Turn a vague request into an image prompt

Turn this vague image request into a usable prompt.

**Step 1 — What the request does not decide.** List every undetermined slot, with a recommended default so each can be answered quickly:
- Medium: photograph, illustration, 3D, flat graphic
- Subject: who or what is in frame, and how many
- Setting and time of day
- Camera distance and angle
- Mood, stated as something checkable rather than as an adjective
- Palette
- Aspect ratio and where it will be placed
- What must NOT appear

**Step 2 — Decode the adjectives.** For each vague word in the request ("clean", "premium", "warm", "modern"), give two or three concrete renderings it could mean, and ask which. *These words feel specific to the person saying them and mean nothing to a generator.*

**Step 3 — The prompt**, with every assumed value marked `[assumed]`.

**Step 4 — The one question** whose answer would change the image most. If only one thing can be asked, ask that.

Rules:
- Do not guess the brand or the industry's conventions unless I told you.
- Say when the request implies something that cannot be generated reliably — exact text, a real person, a specific existing product.