Sometimes a project teaches you something before you've even finished it.
I've been working on an illustrated comic project recently. The first page was ambitious: multiple characters, a detailed environment, physical comedy, changing emotions, collapsing cardboard structures and a surprising amount of continuity to keep track of.
It was also great fun to make.
The illustrations are AI-assisted, with the final artwork directed, edited and assembled by me. Going into the project, that made the scale feel manageable. AI can generate an image in seconds, after all. Except a finished illustration isn't the same thing as a generated image.
Sixty panels aren't sixty prompts
The first page took me around 14 hours. Some images appeared remarkably quickly. Others absolutely refused to cooperate.
A character needed to hold a particular object in a particular way. Another needed to interact with an established piece of scenery without that scenery mysteriously changing. Characters needed to remain recognisably themselves. Objects had to stay where they had been left. Scale, lighting and perspective needed to remain coherent.
Sometimes generating another image was the quickest solution. Sometimes Photoshop was. By the end, I wasn't simply prompting an image generator. I was art directing, editing, compositing, maintaining continuity and designing the final page. The first page became something more useful than a deliverable, it became a prototype and prototypes give you evidence.
Writing has a visual cost
One of the biggest lessons from that prototype was that a tiny piece of writing can represent a surprising amount of production work.
"The character picks up a broom and pokes the structure."
Simple sentence.
Visually, it means introducing a broom, creating a pose that holds it convincingly, positioning it correctly against the structure, maintaining the established character design and then remembering where the broom goes afterwards.
And that broom. THAT BROOM... turned out to be particularly troublesome. Getting the character to hold the correct end, point it in the right direction and interact convincingly with the structure took far more iteration than that one sentence would ever suggest.
The frustrating thing is that I already knew there was a much simpler solution; he could poke it with his toe. No new prop, no complicated grip, no dropped broom to account for afterwards. The character could remain close to an existing pose and the exact same narrative event would occur.
Arguably, it suited the character better too, but I didn't do it. I was trying to stay true to the script.
That revealed another useful distinction: being faithful to the story isn't necessarily the same thing as being literal about every visual instruction.
What did the story actually need at that moment? It didn't need a broom, it needed the character to interfere with something another character had carefully built. Once I started thinking about it in terms of the narrative beat, rather than the prescribed action, the production problem became much simpler.
"The character scrambles backwards up an apple tree."
Okay, so now I need an apple tree.
The tree has to belong naturally in the established environment and the character needs a new climbing pose. Scale and perspective need to work and once the tree exists, it needs to continue existing in subsequent views of that location. Again, the question becomes: does the story specifically need an apple tree, or does it need the character to retreat somewhere in alarm?
Sometimes the answer genuinely will be the apple tree. A particular object, action or location may matter later, carry meaning, or be essential to the activity, but sometimes it is simply one possible visual solution to a narrative problem. That has changed how I think about adapting scripts for illustration.
Rather than asking:
How do I illustrate exactly what this sentence says?
I can ask:
What does this moment need to communicate, and what is the clearest, most efficient visual way of communicating it?
That isn't taking liberties with the writer's work. Done properly, it's part of translating writing into a visual medium. None of this means stripping interesting objects, actions or environments out of a story simply because they're harder to make. That would quickly result in a very dull comic. It means recognising that narrative complexity and production complexity are not the same thing, and deciding deliberately where that complexity is actually earning its place.
The broom wasn't. The toe would have done just fine.
Designing the script for production
After completing the first page, I fed this back.
Could existing locations be reused?
Could visual descriptions intended to guide the designer be separated from the words that actually needed to appear on the finished comic?
Could artwork carry some of the descriptive storytelling?
Could we identify the essential visual beats before production began?
Could established character poses, props and environments become a reusable asset library?
The revised script did exactly that.
It separated visual direction from reader-facing text. It identified must-have panels. It deliberately reused locations and compositions. It considered where one character pose could support several pieces of narrative.
It was a much more production-aware script. That should save a considerable amount of time, but then I started estimating it properly.
One asset isn't always one asset
This was the next lesson. On a script, a rocket is one thing, for the designer, it might be:
an exterior design
a cockpit
a dashboard
a cargo bay
several viewing angles
something that needs to remain recognisably the same rocket throughout
Likewise, "a new planet" isn't necessarily one background. It can mean establishing a visual language for an entire world, designing its landscape and environmental features, creating a new character that belongs there, and then making all of those things work together consistently across several panels. AI makes that process considerably faster than creating every element manually from scratch, it does not make the process disappear.
Asset reuse saves time after the asset exists.
Someone still has to create it first.
Then one clarification changed the project
My understanding had been that the sessions were individual resources. That gave me a production strategy. Create the early sessions first. Deliver those. Continue producing the later resources as the programme progresses. Staggered production made a large project manageable.
Then came a clarification. They weren't ultimately separate resources. They were pages of one publication and the complete publication needed to go to print together.
Interestingly, almost nothing about the artwork had changed; it was the same characters, the same panels, the same script, the same number of illustrations.
But one constraint had changed, and suddenly my entire production plan no longer worked.
That's when I realised:
Sometimes the scope doesn't change. Your understanding of the scope does.
And that can change everything.
Possible isn't the same as deliverable
I went through the remaining pages individually and estimated them using what I'd learned from producing the prototype; some benefited enormously from asset reuse and might take only a few hours, However others introduced an entirely new environment, character and set of interactions.
My estimate for the remaining production came to roughly 63–93 hours, with somewhere around 75–80 hours looking realistic. That number made the decision surprisingly simple; Two conventional full-time working weeks are about 80 hours. Even if I abandoned everything else and treated the project as my full-time job, the realistic production estimate would consume essentially every working hour available before the deadline.
No contingencies, no difficult image that unexpectedly takes two hours, no client revisions, no final corrections, no print preparation.
No life.
And, more importantly, I don't actually have two empty working weeks. That's when the question changed. It was no longer:
Can I make this?
I can.
It became:
Can I responsibly promise to deliver this within these constraints?
I can't.
Those are very different answers.
AI doesn't create capacity
I think this is particularly important as generative AI becomes part of more creative workflows. AI dramatically changes what one person can produce. This project would be a substantial undertaking for a traditional illustration and design workflow. AI assistance makes it possible for me to contemplate producing something of this scale independently.
But possible is doing a lot of work in that sentence. AI doesn't give me an extra eight hours in the day. It doesn't remove client communication, decision-making, art direction, continuity, layout, editing or revision. It doesn't guarantee that the complicated image will work on attempt one rather than attempt twenty-one. Most importantly, it doesn't turn a substantial publication project into a small job simply because one part of the production process has been accelerated.
A professional design studio can distribute those responsibilities across several people. Working independently, even with very powerful tools, I'm still one person.
Scoping can tell you when to stop
I've traditionally thought of good scoping as something that helps make projects happen. Define the deliverables. Identify the assets. Estimate the work. Find efficiencies. Build a production plan.
This project has taught me that scoping has another equally valuable purpose. Sometimes it tells you that the project shouldn't happen under the proposed constraints. That's not failure.
It's useful information. There is always the temptation with an exciting creative project to think:
I could probably manage it.
Maybe I work a little later or give up a weekend or move something else and hope the difficult bits behave themselves. And perhaps I could.
But that's different from making a professional commitment based on a realistic understanding of your capacity. The prototype gave me data and the revised script gave me scope. The production estimate gave me a boundary. Whatever happens next with the project, that lesson is already valuable:
Knowing how to make something is a creative skill. Knowing whether you can responsibly promise to deliver it is a professional one.