One park, one language — compiled into prompts

01  The System

Prompts as a design system, not a folder of text

Every render of Al Adawi Park is generated from one grammar: a fixed subject library, six groups of brand-DNA tokens, a site-accuracy clause bound to the real 109-vertex boundary, and a universal negative prompt that keeps Dubai, fantasy and war imagery out. The compiler below assembles them the same way every time — which is why 100 images look like one park.

02  Prompt Compiler

Compose a render — the prompt writes itself

Choose a subject and conditions; the compiler assembles the full production prompt — subject, Damascene context, scale cues, quality clause, site-accuracy rule and negative — in your engine's dialect. Add DNA tokens from below for extra flavour.

03  Brand DNA

Forty-eight tokens — click to add to the compiler

The vocabulary the park speaks in. Each chip is a tested fragment: real materials from the Keturah Stones catalogue, the verified planting palette, Damascus light, honest people. Click to inject into the compiled prompt; click again to remove.

04  The 100

The production library — ten sets, one park

The master production run from the project's own prompt bible: 100 unique views, no repeats, every one bound to the same boundary and language. Search, filter by set, expand, copy.

06  Export & Use

Take the whole system with you