DeconstructAI
🪙3

Convierte cualquier imagen o video en prompts y guiones listos para producción con IA

Deconstruye fotos de Pinterest e Instagram o videos de TikTok y Reels en guiones virales, prompts para Kling / Runway y recetas para Midjourney y ComfyUI en segundos.

Probar ejemplo:
3 Free Generations
No Credit Card Required
HD Resolution
Original Scraper Source
3 Model Engines
SDXL • Midjourney • ComfyUI
< 2.5s
Average Deconstruct Speed

The Reverse Engineering Pipeline

From Social Post to Production AI Model

01

1. Paste Post URL

Drop any public Pinterest pin or Instagram photo link directly into the prompt decoder.

02

2. Multi-Modal Vision Analysis

Gemini 2.5 Flash analyzes composition, lighting ratios, subject textures, lens optics, and aesthetic tone.

03

3. Export Ready-to-Use Recipes

Get clean SDXL positive/negative tags, Midjourney v6 parameters (--ar, --stylize), and downloadable ComfyUI nodes.

Engineered For Every Workflow

Tailored Prompts For Leading AI Generators

Stable Diffusion & SDXL
Weighted Tokens

Formatted with attention brackets (tag:1.2), Danbooru tokens, negative quality sanitizers, and photorealism tokens optimized for SDXL 1.0, Juggernaut XL, and Realistic Vision.

(photorealistic:1.2), raw photo, 85mm lens, golden hour rim lighting, 8k resolution...
Midjourney v6.0
Natural Language

Eliminates keyword stuffing in favor of rich, evocative cinematic syntax, aesthetic framing descriptors, and automated flag parameters like --v 6.0, --ar 4:5, and --style raw.

cinematic editorial street photography, Hasselblad medium format --v 6.0 --ar 4:5...
ComfyUI Graph (.json)
Plug & Play

Downloads a complete node graph incorporating KSampler, positive and negative CLIP encoders, latent space conditioning, and VAE decode with optimal steps and CFG settings.

{ "nodes": [ { "type": "KSampler", "widgets_values": [30, 7.0, "euler"] } ] }

Frequently Asked Questions

Everything About AI Image Reverse Engineering

How do I get a Midjourney prompt from a Pinterest picture?

Simply copy the URL of any public Pinterest pin (including pin.it short links) and paste it into Deconstruct. Our scraper retrieves the highest-resolution original asset, passes it to Google Gemini 1.5 Flash Vision, and converts the aesthetic, lighting, and composition into a Midjourney v6-ready prompt formatted with appropriate aspect ratios and style tags.

Can I convert Instagram photos into Stable Diffusion SDXL prompts?

Yes! Paste the public Instagram post URL. Deconstruct isolates the photo subject, outfit, camera lenses, and environment to generate Danbooru-weighted tokens specifically designed for Stable Diffusion and SDXL checkpoints.

How does the ComfyUI workflow JSON download work?

When an image is deconstructed, click the "Download ComfyUI Workflow (.json)" button. You'll receive a standard ComfyUI schema graph with pre-configured positive and negative CLIP text encodes, Checkpoint loader, and KSampler nodes. Drag and drop the downloaded JSON directly onto your ComfyUI canvas to run immediately.

What is the difference between Plan Free and Starter Pack?

The Free tier provides 3 deconstructions per session with standard downloads. For creators needing volume, our affordable credit packs (starting at $2.99) provide permanent image and video prompt deconstructions with zero ads and credits that never expire.