Version 2 · 2026-07-31 · Round 2
Round 1 leaned into emoji-cute and knowingly broke the written brand rules. This round obeys them. Ten directions that stay inside the original commission brief: line-driven, editorial, adult, restrained, in Fox Russet with cream and ink. Sharp, stylish, warm, calm; a New Yorker spot illustration, never a sticker. Same discipline as before: every direction drawn twice, settled portrait and the peek, because a character must survive re-posing.
The constraint
The three brand docs agree on a register: competent without being smug, stylish without
being precious, warm without being saccharine. So this round's shared style contract
swapped Round 1's cuteness machinery for the brief's own rules: modest slightly-tilted oval
eyes instead of big rounds, adult proportions instead of big-head charm, ears at 30 degrees,
the tail doing silhouette work, Fox Russet #C8552D on every candidate.
The peek pose survived the translation, and that is the round's most useful finding. It just changed meaning: instead of a fox popping up to say hello, he is watching, composed and unhurried. The one in the group chat who calmly says "third drawer, behind the spatulas" would peek over a wall exactly like this. The whimsy the first round got from cuteness, this round gets from situation: a dignified animal calmly surveilling your onboarding is quietly funny on its own terms.
The face-off stage
Same live test as v1, now with both philosophies in one picker: the ten brief-faithful foxes, plus the four Round 1 leaders for direct comparison. Flip between a cute one and an editorial one on the identical screen; that flip is the real decision this exploration exists to make.
Step 2 of 4
One photo. He catalogues what is in it, so you can ask for it later in your own words.
Round 2 · the brief's fox
Round 1 leaders · the cute fox
The field
Round 1's lesson applied: directions diverge by construction, not by adjectives. Ink line, clear line, colour blocks, silhouette, geometry, continuous line. Each card notes how tightly the result honours the brief, because that is this round's whole premise.
The 120px test
The quiet headline of the round: all ten survive at 120px, where Round 1 lost three of ten. Editorial economy is not just taste; it is engineering. When a drawing spends nothing on fur and blush, everything it does spend survives miniaturisation.
Strongest at this size: GM1, MC1, ZB1 and SL1, the four with the boldest shape economy. CO1's single line thins out furthest, as a single line must.
Where this leaves the choice
The exploration has done its job when the decision is clean. It now is:
A workable middle exists and two candidates already point at it: GM1 keeps geometric warmth close to cute without crossing into it, and ZB1 is the most ownable single drawing across both rounds, the one a competitor could not accidentally reproduce. If one from each round goes to a head-to-head refinement (side-peek drawn for real, a "found it" pose, a pre-coffee pose), the choice will make itself on the screens.
Both rounds reproduce from generate_fox.py in the vault record. This round:
20 renders, roughly $1.70. Running total: 40 renders across two rounds.