THIS BLOG POST CONTAINS SPOILERS FOR THE "MoeChakkaFire - Tristan's Cover" PROJECT. IF YOU HAVE NOT WATCHED IT YET, PLEASE DO SO HERE.
With the release of the "MoeChakkaFire" cover, I decided to write down my thoughts as to how this project was created, released, and what I could have done better with it. I also wanted to write this post to give people an idea of the creative process of this project, including the motion graphics, animation, and song.
I am once again using a translation service to write the Japanese translation. This time however, the translation will be made separately from the English Translation. It may not be accurate, so if you find any mistakes, please contact me at tristan.gaylord7@gmail.com with the correct fix.
This project was hell.
Or at least some equivalent to it. I lost a lot of sleep during this project, as most of this project was crunched in a two-week timespan. Several nights, I stayed up until 2:00 in the morning in my time zone, working nonstop on this project. Some nights, it was 3 in the morning.
And the worst part was that this project was not even originally meant for the competition at all! I saw the song on YouTube one day, and I thought to myself "Hm, this song isn't bad. It's kind of catchy." Before I knew it, those thoughts turned into "Hey, this is actually pretty good" to "I kind of want to try reanimating the chorus with my own character."
Then, while I was in the very early stages of animation, I saw the Twitter post with the requirements: Create your own fanart around the MoeChakkaFire song. Each submission is judged by the original song creator issey and the MV's animator hatsumino . There are only 10 winners for the Creative Prize, and each winner gets $300.
Before I knew it, I was thinking to myself "...hey, how much is each of my upcoming college loan payments? $300? Well, now I need to make sure that this animation is as good as humanly possible."
What started as a simple reanimated segment quickly turned into a full song cover, completed with:
more animated segments
my own voice singing the song in Japanese (lyrics weren't too bad actually. i'm good with pronunciation, i just can't read hiragana/katakana/kanji at all)
mixing the vocals and mastering it all together
color correction & grading the final results
clarifying that the email that I am submitting the song cover under is a different email than what is connected to my actual Hoyoverse account, and that I own and manage both email accounts (trust me on this one, I did have Notepad open explaining everything in the screenshot I provided)
and hoping and praying that I can get all of this done within 2 weeks.
So without further ado, here's the full breakdown from start to finish of how this project was crunched within a two-week timeframe.
Character Model Preparation
Before the animation even started, I wanted to create a custom dress for the avatar that I usually use. I wanted to make it close to Ellen's outfit as possible, even using her reference sheet to help with answering questions such as "How short is the maid outfit at the hem?" and "Do I really need to make the outfit cut off near the shoulders?" Keep in mind that I'm not a clothing artist, and I knew very early on that I didn't have the willpower to recreate the outfit 1:1.
So I just used the default VRoid Studio outfit textures and adjusted from there. Instead of high heels, I used sneakers with platforms on them. Actually, I did try my best to recreate Ellen's shoe texture on my own, as a homage to her original design.
Once the design was complete, I took it into Unity, and copied my original model's facial blendshapes onto the new model. This was done with ReForge Mode's NotEnoughBlendshapes plugin, better known as NEB. I used the Lite version because you can still copy Blendshapes from one model to another without paying anything. However, if you want to create your own Blendshapes, you will need to pay for it. You can buy it by clicking here.
Once that was complete, it was time to import into Blender. I used the VRM Add-on for Blender 4.1, since that is the version I used to animate in.
Character Animation
The character animation was divided into sections. To keep myself motivated through this tasks, I decided to post progress updates in a special channel in my Discord server (you can access these Behind the Scenes videos by becoming a member through either Discord or my own Kofi page). Each video posted keeps track of how many seconds I've animated so far, the total percentage of animation complete, and any changes that I needed to adjust.
For animation, I wanted to recreate the original motion completely, without trying to rely on any MMD or motion data already created. This was achieved by using Blender's "Background Images" setting in a camera, which allows me to use a reference clip for matching animation as closely as possible.
As a result, the process went relatively smoothly. Because it is stepped animation, each bone is posed manually and is animated precisely in time with the music video. Hair animation was not referenced from the music video, and was animated based on the character's movements. In some sections, you could find the character's hair rise and fall when performing a jump.
While the animation itself looks realistic and matches almost perfectly, the behind-the-scenes footage is actually quite a mess!
What happened? Why are the character's limbs flying all over the place? Why are some parts of the model not moving at all at times? To answer this, I need to explain two different techniques: screen space and world space.
World Space is anything that can be seen and animated within a 3D environment. Professional animators usually animate in world space whenever working on something that a person can interact with, such as when playing a video game. Screen Space, on the other hand, focuses more on 2D environment, and is limited to only what someone sees on the camera.
Therefore, what looks normal to a viewer while watching from a specific angle can look hilariously broken when viewing at another angle. And many professional animators use it all of the time, from Disney animators to popular 3D community artists.
