Progress Report - January 9, 2023 (Patreon)

Hello, everyone!

It appears I missed last week's Progress Report. That's starting to become a habit, and I'm not terribly pleased by that. I won't lie, all the days have been bleeding together for me of late - so much stuff getting done, all the time!

The whole first episode of Primordium has been fully blocked out, and is awaiting voice lines. Half the voice cast got sick over New Years, so the January 7th stream ended up being spent working on a title sequence for the series. We put together 14 animations in about 8 hours on-stream for it, which is a pretty good haul in my book.

In sum, Primordium Episode 1 currently has a runtime of 7:23, and you can click here to watch the full WIP. Again, it isn't fully voiced yet - only Ciri's lines have been installed so far. The title sequence currently sits at a runtime of 1:13, and you can click here to watch the current WIP. The title sequence WIP isn't finished, as the still images, unfinished scenes, and lack of lighting probably suggests, and the music very likely won't stay in its current form.

The purpose of the title sequence is to sell the series forward. Each two-second animated sequence is a tiny vertical slice of a different episode idea. 15 episodes in total are teased by the sequence, and I had a hard time picking them - I had to leave a lot of tantalizing ideas on the cutting room floor, because they were either too abstract, too high-context, or not high-octane enough to accurately represent in a tiny two-second clip.

I have a few concerns about the title sequence, and they are frankly at odds with one another. The first concern is I fear the title sequence is too long - Primordium is targeting 8-minute episodes, and I feel like a 1-minute title sequence slapped in front of each episode is a bit... uncouth. The second concern, paradoxically, is I fear the title sequence is too short - the animation clips are only two seconds long each, and you barely get a good glimpse at them when half their runtime is consumed by transitions.

I've had a few people suggest that the title sequence isn't too long and I should extend it further, but I'm still not convinced. If anyone has any input on this conundrum, I'd love to hear it.

Much of last week went into storyboarding the title sequence, which was... not the plan. I intended to spend a single afternoon getting the poses put together in preparation for the stream. I ended up spending four days on it.

As a final note on Primordium, the storyboard for episode 3 is almost finished. It's running long, and will probably end up right around the 10-minute mark. And that's after I had cut out the planned ending sequence and instead merged it into the previously-penultimate sequence. If the Episode 1 storyboard is anything to go by, where the storyboard suggested an 8-minute runtime and the video came out to 7.5 minutes (probably closer to 7 once all of the audio comes in), then it stands to reason that this would put episode 3 around the 9-minute mark. Which I'm willing to accept, especially since I have already aggressively cut down the narrative sequence, relegating an entire character exploration arc to some later episode.

Beyond Primordium, my attention for the past two weeks has been focused mostly on Claire/Jill. More specifically, it's been focused on the auto-jitter tool I am codeveloping alongside Claire/Jill. I discussed the auto-jitter tool in great detail in the last Progress Report, specifically its advantages and its challenges. I predicted the challenges to be challenging, and it was an apt prediction. But I think I've finally worked out most of the kinks to the point that I can actually fly through the rest of the project. Most of the kinks.

The biggest challenge, as I had predicted two weeks ago, was the hop from jittering vacant animation to jittering existing animation. I, uhh, had to basically rewrite the entire jitter algorithm three separate times to finally get it. And it's still not quite right - the algorithm I came up with has a tendency to degenerate, with the animation just completely jumping off the rails to some rather hilarious results. To compensate for this, I added some correction to the algorithm, to try and keep the jittered animation around the original animation.

The first few results had some unacceptable issues, but the final attempt resulted in me biasing the random jitter toward values that pull the animation back to baseline - basically, the larger a value it generates in one direction, the more likely it is to generate the opposite direction. The results are pretty good, but there are two problems: first, I have some bizarre math cases that, if left unchecked, causes the animation to explode like the link above. I don't know what causes the values to explode, so I've sort of just forced the values within a box. This means there is a non-zero chance the animation can degenerate just like before, if it keeps slamming up against the edges of that box. I'm not pleased about that.

The second problem is the correction factor tends to have a sinusoidal motion. It's not too bad if it's small values over a long span of time, but for more intense motions, the seesawing wave becomes pretty apparent. I want to mitigate this by using nonlinear correction, so that small deviations are very lightly corrected and large deviations are crushed by the fists of God. My efforts to correctly implement nonlinear correction have been... less than successful.

Ultimately, though, I decided that the correction is good enough for now, and I've decided to move forward with progressing the animation side of things. My hope, then, is to have all of the jittering done before the weekend. Could be as soon as Wednesday evening, but I am being kind to myself and giving myself until Friday.

That's all for now! Again, I apologize for missing last week's Progress Report.

I'll try to make next week's. As I said, I don't want this to become a habit. Until next time, everyone!