Dealing With Loss of Effort; or, Failing to Make the Thing was Part of the Process
Over the past while I've been talking a lot more seriously about returning to music production in a big way. One of my partners got me absolutely obsessed with Project Sekai, and vocaloids as a result. This has led me down the path of finding a lot of excitement in working with Kasane Teto's voicebank, and thus follows an excitement to make original music for her to sing to. Thus, the virtual band project was born which currently has the project name of STRYX9.
This project has been a hyperfixation of mine lately, with a lot of work being dumped into VRoid and Blender to create character assets. The thing is, after finally completing most of the assets, I'm finding that the art direction I've taken isn't inspiring me quite as much as I'd hoped, and has some drawbacks I'm not a fan of.
Having used VRChat for a long time, I know that anime models can have some limitations to their expressiveness and making an expressive, visually striking 3D anime model is extremely difficult. It therefore follows that 3D models made out of a create-a-protagonist program like VRoid are going to make compromises on the expressiveness of the models just to make sure each model at least maintains a minimum level of quality. Beyond that, trying to find clothing for each character has been a hassle, and concessions have had to be made on each character's identity just to make sure each character has an identity that separates them from each other, but still allows them to exist as a cohesive group of characters. Further than that, metalcore kind of requires more expression in the faces of the models I use because of the aggressive rhythms, riffs and vocals used within it. Overall, they just wouldn't look right in motion while performing music videos for the music I make.
And so, knowing full well that I have put an extreme amount of effort into each character's model and assets, I've made the decision to move my character designs to a different art style entirely.
The 'me' in my early-20's who was scared of failure would have seen this scenario and gone "well all this work was for nothing so what's the point," and would have given up entirely. The current 'me' just has more confidence in the project now that I've determined a strong new direction for the project.
The thing is, I went into any type of creative work on this project knowing nothing was gonna get going in earnest until Kasane Teto's SV2 voicebank update is released. Rather than purchase the software now and deal with a technically worse version of Teto than others would after V2 releases, I'm going to wait until the upgrade is published and have SynthV2's full workflow at my disposal. Since we're just sat here waiting for something to happen before any of the real musical work happens, I figured I would make some use out of the downtime and try to pin down their designs. My goal wasn't to set out and make the models I'd end up using, it was just to see if the process I had in my mind would even work. Since the goal had nothing to do with the look of the characters, I was willing to abandon all the work I'd already done in service of finding a much better workflow.
Something I noticed when work was finished on the 3D anime models was a lack of expressiveness in their faces, as I alluded to earlier. It was very difficult to make them look like they were belting, screaming or playing with passion without some part of their geometry looking really strange. Some might not notice it in a music video, but I absolutely would. I want to make the production of music videos as painless as possible, and having to go into post-processing and digitally paint multiple frames of a fucked up weight-paint out or creatively cropping out a weird clipping issue would burn me out on the process fast. However, one place that most people just expect clipping, weird geometry freak-outs and a little less expressiveness to happen is in the PS2 era. Of course, there are games that dig right into a toony style and use squash-and-stretch to bend their models into an extreme expression, but any weird clipping issues that occur from doing that are kind of accepted as part of the medium. Clipping is found in major animated affairs in the same way brushstrokes are found in beautiful paintings; it's just how the format is. Notice how Mermaidman and Barnacle Boy's eyeballs in the timecoded video link are literally just spheres clipping into their heads, and the animation is stiff, and yet they're still very expressive?
Sure, I could just deal with the clipping and lack of expressiveness; most MMD videos just inherently have hair clipping through the face, or skin clipping through clothes in them. However, when the models are that high-fidelity and the art style those models are inspired by is so heavily dictated by expressiveness and in-motion readability, it logically follows that the 3D anime models require an extremely delicate, patient and knowledgeable hand to animate properly - that's a hand I do not have. Rather than doom myself to those same problems only a few professional studios have figured out, I'd rather change the artstyle to one that is already revered for being imperfect. Low-poly models often clip into themselves in games that are highly-regarded among gamers, it's just naturally seen as part of the aesthetic due to the limitations of the hardware at the time. People were just impressed they were seeing their favorite characters with higher visual clarity at all. I mean shit, most people consider Gran Turismo 4 to be one of the best racing games of all time, and the pit crew animations themselves are stiff and robotic, with an impact wrench often clipping through the wheel it's trying to remove.
