DVP Tools / Prompt Engineering

Structured JSON Prompt Builder

Bridge the gap between natural language and structured AI reasoning. Compile precise JSON prompts for Gemini, GPT-4o, and FLUX.1 with total control over cinematography, lighting, and optics.

Start building Quick start Schema Logic

Quick start

Build your schema in three moves

JSON Prompt Builder is a free browser tool. It translates your narrative description into a structured object that AI models can interpret with near-perfect adherence.

Describe the Scene

Type your vision in plain English. The tool auto-detects subjects, lighting, and weather, mapping them instantly to the JSON schema on the right.

Inject Style Chips

Use preset chips to layer cinematic styles like Kubrick Symmetry or CineStill 800T. These inject complex optics and grading parameters into your code.

Export and Deploy

Copy the live JSON and paste it into Google AI Studio, ChatGPT, or your custom API pipeline. The model will respect the structured hierarchy of your request.

Methodology

Schema Architecture

The JSON output follows a hierarchical structure designed to provide high-signal data to LLM-based image generators.

The Core Dimensions

Your prompt is broken down into modular objects, allowing the model to process specific technical requirements independently.

Scene Subject, atmosphere, and narrative context.
Camera Focal length, aperture, and ISO settings.
Lighting Intensity, direction, and color temperature.
Composition Rule of thirds, symmetry, and framing.
Rendering Aspect ratio, resolution, and film stock.

Why Use JSON?

Modern models like Gemini and GPT-4o are trained on code. Providing prompts in JSON format reduces "token drift" and ensures the model weights technical specs correctly.

Tip

For FLUX.1 or Midjourney, extract the "prompt" string value for maximum adherence to the generated technical metadata.

Controls

Mastering the technical override

Visual Style Presets

Instantly load director-specific palettes. From the green-noir of Fincher to the pastel symmetry of Wes Anderson, these chips set the baseline for your JSON grade.

Optics & Glass

Control the Aperture for shallow bokeh or select Anamorphic to inject oval flares and barrel distortion into the prompt logic.

Lighting Rig

Override auto-detection to specify Golden Hour, Rim Light, or Hard Shadows. This forces the AI to prioritize the lighting engine.

Film Emulation

Choose from Kodak Portra 400, CineStill 800T, or Kodachrome. Each triggers specific grain and color response curves in the rendering schema.

Character Consistency

Use the Identity ID field to maintain character likeness across multiple generations—a critical feature for storyboarding and serial content.

Aspect Ratio

Switch between 21:9 Cinematic, 16:9 Widescreen, and 9:16 Vertical. The JSON handles the math for the target model automatically.

Professional Workflow

Production Pipelines

LLM System Prompts

Paste the JSON directly into the System Instruction of Gemini Pro. Tell the model: "Generate an image using the following technical JSON schema."

Consistency is Key

When working on a series, keep the Camera and Rendering objects identical. Only modify the Scene object to ensure visual continuity across your shots.

Iterative Refinement

If the result is too "digital," use the Film Stock chips to add organic grain. The JSON structure makes it easy to spot which variable needs adjusting.

API Integration

Download the .json file to use as a template for your own Python or Node.js scripts. It provides a clean baseline for automated image generation.