Same grotesque-comic prompt fed to three different models, the styles came out wildly different

Got curious and ran a little comparison: same prompt, thrown at three different models. The ask was to draw a game character in that extremely exaggerated, near-grotesque Western satirical comic style, and I specifically named a certain veteran cartoonist’s dense, rough, and quirky linework as a reference.

The three results came out wildly different — one nailed that manic energy, lines chaotic but full of character, another turned out as generic moe-style illustration, completely missing the point of 'grotesque.

’ Same prompt, this much interpretation gap, really shows how each model’s sensitivity to style words differs. Images are below, which one do you think nails it?

Same prompt across 3 providers really does vary a lot. The more exaggerated the deformation you want, the more some models tone it back down.

Post the three sets side by side so we can actually see, just saying the gap is big isn’t satisfying :eyes: