End-of-July Update (Pixiv Fanbox)
Content
Hey guys! I hope everyone had a good month. I have some news and updates about all things VLVT.
Belly of the Beast
Things are finally moving on this animation!
One of the biggest things holding me back from completing this has been the audio. For a long while, I just haven’t been happy with it. Lots of people tell me that sound design is the ‘thing’ that sets my animations apart from the rest, so it kills me when the stuff I have just doesn’t hit the mark.
To remedy that, I had all the voice acting redone with a new actress. She was open to getting lots of direction and feedback, so what I wound up with is very close to the vision. The new lines sound a lot more authentic, too.
Another addition to the soundtrack has been non-diegetic elements—sounds which aren’t being created by something in the scene. I’ve been hesitant to use audio cues like this in the past, but it’s working really well in BotB.
I’m currently animating to this revised soundtrack and it’s going more smoothly than before. I’ve also done several revisions on the completed animation, both to add realism and make sure it’s all synced. Currently, I’m working my way into the final third.
Box 2.0 Design Poll
Thank you, $5-and-up subbers for your input! 🎉
I’ve learned all sorts of interesting things from this. The votes and comments actually changed my mind on a couple things, and made it clear where the mark needs to be for Box 2.0. I’ll have a full breakdown of the poll results in the next few days.
Future Polls
I have ideas for two additional polls, so stay tuned. I’d like to make polls in general a bit more regular going forward.
Research
What a rabbit-hole this has been. First and foremost: audio synthesis.
Last year, I used 15.ai to make GLaDoS speak in GLaD to Have You. That AI, as well as most AI TTS systems, I’d describe as “not even good enough to trigger the uncanny valley effect”. It’s obvious right away that you’re hearing a computer speaking. This is why I chose GLaDoS in the first place—her affectations were already ‘computery’, so making her sound real wasn’t a battle I’d even have to fight.
But I’m more interested in the opposite: realistic, AI-generated human voice. Sadly, everything I’ve tried falls just a little bit short. I did, however, come up with a fascinating use-case. When a person’s voice is muffled by, say, a thick, lined wall—say they’re inside a box for some reason—it’s a lot harder to distinguish artifacts in their speech. That said, getting an AI to emote correctly is still very challenging, especially if they’re supposed to be in distress. Actual voice actresses (plus a little post-processing) will continue to be useful for the foreseeable future.
I’ve also been looking into AI-generated images. I’m sure some of you have been seeing DALL•E 2 cropping up everywhere lately. You best believe I’m on board. There are fairly strict guidelines about producing sexually-explicit material, but there’s a lot I can do to generate worthwhile assets for 2D and 3D. Heck, Haruka almost got an AI-generated poster on her wall in the last image set.
Even further out there is a bit of AI tech that can animate the face of static rendered characters based on supplied footage. How fucking awesome is that? I’ve already got the system set up and I’m going to do a series of tests with this. The finished results won’t be as “clean” as an actual keyframed animation, but there are loads of ways to take advantage of this. I’m looking into using another AI system (such as EbSynth) on the output to change the style into something more artistic/expressionistic. But hey, even if none of this pans out, it’s still pretty exciting!
Subtitled Animations
You should be seeing a lot more of these in the coming months. I’ve procrastinated on these long enough, and I think these offer something fun for fans, hard-of-hearing or otherwise.
Writing
Alchemist Ch. 3 is done. I’m holding onto the release until I can do a poster for it. There’s a chance I’ll just release it anyway if I can’t think of anything good and just want to move on with releases.
Personal
I’m honestly a little apprehensive to even post this, but I’m working on a project that’s a bit bigger than anything listed above. Soon—hopefully some time this year—I’ll be able to quit my job and go full-time as a 3D content creator. My body is ready, but I’m not sure my bank account is.
And that’s just about everything. Thank you all for the support, and I hope you’re all excited for some juicy new content.