[!Preface] I made Subaru Tap: a Subaru Natsuki scream-scale cheer tool | Amiya’s Desk. Go give it a try if you feel like it, and maybe support Bilibili Moe too, hehe.
How SubaruTap Came Together
The spark
Bilibili Moe has been unusually lively lately. I got back into Re
As someone who also likes Subaru Natsuki quite a lot, I started wanting to join the fun, cheer for him, and catch a little traffic. I do not have the art skills those artists have. Even if I did, that would not automatically mean I could make a handwritten-style video. I also do not have much experience, money, or time for rolling the dice on AI video generation. It was a sudden idea in the first place…
Then I remembered Dagou Dagou Jiaojiaojiao, a Big Dog Tap made by Mark Cup. The first time I saw it, I thought: wow, this is such a good way to find an audience. Of course, the creator really is skilled, open-sourced it, and has a sharp eye for entering through a hot topic. They earned that attention.
So I thought Subaru could probably have a similar scream tap too. He has a ridiculous number of scream clips, and so many of them are classics (lol). It also gave me a way to do something for a character I like.
The build
Once I decided to make it, I started gathering material. I searched Bilibili for scream compilations. Sure enough, plenty of Subaru fans had already done the filtering for me (fog), so I downloaded videos and extracted the audio.
For the images, I happened to see two that worked in a comment section: one Return by Death image for the cover, and another for the main in-game visual.
Then I borrowed from an open-source project and built the thing with a lot of GPT clanging around in the background. Having a public repository to learn from made it much easier. GPT still had some truly silly moments, though. It would ignore the existing library and invent weird rules of its own. Only after I made it follow the open-source implementation strictly did it finally start behaving.
After it was done
Since this was made to cheer people on and catch traffic, the next obvious step was recording a video longer than thirty seconds. I had just bought a microphone too, so of course I wanted to try voicing it myself. That is how this came out: Re:Zero: After MikuTap and Big Dog Tap, SubaruTap makes its shining debut!
As usual, I shared it in various social platforms and communities. The early numbers did not look great, and I started worrying that the video had flopped.
Then I checked again a few days later and, somehow, Bilibili had started recommending it. I was honestly surprised. I think this may be the first time I have received a real recommendation push there (sad). Still, I was happy. I had almost convinced myself my account was being throttled… now it looks more like my older work simply was not interesting enough (sad). It is still good news: I just need to get better at making things.
Thoughts After Finishing It
After making SubaruTap, I started looking at my own identity again. What exactly should I call myself?
A community regular? A game developer? A blog operator? A VPS tinkerer? A video creator? Someone who uses AI? A web developer? There are even more labels here.
Maybe I am a content creator. I like creating itself. Honestly, I still want to learn to draw, write music, write fiction, play instruments, and pick up all kinds of skills.
But I also like collecting resources: finding free stuff, digging up useful sites, keeping good material around. That does not seem directly related to creation itself, so what does that make me?
Confused, I asked AI for help. It pointed out that collecting and creating do not conflict at all. The process is more like: collect -> extract -> rework -> create or share.
So after deciding that I really am a content creator, the next question became how to become a better one, and what qualities creators usually need.
How to Become a Better Creator
Good qualities for a content creator
-
Making connections: moving from A to B, C, and D. SubaruTap was inspired by Mark Cup’s Big Dog Tap, and Big Dog Tap was itself inspired by the classic MikuTap.
-
Strong execution: when you get an idea, make it. Download the materials, tune GPT, write the code, record the voice, post the video. Do not leave yourself too much room for regret or hesitation, and you may suddenly find that it is already finished.
-
Knowing how to borrow tools: right now, AI is one of the most useful tools around. It helps with research, development, and expanding ideas. There are other tools too: predecessors’ open-source work, AI resources from communities, even images in comment sections. Without the resources I had collected before and the ability to find them, SubaruTap would have cost more money or much more time.
-
A little shamelessness: finish it and post it. After posting, share it in social platforms and communities. Good wine still needs an alley people can find. If you never show your work, nobody gets to see it. Build in Public can lead to better results.
-
Doubt yourself, but keep making things: see Creator Confidence Cycle. When you are not feeling good, do not abandon what you are making. Even if the work becomes mechanical for a while, keep going.
-
Think long-term: a break does not mean your creative life is over. If you did not make anything today, make something tomorrow. Do not punish yourself too harshly for a tiny lapse. Before SubaruTap, I had also gone several days without making anything public.
-
Keep learning for life: creators need a steady source of things to make, and learning is genuinely fun. If I had not tried building websites, buying domains, and running servers a few months ago, I would not have made SubaruTap or written this post today (lol).
To be fair, DeepSeek helped me write some of the above too (fog). Looking back, I have some of these traits already, such as execution and shamelessness; others, like long-term thinking and lifelong learning, are still things I am practicing. If you made it this far, maybe compare them with yourself as well.
So what should we do next?
Yeah. What should we eat next?
For me, I am currently closing out the final stretch of a Godot Hub Jam project: [heart] [cracks] [ending puzzle] [ending sequence]. The true ending starts to play inside the game, which means Delay Trace wins! Next episode, damn it.
Then I will finish this post and turn it into a video.
I also have another video planned about my home server, plus a proper introduction to my cute Sayori bot.
After that, I can try making my Steam game: a new puzzle game mixed with ARG ideas, and maybe some passionate combat too.
If you made it here, I want to ask you a few things too:
-
When was the last time you had an idea and immediately made it? What happened afterward?
-
Which quality do you think is strongest in you? Which one do you lack most?
-
What is your next step?
Replying in the comments would be great. But even if an answer only flashes through your mind, that at least means you are thinking about creation too. That is what I most want this post to do: not only reflect on myself, but prompt other people to think.
I think creators have a responsibility to guide others.
Extra Material
Here are a few videos I like that fit this topic. I am leaving them here as extra material, and I might add more later.
- A video by Naoki Saito. It is about drawing, but there is still a lot to learn from it. A very good creator; worth following.
Some information may be outdated