one crucial part fell behind, that could affect everyone else, and we may have to to replan or adjust our milestones. For the most part, I'm happy with my efforts on the leadership end. It's not easy organizing online efforts, and I think I did an alright job pushing people to complete their tasks.

By the end we had a complete product, though perhaps not the most polished one. I think I definitely should have cut things shorter and left more room for iteration and improvements. Some of our later problems probably could have been avoided if I was more strict with my decision making. These details I'll mention later on as we near the end of the project.

But let's get into my other parts I was assigned. The biggest one early on and which I spent probably a good month or more on was the background illustration. Different from my other beatmap projects, I would be responsible for every part of the drawing, from sketch to finish.

When I started working on it, I took a lot of inspirations from different sources. The setting of the picture was dictated mostly by the song. The first verse talked about how GUMI wanted to eat some cool pudding in the hot morning while waiting for Miku to arrive. The eventual concept I landed on was having GUMI look wistfully into her refrigerator at some pudding.

GUMI herself was inspired from several sources. Of course, there was the original character that I referenced a lot off of, but I took a lot of liberties. I could have chosen to use her traditional costume, but I decided to throw that out and go with something more hip and modern and deck her out in accessories.

Much of this came from my love of a particular illustration that I came across on DeviantART. This picture was Your Glance, by dark134. Not only did I really love the soft coloring and blending, but I found the girl particularly attractive with all her sick accessories, and I wanted to emulate the same style.

So I kept her color palette (green, orange, and pink) and kept her short hair but just went wild with everything else. I guess if you wanted to make it a stretch, you could say I also kept her signature red goggles as well, just now in the form of glasses and a little pinker.

You might be wondering why I don't have the soft blending style like Your Glance. Part of the reasoning was that I just wasn't that good at it. It's not something I have too much practice on, and since I was already spending so much time on this drawing already, I didn't want to drag my work out any further. The other reason, and probably why I didn't have a lot of blending practice, was because I just really liked heavy black shading and thick lines.

Inspired by a lot of manga, a bit of traditional comics, some of my favorite artists, and a few anime that adopt this style, I really love the pure black shading style. I don't know if my preference is something I can put easily into words. It's just something that I think is super cool and that I've been trying to pursue for a while now.

I didn't notice it at first, but this version of GUMI actually takes a lot from one of my favorite anime of all time, and not to mention one of the few anime that has that black shading style. She reminds me a lot of Sonoshee McLaren from Redline. It's kind of funny actually because I didn't notice the resemblance at first until one of my friends remarked about it. I think it's probably just the hair, which is quite warped from the traditional GUMI style, but you could include the girl's bust and fashionable nature as additional links I suppose.

After working on the picture for a while, I ran into an issue where I wanted to fill the refrigerator with some items, but I didn't know what to put in there. I put in a few generic objects in there like ketchup, mustard, barbecue sauce, pickles, and milk, but there was still quite a bit of room left to fill.

That's when I stumbled upon a page talking about vocaloid items, items commonly attributed to certain vocaloids. I was originally going through this list to see what items fit GUMI, but after coming across a wide variety of them, I tried to fit as many as I could into the refrigerator. The octopus, leek, orange, carrot, and alcohol were taken from here.

I actually could have spent more time on the picture, but I probably already spent a month on it already, and I wanted to wrap it up. I worked pretty hard on the black shading, but after that I went very simple and easy with the coloring and highlights. The coloring was just one shade of color, and the highlights were very light patches of white gradients here and there.

I think the overall result was okay. Some things lacked detail and looked a lot more simple than they should, for example, the beer bottle and sauces. There're also a few composition problems. Most of the emphasis of this picture lies in the left half, focusing on GUMI and the pudding. The right side is pretty bland and doesn't really have a reason to be there.

Because of the way I colored things, everything in the picture is pretty bright and bold. This takes some focus away from the main points of the picture: GUMI and the pudding. You might not even notice the pudding at all on first viewing. I don't know if this was necessarily a bad thing. I played around with some of the colors to be more washed out, but the picture seemed a lot less vibrant. I prefer it the way it is.

