The last post ended on a sentence that sounds harmless: take a canonical story and turn it into a comic. That sentence hides two assumptions, and both of them are wrong.
The assumption
The plan, as far as I had one, treated the stories as the easy part. They already exist. They’ve existed for centuries. Somebody else did the hard work of making them up; my job was just to draw them.
So the pipeline in my head started at script. Research wasn’t a step. Research was the thing you did on a Sunday afternoon before the real work started.
That survived exactly as long as it took to ask two questions: which story? and whose version of it?
Half one: there is no list
I could name maybe fifteen stories off the top of my head. That’s the famous handful — the ones that get retold constantly, which is precisely why retelling them again isn’t very interesting.
The rest, the hundreds I actually wanted to reach, I couldn’t name. Not because they’re hidden, but because a living tradition isn’t shaped like a catalogue. It doesn’t come with a table of contents. Stories nest inside other stories. A character wanders in from somewhere else with a history you’re expected to already know. Ask where one story ends and you get a reasonable argument in three directions.
For a single comic this is a non-problem. You pick a favourite. You already know which one before you start.
For a series, “pick a favourite” doesn’t scale past about five. You need some sense of what the whole field contains and how the pieces sit relative to each other — otherwise you’re not building a series, you’re building a pile.
Half two: fluent is not faithful
The second question turned out to be the sharper one.
Ask for a story and you will get one. Immediately, confidently, well-structured, and reasonable. The problem is that “reasonable” is not the bar. A retelling can be perfectly fluent and still have quietly blended two regional versions, promoted a minor character, resolved an ambiguity that the tradition deliberately leaves open, or smoothed off the strange specific detail that was the whole reason the story survived.
None of that announces itself. It reads fine. It reads better than fine — drift almost always makes a story smoother, because the rough parts are the parts that get sanded away first.
And in one comic, it genuinely doesn’t matter much. One story, slightly off, told well.
Across hundreds it matters enormously, because the errors aren’t random — they all lean the same way. Every story drifts a little toward the generic. Do that at volume and you haven’t adapted a tradition, you’ve produced a large quantity of myth-flavoured content that could have come from anywhere. Which is the one outcome that would make the entire project pointless.
The insight
The fix wasn’t to research harder. It was to stop thinking of research as something that happens before the pipeline.
Research is the first step of the pipeline, and like every other step it has an output. It doesn’t produce a vague sense of having read around the subject. It produces a specific thing: this story, bounded here, in this version, anchored to actual source material, with the choice written down.
That last part is what makes it a system rather than a habit. A decision I make in my head is a decision I have to make again — slightly differently — every time the same question comes up. A decision that’s recorded is one that later stories can be consistent with.
What it actually looks like
The step does four things, in order:
- Locate the story — establish that it exists as a distinct, adaptable unit and isn’t three stories wearing a coat.
- Situate it — where it sits relative to everything else, what it assumes you already know, who walks into it from another story.
- Ground it — anchor it to real source material rather than to a confident summary of source material. This is the step that stops the drift, and it’s the one I’d have skipped if I hadn’t gone looking for why retellings felt subtly off.
- Record the version chosen — so that when the same character or event turns up eleven stories later, the answer is already settled.
Nothing here is clever. It’s the unglamorous half of the work, and it happens before a single panel gets drawn. It’s also the step that decides whether the finished series is faithful, because nothing downstream can fix a story that was wrong when it arrived.
The lesson
At series scale the expensive mistakes aren’t the loud ones. A bad panel is obvious and you redraw it. The expensive mistake is the small, defensible, invisible one — made once, then repeated silently a few hundred times, always leaning the same direction.
Those are worth spending real effort on up front, because they’re the ones you can’t spot by looking at any single output. You can only spot them by looking at all of it at once, and by then you’ve made them everywhere.
Update log
- 2026-08-18 — Post #2. The first pipeline step, and the one I originally didn’t think was a step at all.
- 2026-08-20 — Added the pipeline map at the top. Readers shouldn’t have to reconstruct the shape of the system from the posts that came before.