As for why some parts don't move, I use the age old saying of "out of sight, out of mind". Basically, if it doesn't show up in the frame, there's no need to worry about it. Of course, this saying only works when working in Screen Space. It doesn't work with World Space animations though, because someone may want to view an object from multiple angles.
Actually, while typing this part, I remember that I used this technique again when making the Thumbnail for the video. While the hands seem to be making the heart pose, they're not connected at all.
Moral of the story: Break your rigs when working in Screen Space. It's not only funny, it's encouraged!
Additional Posing and Animation
Once the main animation was done, the next step was to incorporate a way for the character to appear in the music video and have it look somewhat natural. As a result, I made more poses with the character entering the video, holding a broom that I modeled live on stream. There will be a link to download the broom model down in the last section of this video.
As you may have guessed, even the image of the character getting ready to swing the broom has their rig broken.
Now that I look back, I just now realized that the arm is extended farther out than usual. The worst part is that I know exactly what the problem is- I didn't properly keyframe the size of the arm- and yet the video is already published with this mistake. Then again, this video was rushed, so I probably wasn't thinking straight. At least that's the only mistake that I found.
After all of the posing was complete, next was compositing. Most of the shots use two filters: Lumetri Color, for color correction and grading; and a separate copy of the image or animation clip with the Find Edges filter. This is used to give the image and/or animation a solid and distinct outline, rather than rely on Blender's broken tools for line art or outlines.
And now, with 3 days left to spare for the deadline, we come to the hardest part of this entire project: finding the confidence to sing the vocals.
Singing, Cleaning Audio, Mixing, Mastering, & Stressing Out
I don't really like singing that much. I don't live in an area where I can reliably sing a whole lot. I especially don't like to sing when other people are around, since my own family members have made fun of my singing before (granted, that happened years ago, but it still left a mark.)
At the time that I was recording vocals, I would have preferred to do it without anyone else around. Unfortunately, my own family members were around, and I didn't want to interrupt them with any work they were doing while they were in the house. I also had this unconscious fear that I would get made fun of again.
So when the time came to record the vocals, I was very nervous and quiet at first. (You can actually hear this towards the beginning of the song, since I recorded everything in order from beginning to end.) It was only near the end of the song that I fully got the confidence to sing, but by then it was too late. I still needed to clean the audio to filter out any pops and clicks, master the vocals to make it sound clearer, and mix it all together.
Finally, with one day before the deadline, I finished the mixing and mastering of the song. Personally, I didn't like how the mix sounded. I knew that if I had more confidence with singing, it would sound a lot better. Additionally, I don't have that much experience with mixing audio for songs. But considering that I got it all completed within a relatively short time period, I was proud of myself for stepping out of my comfort zone with this project.
What Could Have Been Better/Improved
For starters, I do think that there were some parts to the moving character animation that I wanted to improve upon. I don't really like the hair animation that I did, because to me it feels just a bit floaty and unnatural. Perhaps I would have gone back and smoothed it out more, and made it less "snappy".
As for the singing, I know that I mentioned it before, but I do want to try and step out of my comfort zone and sing more often. Not sure with what song yet, but I do know that if I ever want to try singing again, I don't want to be put under the pressure of a deadline to finish it on time.
Actually, now that I think about it: most of my projects are considered to be "rushed" projects. I don't think I've worked on something that wasn't rushed or under a deadline in years, outside of music covers. Even so, I do want to create something without the feeling of a deadline looming over my head, and actually make sure that I make something that is at an acceptable level to myself.
Or maybe I need to stop being a harsh self-critic to myself, who knows.
Aftermath
Unfortunately, this is the part where I must ask you for your support. The day after I published the cover to YouTube, it was immediately dislike-bombed. The reason was because of several internet trolls that have been making me uncomfortable to make content for years. When I finally confronted them about it, these trolls then proceeded to dislike-bomb a number of other videos on my channel.
This is significantly harmful because my channel had been effectively destroyed in terms of engagement. To my understanding, YouTube doesn't show videos with a strong dislike ratio in their algorithm, and even after asking other people to help share the cover around (and fighting with YouTube Support about this manner), practically nothing has been done to fix this.
So all that I ask is that you share my works around. I know that some of these works may not be of the best quality. And I know that some of these works are very old and outdated. But I do want to improve a lot, not just in singing or animation or even just video editing.
I just want to make something that others can enjoy. So if you enjoy my works, it would be very much appreciated if you shared them around to others.
Also, if you would like to download the broom that was used for this cover, it's available on my own Kofi store. It's free to download, but you can throw in a donation too if you want. Just don't combine the meshes together. It'll break the broom model and add hair all along the broom's exterior. Trust me, I've tried to find a workaround. Nothing I can do to completely fix it, sorry.
Thank you for reading this post!
Update 12/30/2024: We didn't win. But that's okay. I learned a lot from this project, and I definitely want to get better at mixing audio and singing for any future projects. I'll get back to work now.