Prompting like an art director — one anatomy, four dialects — step 8 of 8

Tuesday morning: three briefs, three different tools. Write build_prompt(brief) that renders one brief dict into the right dialect.

Every dialect starts from the same core string:

core = f"{subject}, {medium}, {style}, palette of {palette}"

Then, by brief["tool"]:

• "midjourney" → core + " --ar 4:5 --no " + negatives where negatives is the list joined with ", ".
• "ideogram" → 'the headline "H" in bold display type, ' (H is brief["headline"]) + core + ". Do not include N." (N = joined negatives).
• anything else (Flux and friends speak plain prose) → core + ". Do not include N."

(Ideogram does ship a native negative-prompt field — prose negatives here keep the drill to one rule per dialect.)

Print {id}: {prompt} for each brief. Expected output:

P1: rooftop espresso bar at dusk, flat vector illustration, 2px uniform stroke, isometric, palette of terracotta, deep teal, off-white --ar 4:5 --no gradients, photorealism, watermark
P2: barista's hands mid pour-over, photograph, shallow depth of field, window light, palette of muted earth tones. Do not include text, plastic skin.
P3: the headline "SLOW SUNDAYS" in bold display type, sunrise over rooftops, flat vector illustration, halftone texture, palette of terracotta, deep teal. Do not include gradients.