Given all of that, it's a no-brainer to move to low-poly. While not openly stated as the case by the creator of Nightmare Kart, it's obvious to me why she and many other independent/solo developers use low-fidelity artstyles for their video games. It's just that much easier to work with, and much less time is required to make low-poly stuff look great when compared to HD! It puts the viewer in the same mindset they were in as a child playing low-fidelity video games, taking all of the details the art style is missing and making it up in their heads instead. As expressive as Crash Bandicoot is, I still keep underestimating just how much Naughty Dog squashed and stretched that motherfucker - and how many times Crash's hand clips through his own body.
The fact remains though, I'm already going a lot further than many other vocaloid producers go with my marketing strategy. Almost every other vocaloid producer I'm aware of just uses simple lyric videos that show some artwork of the vocaloid, have the lyrics slide in and out of screen as they're sung in the song, and leave it at that. It's not lazy by any stretch of the means, it's just the nature of things; if you're doing everything from the music to the branding yourself, some step of the process is going to suffer because you've focused on what you're good at instead. Other producers often form groups with artists to create visual effects and original animated artwork for their music videos, but often the most dynamic part of their music videos is the typography. Words will splash, squash, stretch, bend, twist, rotate, flip, swing, sparkle and shine as the artwork of the character singing those words remains simplistically animated, moving its arms or hands around without actually singing the lyrics being flashed on screen.
This isn't to say those producers and artists are lazy; far from it, actually! Animation is really fucking hard work, and making music based on fictional characters inherently requires either animation or cosplay to make that character come to life. Given that most vocaloid producers choose to remain faceless, animation becomes the choice to be made. And when you don't have the money to pay a studio to animate a music video like the ones by Gorillaz or Alvin and the Chipmunks, you're stuck doing what you can manage on a tight budget and extremely small workforce.
Ultimately, I want a particular standard of quality in my work that doesn't pull people back and think "ahh, she's trying her best with what she has." I want people to look at my work and say, "holy shit how the hell does one person make all of this look good?" 3D anime wasn't gonna do that for me. Low-poly 3D will.
Doing all of the work myself is still a daunting task, though. I'm not just producing the music, which is already a significant challenge: I'm also using storyboarding, choreography and cinematography skills for the music videos; I'm using graphic design, storytelling and marketing skills to create the project's branding; and I'm using engineering and art skills to try and bring these ideas to life in one way or another. Most musical acts hire contractors because not many people can do all of these things in a single package. For each music video there's usually a director with their own workforce of cinematographers, costume designers and lighting directors to make it look great. For each album there's a team of branding experts, graphic designers and pressworkers making each physical release look and feel good in the hands. For each song there's a number of recording artists, producers and mixing/mastering artists who try to make every song within an album feel as big as it can be. So really, what I'm trying to say is, "if I wanna do all of this shit myself and do it at a high standard, I need to want to change ANYTHING in the workflow to make the process easier while maintaining its quality."
In fact, the reason I'm deciding to move toward low-poly artwork for my characters and move away from the 3D Anime style is specifically because i know how much of a struggle it is to make animation look good enough to match the quality of the music around it. Being able to mask the animation errors with some pixelation and reduced-colour dithering in post-processing will actually just enhance the experience rather than kill it, which gives me enough forgiveness in my production method to just keep rolling forward. I was previously giving up on things I failed at - not just because of rejection sensitivity, but because I was constantly setting goals up for myself that I was doomed to fail. This time, the goals I've set aren't just within reach, they're actively designed to keep me from failing. Rather than bash my head against a wall trying to make something work that even talented animators struggle with, I'd rather change things up and create things in a framework that's more realistic for my skill level. That's why I can look at all the work I've done on the 3D models I've made so far and kick those models to the curb. Besides, it's not like it was all a waste; thanks to those 3D models, I've settled on a general design that can be translated into low-poly styles.
The work I've done on this project up to now isn't a failure, not at all. Instead, it all was part of the process of trying to make something fucking amazing, and I couldn't be more excited to say, "I tried that and it's not working out, let's try something else."
