MOGGEDNYC CRANIOFACIAL PROPORTION ENGINE
LIVE

MOGGEDNYC V1.0 // LANDMARK MESH 68PT // CLIENT-SIDE PIPELINE // NO UPLINK

Computational craniofacial morphometry

Derives twelve classical proportion indices from a 68-point facial mesh: vertical thirds, golden-ratio span, interocular harmony, nasolabial segments, gonial vectors, bilateral symmetry residuals, and zygomatic–gonial breadth ratio versus reference ideals. Output is descriptive morphometry only.

Procedure

  1. 01 Frame face frontally; neutral expression; diffuse lighting.
  2. 02 Specify acquisition: library file, device capture, or live feed (localhost).
  3. 03 Mesh extraction and ratio quantization versus classical baselines.
  4. 04 Symmetry raster synthesis and harmony composite weighting.
  5. 05 Review dossier: radar breakdown, corpus heatmap, per-index deviation.

Image never leaves your device. All tensors resolved in-browser.

Acquisition channel

Tap the box or use Photo / Selfie in the bar below on phones.

Ingress buffer — front-facing raster

DROP · BROWSE · TAKE PHOTO / LIVE WEBCAM (HTTPS OR LOCALHOST)

TAP · BROWSE · TAKE PHOTO

Mesh overlay

Awaiting specimen. Use acquisition channel above.

Harmony composite

HARMONY — / 100

Weighted aggregate versus classical proportion baselines. Not diagnostic.

Appeal (experimental)

Appeal score

—/ 100

Heuristic estimate from a single photo. Not a fact about how people will treat you.

Confidence

—

    To get a cleaner reading

      What helped

        What pulled it down

          How we score (transparent)

          This uses a transparent weighted blend of symmetry + proportion/ratio cues. It does not use your photo outside this device. 

          
          
          
            Sources used to choose factors and framing:
          
          
          
            
        
            
      
            
    
            

    
            
      
            Telemetry
            
      
            
        
            
          
            Radar — index deviation
            
          
        
            
        
            
          
            Corpus heatmap
            
          
            Bright: lower mirror residual. Dark: higher asymmetry signal.
            
          
        
            
      
            
    
            

    
            
      
            Dossier — per-index readout
            
      
            
    
            

    
            
      
            Ask Clav
            
      
            
        Checking which assistant is active…
      
            
      
            
        
        
      
            
      
            
        On-device: no API key, no network for chat — pattern-based answers only.
         Claude: put your Anthropic key in the environment as ANTHROPIC_API_KEY and run
        npm start from this folder (not plain Python). The browser never sees your key.
      
            
      
            
        Questions about your scores in plain English — plus the math when you want it. On-device mode never sends data out;
        Claude mode sends only your text + numeric summary to Anthropic’s API (still no image).
      
            
      
            Tap to ask, or type anything below: