The Forensic Pipeline

Discover exactly how Truth Lenses detects the undetectable. From zero-trust memory ingestion to high-speed multimodal neural scanning.

Secure Ingestion & Encoding

When you upload media, it never touches persistent storage. The image or video is immediately converted into a secure Base64 data stream in memory, authenticated via zero-trust JWT tokens, and prepared for neural analysis.

Forensic Parameterization

We inject a strict system prompt that forces the AI to adopt the persona of a forensic analyst. The prompt specifically instructs the engine to look for anatomical inconsistencies, diffusion model artifacts, unnatural lighting, and temporal flickering in videos.

Multimodal Neural Scanning

Powered by our proprietary multi-modal neural engine, the system processes the visual data alongside our forensic parameters. It scans at the pixel level for signs of Generative Adversarial Networks (GANs) and manipulation that the human eye cannot perceive.

Deterministic Verdict Output

The analysis is forced into a strict JSON schema. The output provides a definitive 0-100 Authenticity Score, a label (REAL, FAKE, or ALTERED), and an array of specific indicators pointing out exactly where and why the media was flagged.

truth-lenses-pipeline ~ ingestion.ts

1// Step 1: Secure API Ingestion
const prepareMedia = async (file: File) => {
  const auth = await verifyJWT(token);
  if (!auth.isValid) throw new Error("Unauthorized");
  
  // Convert to Base64 in-memory
  const base64Data = await convertToBase64(file);
  return {
    mimeType: file.type,
    data: base64Data,
    secure: true
  };
};

How Truth Lenses Works

Our multi-stage pipeline ensures accurate detection.