And lastly was a problem with tone. By this I mean that I don't think this drawing fits the song very well. Sweet Regret was a funny, fun, upbeat song about the foolish GUMI making mistakes overeating sweets. This more serious and sexy GUMI in this picture though doesn't pair up with the song very well. The coloring and shading are partly to blame.

But regardless, this was the style I liked, and this was what I envisioned in the beginning, and I'm happy with how it turned out. Nick really loved the picture too, and you can see it making a small cameo within some parts of the animation. I actually told him that I didn't think this was a great idea since the picture synced up even poorer with the cutesy animation, but he insisted that it was good.

That's pretty much it for the picture. It was a long, arduous process that took many nights to complete. Every day I got back from work, I tried to spend some time working on it. I'm a really slow worker when it comes to making full complete illustrations. I don't know if it's because I'm extremely meticulous, if I'm not doing something right, or if I'm just not good enough.

The very lopsided time/effort to results ratio was one of the reasons why I decided to step away from art and focus my time elsewhere. I just can't handle the time consumption that art drains from me. For your pleasure and mine, here is a progress collection of this picture. I frequently sent updates to my team and saved separate PSD files to capture snapshots of my work.

Once the image was finished, however, my work was hardly done. It was time to move onto storyboarding. From working on Silicon Heart, I had a general idea of how storyboarding worked, although at the time I was using the 3rd party Storyboard Generation Language made by someone else to help me with my work. For this project, I chose to do everything from scratch with Python.

There were some constraints with storyboarding. First off, the storyboard shouldn't take precedent over the map. The main focus should be concentrated on Nick's animation. The role of the storyboard then was to do some smaller effects. After brainstorming some ideas, we agreed that the storyboard's biggest role would be to display the song lyrics.

This was a lot more complicated than one might imagine, for a bunch of different reasons that Royce and I would soon discover. We could have done a ton of things to make our job 10x easier, but there was a specific effect I wanted to achieve. Since displaying these lyrics was the only responsibility I needed to cover with storyboarding, I wanted to do a good job of it.

What we wanted to do was display each kanji character as they were pronounced. Whenever GUMI said a word, the character would show up on cue. The first thing we needed to do was generate the actual kanji images that would show on screen. What I could have done was go into Photoshop and draw each character individually, save it, and then use that, but not only would this take a long time, it would be difficult to iterate because I'd have repeat the process again.

Keep in mind, my life would be made much easier if all I had to do was display line by line what GUMI said. This would involve a lot fewer images and timing wouldn't be so difficult. In order to generate the kanji efficiently, I needed to programmatically draw and save the characters. I looked around and eventually got Python Imaging Library to work with my program.

Just outputting the the character by itself was not enough, however. I realized just having a single color text was not a good idea because it would get washed out by its surroundings. So after tinkering around, I decided on doing a double border with GUMI's green and orange colors. This wasn't as simple as it sounded, and if I recall correctly I did some hacky business in order to get that effect.

Holy moly, I just looked at my code right now, and I basically did the craziest solution to the border problem I could think of. So apparently what I did to make the borders was that I would draw a character 360 times radiating around a center point. These radial images were offset by a border width that determined how thick the border should be. After going around in a circle, you now had a thicker border image.

I wasn't experienced with graphics, so this was the best solution I could come up with. Note that you can't just use a larger font for the border because there're always spaces unaccounted for and the scaling is off. While definitely more time consuming, at the very least this radial drawing solution did its job correctly.

Also I think it's worth noting that working with unicode is kind of dirty. It's much easier just working with ASCII English characters. I probably also used a hacky solution here that I don't really understand just to have Japanese characters printed out working. Of course, I should have looked more indepthly about how to properly handle it, but time was running short, so I wanted things done as soon as possible.

Now that I generated the images, the next task was displaying them at the correct time. This almost came down to having to specify the exact time to display each character, but luckily I was able to bypass that by using the vocaloid VSQX file. This file was generated from a vocaloid program to indicate when, what, and how the vocaloid should speak.

