Prepping To Storyboard Wasps & My Take on LLMs

I'm likely to finish Episode 1's screenplay by new year's eve. 2025 will be the year of the storyboards. I have limited experience boarding, so will force an iterative multi-draft process for the sake of learning. Here's "an Undertale" I did almost 8 years ago, to demonstrate the drawing-detail level I'll be aiming even lower than, probably.

A serviceable storyboard page featuring Undertale fanon of Alphys' discovery of Frisk's journey by security feed on her PC. The boards have no shot length indications, but the pace comes through enough for a 90-ish second sequence.

I want to get accurate pace of shots asap, even higher priority than composition, but might just omit specifying shot length until software-ing a rough animatic. I always think of the comically simple boards of Knives Out, to prevent myself overpolishing individual frames.

Anyways, I can't deliver on the promise of my last post: Viewable pitch decks for Charade of the Paper Wasp & Charity Case. I got distracted by discovering a great use case for LLM services: Story Bible chat bots.

I've spent months educating a few powered by Claude 3.5 Sonnet with my project notes and outlines, having used it many times as a co-writer already.

LLMs ( large language models ) are of ludicrous use to writers, but a detriment to poor writers. I also notice pantsers tend to hate dealing with AI slop, more than outliners who see it as no different from editing.

"But what about tunnel vision?" "But what about missing out on some mystical thing only achievable from the blank page?" Those sound like copes, because you're not confident in your vision as a storyteller, or you don't outline, or you literally only enjoy writing when you don't know who your characters are.

The simple use case of "I know what I want when I see the AI slop getting things wrong." doesn't take much of the fun out of execution of the written word. I like problem solving just a tiny bit more than discovery writing, and any pantsing parts of me are tethered to my drawing, not my typing. I like rationalizing plot based on visual flashes from my vision. I "have it" and need to release it. Discovery has already happened by the time I decide to make something.

The whole first half of Charade's script was done with no LLM involvement, before I got fed up with my slow pace and occassional procrastinating and just tried feeding my chat bot my scene cards, and immeditely realized how much time I've been wasting. I'm trying to make a film all by myself, with limited time on planet earth, not impress an imaginary purist script agent in my head.

That's not to say there aren't unique challenges to introducing AI slop into the pipeline. I definitely catch myself slacking due diligence to not accepting anything but the best string of words with the proper character of a character, or tone, or the vision / vibe. I see this as a trade-off to draft things at lightening speed. Writing is rewriting... right? As an outliner, I personally consider most of the HARD part done much earlier than most, such as filling BS2s and scene cards, and the months of introspection and structured daydreaming that proceed my commitments to put things to text.

And no, I don't use AI to write these posts... though the option for that is literally built into this webhost. Yikes! Cringe in-app service vs chad manual inclusion.