Guide: Midjourney Prompts from Images

Midjourney v6 and Black Forest Flux are incredibly capable image generators, but they require precise vocabulary to achieve specific artistic styles, lighting setups, and compositions.

Why Reverse Engineering Works

Instead of guessing which keywords will trigger the desired style in Midjourney, you can use ProPeek to reverse-engineer an existing image that already has the look you want. Vision AI models are excellent at identifying specific art styles, rendering engines (like Unreal Engine or Octane Render), and photography techniques.

The ProPeek Image Workflow

1. Upload Your Reference

Drop any image into ProPeek. This could be a photograph, a 3D render, a vector illustration, or an oil painting.

2. Image to Prompt Analysis

ProPeek analyzes the image and constructs a prompt that typically includes:

• Medium: (e.g., "Macro photography", "3D CGI character design", "Flat vector illustration").
• Subject Details: Highly specific descriptions of the subject's appearance and posture.
• Lighting: "Rembrandt lighting", "Rim lit", "Soft diffused sunlight".
• Technical Specs: "Shot on 85mm lens", "f/1.8", "Depth of field".

Prompt Lab for Tweaks

Once you have the base prompt, you can use ProPeek's Prompt Lab to refine it. Use the "Variations" tool to instantly generate 5 slightly different versions of the prompt, allowing you to A/B test which one Midjourney prefers.