Most important for me was the timing section of the VSQX file. It used an internal timing system to designate when the vocaloid should say something. I parsed in all these values, and after fiddling with the timing and multiplying it by some 180 BPM against some other random value, I was able to sync the two up properly.

But when I started to time and test the lyric images in game, I immediately ran into some issues. The problem was that while I had the timing correct for when GUMI said the words, it turned out that a one syllable sound could actually indicate multiple characters or the opposite, that multiple syllables could indicate one kanji character.

I talked to Royce about it, and the only conclusion I reached was that Japanese was weird. In order to time the characters to sounds correctly, our solution was to go into the lyrics file and indicate where we needed to elongate or shorten characters. A tilde after a character signified that we needed to hold a character for an extra syllable. A backslash signified that a character needed to appear at the same with the character after it.

This was an awkward manual indication system that required a lot back and forth exchanges between Royce and I. Royce was the expert on the language, so he was the one responsible for putting in the notations or changing out characters. Once he made a few changes, he had to pass the new lyric file to me, and I would regenerate the storyboard. Then, I would be the one to check for new mistakes, and if I found any, we would repeat the process again.

Granted, finding mistakes wasn't that difficult since you can immediately notice when you're on the wrong line in the song. Despite that, we still went through at least a week of constant back and forths between us to fix the problems. Also, a bunch of random issues came up that we didn't expect and had to address.

For example, some characters needed a double tilde following them to denote holding a character for another two syllables. We also didn't realize for a while that there were some single syllable sounds with multiple characters attached to them. Fixing these issues required extra coding and cases on my part. By the end, my code got into somewhat of a hot mess dealing with random problems.

It took a long time, but we eventually reached the end and fixed up all the issues. Again, I'll reiterate the difficulties of doing the storyboard like such and displaying each character individually. Timing was difficult because we needed to display each character on its own. I didn't know the language well, so I required third party help to look at my problems. If all we needed was line by line lyric display, this would have been a metric ton easier.

With the timing down now, most of all the difficult parts were finished. I made some rotation, pop in, pop out, fade in, fade out effects to give the characters some flair. And I added in some random star/heart explosions in the background that went off whenever GUMI said something. And that was pretty much it on the storyboard side.

My last task was to finish a skin for a game. I could have worked a ton more on this, but we only had a little bit of time left so I adapted the skin I was currently using for the map. It's basically just a pretty big edit of the popular aesthetic skin. I kept the HP bar and cursor from the original and just mainly modified the hit circles and coloring. Not particularly fancy, but it gave the map a little more flair.

Now it was time to put everything together. By now we were hitting the end of the summer, and while we had most of everyone's parts completed, we didn't have time to iterate on our work and go over issues we had. Wes released a final version for us to use, and this was given to mappers and Nick.

This final song included the final version of the vocaloid which Royce did some extra "tuning" on. I use the word tuning lightly because according to him, he had a lot of trouble figuring out how to do the tuning in the first place and figuring out if his changes even made the song better. After some attempts, I think he just did a very simple filter through the whole thing and applied a general effect to everything.

That, and he also switched the vocaloid to using Sweet GUMI. Personally, I only liked Sweet GUMI because it just fit with our song title so well, not really for any musical/vocal preferences. Wes was kind of upset about it and liked another version better, but the group as a whole preferred Sweet GUMI.

There're a lot of awful vocaloid songs out there, but I don't think our song is one of them. Maybe it's just bias because I physically worked on it, but I really love GUMI's voice in this regardless of how few changes Royce/Justin made. Wes of course deserves a lot of the credit. He made additional mixing changes to make everything sound a lot brighter and more alive. I'm proud to say I actually like this song.

I was responsible for putting everybody's parts together and uploading the map to osu!. Going through me, by the way, broke an osu! ranking rule. One of the rules states that the owner of a ranked map must be a major contributor to the maps, which I was not. As one can see, I didn't make a single map for Sweet Regret.

