🤯 How to Reverse Engineer ANY Image into a Perfect JSON Prompt (For FREE)

Yup, that’s how you steal viral image prompt 👇🏻👇🏻

🤯 How to Reverse Engineer ANY Image into a Perfect Prompt (For FREE)

I stopped guessing prompts months ago. Here is the new workflow.

THE TRUTH 🕵️‍♂️

Most people stare at an AI image and guess: “Uhh, cinematic lighting, cute cat?”

That fails. The pros use “Forensic Reverse Engineering.”

I built a free tool inside Google Gemini that analyzes any image (pixel-by-pixel) and gives you the mathematically perfect JSON data to recreate it.

Here is how to set it up in 60 seconds:

1.🏁 QUICK START GUIDE

Step 1: Go to https://gemini.google/

Click “Gem manager” on the left sidebar.

Step 2: Create a New Gem

Click “New Gem” → “Create from scratch.”

Step 3: Name It

Name: “Vision-to-JSON”

(Icon: Pick the little code bracket symbol </>)

Step 4: The Secret Sauce 🥫

Paste the forensic instructions into the “Instructions” box.

(This is the magic part that forces Gemini to act like a computer vision expert).

Step 5: Save & Use

Click Save. Now, just upload ANY photo to this Gem. It will spit out the exact camera lens, lighting angles, and style code.

2. 👇 WANT THE CODE?

Here are the instructions to paste into your Gemini Gem!

INSTRUCTIONS:

Role: You are a Lead Forensic Image Analyst and AI Prompt Engineer.

Task: Analyze the visual content of the uploaded image with forensic precision. You must focus exclusively on extracting objective, measurable, and highly specific visual data points. 

Output Format: The output must be a single, strictly structured JSON object. Do not include any introductory text, concluding remarks, or markdown formatting (like ```json). Just the raw JSON.

JSON Structure Requirements:

1. “metadata”: Dimensions, aspect ratio, and dominant artistic style.

2. “technical_specs”: Camera model (simulated), lens type (e.g., 85mm portrait), f-stop, shutter speed, ISO, and film grain details.

3. “lighting”: Key light direction (clock face position), color temperature (Kelvin), contrast ratios, shadow hardness, and atmospheric effects.

4. “subject”: Detailed breakdown of the subject (pose, clothing texture, facial features, micro-details).

5. “composition”: Rule of thirds analysis, depth of field, background blur level, and viewing angle.

6. “colors”: Hex codes of the primary palette and grading style.

7. “generated_prompts”: 

   - “exact_prompt_for_flux”: A natural language descriptive prompt optimized for Flux AI.

   - “exact_prompt_for_midjourney”: A prompt optimized for Midjourney including --ar (aspect ratio) and --stylize parameters.

Constraint: Your analysis must be purely descriptive and factual. Avoid generic terms like “beautiful.” Use technical terms like “subsurface scattering,” “volumetric lighting,” and “chromatic aberration” where applicable.

3. How to Edit the JSON Prompt?

Use this free tool
https://jsoncrack.com/editor

With this app you can easier see the Json prompt in a structured diagram, and you can edit anything you want, finetune it.

💡 PRO TIPS

• Use this to clone expensive ad styles for free.

• Great for recreating specific photography vibes (e.g., “1990s Polaroid” or “CGI Render”).

• The output gives you ready-to-copy prompts for any image generator including Nano Banana, GPT-Image, Midjourney AND Flux.

📉 THE BOTTOM LINE

Stop wasting credits guessing. Let the AI write the prompt for the AI. It creates a perfect loop. ♾️

You can use the same instructions in GPT or perplexity.

But for the best result Gemini 3 Pro has the best image analysis model. So it can really see what’s inside the image.

So I use Gemini Gem for this demo.

Prompts from the Demo:

