How to illustrate a concept with AI
Background
A friend, a very A conscientious one, is about to buy a flat, and that, for her, is an uncertainty-inducing situation.
I thought I would send her an encouraging postcard. Open Midjourney and try:
A courageous cat confronts a monster.one point perspective. Cat from behind in the lower center. The monster a large grey mass in the background.the feeling should be that the cat will win even if the monster is really monstrous
OK: Cat, monster, check. But no "overcoming challenge".
Second try: rerun
Not any better, really.
Third try: fire up Claude
Can you think of something better? I usually render with Midjourney.
Your prompt fails because it tells Midjourney how to feel instead of what to see. "The feeling should be that the cat will win" is instruction, not image. And "a large grey mass" is a literal description of a boring grey blob — MJ gave you exactly what you asked for.
Three fixes:
Don't depict the monster. Imply it. Uncertainty is a visual property: fog, silhouette, something half-resolved, something with no face. The moment MJ renders a full creature, the uncertainty is gone and you get a Halloween cartoon.
Courage lives in the body and the light. Tail straight up, ears forward, paws planted, long shadow thrown toward the threat. Put a warm light source behind the cat so it gets a gold rim against a cold grey world. That single contrast does more than any adjective.
Scale, not size. Make the cat tiny — bottom 10–15% of the frame. Small-against-vast is the whole emotion. Your version has the cat and monster competing for the same space, so neither reads.
Here's the prompt I'd run:
small tabby cat seen from behind, tiny in the lower third of the frame, standing planted on wet asphalt, tail raised straight up, ears forward. ahead the street dissolves into thick grey fog, and within the fog an enormous dark shape only half resolves — an outline, a suggestion of shoulders, no face. low warm streetlight behind the cat casts a long thin shadow toward the fog and rims its fur in gold. one point perspective, low horizon. painterly gouache illustration, cold grey palette against one warm light source, heavy negative space --ar 4:5 --s 400
Two variants worth trying with the same bones:
Graphic/poster — swap the paint for a screenprint: three colours only: bone white, deep indigo, one stroke of orange. thick grain, flat shapes, strong diagonal light. Works better if this is going on a blog header.
Cinematic — shallow depth of field, cat sharp and in focus, colossal blurred silhouette behind in volumetric haze, unreadable and soft-edged. muted teal and ash --ar 21:9 --s 250. The blur is doing the "uncertain" work here instead of the fog.
One more lever: the threat doesn't have to be a creature at all. A wall of storm front, a wave, a mountain that wasn't there yesterday — those read as "menacing future" more honestly than a monster does, because you can't fight them, only face them. Try replacing the shape with a towering wall of slow-moving cloud swallowing the street and see what you get.
I think the third one is pretty good. Except for the alignment along the central axis.
This one – same prompt, but landscape – is quite good, too. The menace is a bit too wishy washy, though.
Next try: Lose the axis
I like the aesthetics of the cinematic still but the center alignment feels like a waste of space. What other "blocking" would show the "small against dangerous" yet use the frame?
Move the cat off-center and give it a trail. Bottom-left third, mid-stride, with a visible wake behind it — footprints in wet ash, a broken reflection, a path already crossed. That's your Gaiman line, literally: the frame shows the cat has already come a long way toward the thing. It isn't sitting and contemplating, it's closing distance.
That also fixes your dead space. In the centered version, everything left and right of the cat is doing nothing. With this blocking, the left edge holds the trail (past), the cat sits on the left thirds line (present), and the whole right side is the mass it's walking into (future). Every part of the 21:9 has a job.
Three things that do most of the work:
Mid-stride, not seated. One front paw lifted, body weight forward. Sitting reads as awe. Walking reads as decision. This alone changes the whole sentence.
Put a scale object in there. A lamp post, a car roof, a road sign half-swallowed by fog. Right now your monster is infinitely big and infinitely vague, which means it's actually unmeasurable — and unmeasurable threats can't be beaten. Give the viewer a ruler and the mass becomes "very large" instead of "cosmic." Beatable.
Let the light win somewhere. The dark should fray where the warm light touches it, not sit as a hard wall. A thin gap of light on the horizon, a seam in the mass. One small visual promise that it isn't solid.
The prompt:
small tabby cat seen from behind, tiny, positioned in the lower left third, mid-stride with one front paw lifted, walking rightward across cracked wet ground. a trail of wet pawprints and broken reflections leads back to the left edge of frame. ahead and to the right, an enormous slow wall of dark cloud fills two thirds of the frame, half-swallowing a leaning street lamp and a road sign for scale. a thin seam of warm gold light breaks the horizon where the dark frays. cinematic still, shallow depth of field, cat sharp, background soft volumetric haze. muted teal and ash with one warm light source --ar 21:9 --s 250
Two variants on the same blocking:
Gateway — keep the cat lower-left, but add dark ruined verticals at both frame edges in the foreground, out of focus. The cat walks through the gap between them. Uses the edges instead of letting them fade, and makes the viewer feel like they're standing behind the cat rather than floating.
Mirrored diagonal — cat lower-left, mass entering as a wedge from the top-right corner. The empty diagonal band between them is the ground still to cover. Cleanest, most poster-like, and it reads instantly at thumbnail size.
One note on your portrait image: that giant shape is unmistakably a cat silhouette. Probably accidental, but it's the strongest thing in the frame — it turns the monster into the cat's own shadow. If you want "you can beat them," that's a gift. Lean into it rather than fixing it.
Midjourney was a bit overwhelmed by the composition in this prompt:
Switch to GPT image
A lot better. Let's see if Claude has an idea how to improve it.
One fix matters more than all the others: the tail is down. That's the single gesture that contradicts everything else in the frame. A cat with a low, trailing tail is anxious or submissive — it's the universal cat signal for "I am not sure about this." Vertical tail, slight hook at the tip, is confidence. You've built a whole image about courage and then put a nervous tail on the protagonist. Fix that and the picture changes meaning without touching anything else.
If you're editing rather than regenerating, one instruction does most of it:
Raise the cat's tail to a confident vertical with a slight hook at the tip. Lift its head so it is looking up at the storm wall ahead. Make the cat a little sturdier and less bedraggled. Add a narrow seam of warm gold light breaking through the dark cloud on the far right, small and distant. Extend the cat's wet reflection across the lower right foreground.
You lost the stride. All four paws are planted. It's standing and looking up, which reads as awe — "look at that." The previous version had a lifted paw and read as advance — "here I come." Awe is a fine image, but it isn't the Gaiman line. Get the front paw back off the ground and you have both: moving forward, head up, tail up. That's the full sentence. (...)
Edit instruction:
Lift the cat's front right paw off the ground into a mid-stride walking pose, body weight forward. Restore a clear trail of wet pawprints curving from the cat's feet off the bottom-left corner of the frame. Replace the bright starburst at top right with a narrow horizontal seam of warm gold light low in the storm wall, near the cat's eye level, where the dark cloud frays open.

Aside from the fact that those are no longer cat paw prints, it’s a pretty cool picture.
- ← Previous
This is my fourth post