In total, we got 4 completed maps from Langston, 1 half map from Justin, and 2 mania maps from Josh. That's a lot of work, and I'm thankful for all the mappers' contributions. I am a bit sad that we weren't able to mod each others maps and give each other feedback. I think everybody would have liked getting some tips to improve and polish their technique. Instead, we had a kind of awkward "hey thanks for everything, kthxbye" atmosphere, and I feel bad for such a rushed ending.

In the future, I'd like to reach our completion milestones much earlier so that we can iterate and go over maps. At the end of the day, the maps are the most important part of an osu! project, and our focus should be getting these up to par over anything else. I suppose our lack of mapping experience contributed to difficulty in giving feedback. Hopefully on later projects we can get some more experienced hands to help out.

Another important part of iteration would be the animation. Nick had some difficulty getting his parts finished because the video was much longer than he was used to making. It wasn't until the very end of the summer that he gave us a finalized animation. He was also on a tight schedule because immediately after summer project he was jumping ship onto other animation gig.

I appreciate all his efforts in making the video to us. This was something completely different from all my past projects and for maps in general. It's not often you know someone who's experienced MMD. It's definitely taken a lot of work getting props, settings, characters for the animation and of course doing the animation, camera work, lighting, and timing involved.

I really wish we still had some time left over because I wanted to give a list of things to look at in the animation. Earlier in the summer, Nick provided us a rough cut of about 1 minute for us, and I think I gave some fair, constructive criticism for animation there. I wasn't able to do that for the overall video since it was handed to us so late.

That's not to say that it's bad quality or anything, but I value the collaborative process between us, and I think he's very easy and fun to work with. There were some small inconsistencies I wanted to address. I probably still could have said something, but when he mentioned that he was planning to switch onto another project and was glad to see Sweet Regret over, I decided it was best to cut the project here and say our goodbyes. By this stage, I was also already looking to the future and planning our Fall beatmap project.

Overall, I'd say things went quite well. We probably did the best we could considering our limitations and ambitions. Working online is always difficult coordinating things, but hey, we pulled things together and had a complete project by the end. Our inexperience and time constraints hurt us, and I think with iteration we could have done better.

This was a very unusual project for me to work on because there were so many moving parts, some parts of which were things I never imagined I'd have to work closely with like MMD. I'm pretty proud of my own work. The background I drew was something I really loved and am proud of. The storyboard was pretty simple, but I learned a lot through programming it, and it looked pretty good and unobtrusive in my eyes. I think I did a fairly good job leading the team, and everyone came out unscathed and pretty happy, I hope.

Last but not least let's put some final links up. First I'll credit the non osu! versions. Nick has the animation by itself put on his Youtube channel. This is just the background video and does not contain any part of the map or storyboard. I was talking with him briefly about maybe somehow transporting my storyboard into video form, but that seemed like too much work so I never did it.

Next we have Wes' song by itself. This was uploaded onto his channel, WasoonMusic, his artist title he goes by. This contains just my background illustration. Also on his channel is the osu! version of the song. This was played and recorded by Aaron/Xenocidel, a friend and officer of osu! UCI. The version played was Langston's Insane.

We actually had to jump through some hoops in order to get the video working on the osu! site. For Aaron's playthrough, we used the full HD video that Nick made for us in the background. osu!, however, has some pretty strict size limitations, so we couldn't upload this gigantic file onto the site.

After a lot of compression and testing, we uploaded a really crappy version of the video to the site. It looks pretty super bad, with a lot of grainy artifacts everywhere. I don't know if there was much we could have done. Maybe if we asked Nick to properly render a lower quality video for us, that would have been best, but by then we were way passed the summer, so I didn't want to bother him.

Oh yeah, as a final note, I guess I'll mention something I think was pretty funny that we never got a chance to make for the project. One of the things we wanted to add at the very end of the song was a cheeky little remark voiced by GUMI in the outro of the song. In our first meeting it was going to be something like "Ah... Miku then pointed over to the ice cream shop." We never mentioned ice cream in the song which is probably the most popular summer dessert, so the end was a fitting place to do that. Shame we never got around to it though.