{ "metadata": { "dimensions": "1080x1350", "aspect_ratio": "4:5", "dominant_art_style": "Hyper-realistic commercial portrait photography", "overall_mood": "Playful, narcissistic, warm, joyful, introspective" }, "color_palette": { "dominant_colors": [ "#EADBC8", "#F5Ccb0", "#121214", "#FDFDFD", "#8B2E31", "#2B3A55" ], "accent_colors": [ "#A65E5E", "#3E4C68" ], "gradient_description": "Smooth, creamy gradient in the background transitioning from a lighter warm beige on the left to a slightly darker tan on the right, mirroring the falloff of the key light." }, "lighting": { "key_light_direction": "Upper Left (approx 10 o'clock position)", "temperature": "4800K (Warm Neutral)", "softness": "Very soft, likely a large octa-box or softbox modifier", "fill_light_ratio": "1:3 (moderate fill to keep shadows open but dimensional)", "rim_light_presence": "Subtle cool rim light on the top right hair texture", "ambient_color": "Warm Beige", "shadow_hardness": "Soft, diffused falloff, particularly under the nose and chin", "global_contrast_level": "Medium-High, optimized for commercial clarity" }, "composition": { "subject_placement": "Man's face centered in the left vertical third; Hand and figurine centered in the right vertical third", "leading_lines": "The angle of the man's forearm leads the eye upward to the figurine; the man's gaze creates an invisible vector line directly to the toy face", "symmetry": "Asymmetrical balance between the large human face and the small detailed object", "negative_space": "Generous negative space in the upper right quadrant, clean background", "depth_layers": [ "Foreground: Hand and figurine", "Mid-ground: Man's face and upper torso", "Background: Blurred beige seamless backdrop" ] }, "camera": { "focal_length_equivalent": "85mm (Portrait telephoto)", "aperture_effect": "f/4.0 (Sufficient depth of field to keep both face and figurine relatively sharp while blurring the ears and shoulders)", "lens_type": "Prime Macro or Portrait Lens", "camera_angle": "Eye-level relative to the human subject", "distance_to_subject": "Approximately 1.5 meters" }, "subjects": [ { "id": "human_male", "type": "person", "appearance_attributes": "Asian male, approx 30 years old, short black hair with gelled texture, light skin with visible pores and slight stubble, faint mole under left eye", "pose": "Head turned slightly right, looking down at hand, right hand raised holding object, smiling with teeth visible", "clothing_specifics": "White polo shirt, pique cotton texture, collar with double stripe trim (red and navy blue), top button fastened", "expression": "Amused, proud, affectionate smile, crinkling at eye corners", "normalized_coordinates": "0-100% coverage, centered left", "relative_size": "Dominant (occupies 60% of frame)", "verbose_description": "An adult Asian male is captured in a close-up portrait. He has short, straight black hair styled upwards with a matte product. His skin tone is fair to warm tan, showing realistic texture including pores on the nose and cheeks, fine lines around the smiling eyes, and a very light shadow of facial hair on the chin and upper lip. He is wearing a white pique cotton polo shirt. The collar features a distinctive twin-tipped design with a thin red line and a thin blue line. He is looking affectionately at a small object held in his right hand. His expression is one of genuine delight, with a soft, open-mouthed smile exposing the upper teeth." }, { "id": "miniature_figurine", "type": "toy/object", "appearance_attributes": "3D printed style chibi figure, large head, cartoon eyes, plastic texture", "pose": "Standing, arms raised in a 'V' for victory or cheering motion, legs apart", "clothing_specifics": "White polo shirt matching the human, dark blue trousers, black shoes", "expression": "Exaggerated happy screaming/cheering face, open mouth, arched eyebrows", "normalized_coordinates": "Held in hand at approx 60% horizontal, 50% vertical", "relative_size": "Small (approx 5% of frame)", "verbose_description": "A small, stylized 'chibi' figurine constructed from what appears to be vinyl or 3D-printed resin. The figure is a caricature of the man holding it, featuring the same black spiky hair, a large head relative to the body, and large expressive cartoon eyes. The figure is wearing a miniature version of the man's outfit: a white polo shirt with the same red and blue collar trim and a small laurel wreath logo on the chest. The figure also wears dark blue pants and black shoes. Its arms are thrown up in the air in a gesture of excitement or triumph. It is being held gently between the man's thumb and middle finger." } ], "background": { "type": "Studio Backdrop", "description": "A seamless, solid-colored background in a shade of warm latte or beige. The background is rendered out of focus due to the depth of field, creating a smooth, non-distracting field of color that complements the skin tones. There is a slight gradient of light, brighter on the left and falling into shadow on the right.", "visible_text": "None", "environmental_details": "Indoor studio setting, controlled artificial lighting, no weather elements." }, "post_processing": { "film_grain": "Low/Negligible", "vignette_strength": "Very subtle darkening at the bottom corners", "color_grading": "Commercial standard, slightly warm bias, high skin tone fidelity", "sharpness_level": "High on the eyes and figurine, rapid falloff elsewhere", "chromatic_aberration": "None visible" }, "micro_details": [ "Specular highlight in the man's right eye reflecting a rectangular softbox", "Fine knit texture of the pique cotton shirt fabric", "Subtle veins visible on the back of the man's hand", "The tiny laurel wreath logo printed on the figurine's chest", "Individual strands of hair sticking up on the man's crown", "Skin texture details including a small mole on the left cheek", "The way the light catches the ridge of the figurine's plastic hair", "The fingernail of the man's thumb showing a clean, matte finish", "The fold of the ear cartilage", "The separation of the teeth in the man's smile" ], "exact_prompt_for_nano_pro": "Photorealistic close-up portrait of a smiling Asian man in a white polo shirt holding a tiny 3D-printed cartoon chibi figurine of himself between his fingers, soft studio lighting, beige background, sharp focus on face and toy, 8k resolution, high texture detail.", "exact_prompt_for_flux": "A high-quality photo of a young Asian man holding a miniature 3D avatar of himself, the man is smiling warmly at the toy, he wears a white polo with red and blue collar trim, the toy is in a cheering pose with arms up wearing the same outfit, studio photography, soft beige background, depth of field.", "exact_prompt_for_midjourney": "A hyper-realistic portrait of a man looking at a miniature figurine of himself held in his hand, Asian male subject, white polo shirt, playful concept, macro photography, soft studio lighting, 85mm lens, beige backdrop --ar 4:5 --stylize 250 --v 6.0" }