Crush analysis that asks more, shows more, and feels more real

LensCrush now uses a deeper quiz to make the result feel more accurate, then optionally blends in a browser-side TensorFlow.js MobileNetV3 visual signal from an uploaded image. The output is displayed in pie-chart style panels so visitors get more than a single random number.

Guided question flow
MobileNetV3 signal
Canvas pie charts
Live LensCrush preview Start answering and LensCrush will build a stronger real-time reading as you move through the quiz.
Mobile-first flow Step-by-step quiz AI-assisted finish
Loading MobileNetV3... The model runs in your browser, so uploaded images do not need to leave the page during local use.

Compatibility questionnaire

Answer one question at a time so the flow feels more guided, then let LensCrush process the full set with the optional MobileNet visual signal.

Step 1 of 10 Names

Start with the two names

LensCrush will use these names in the final reading and score summary.

Important note

This page uses a real TensorFlow.js model for image classification in the browser, but the result is still an entertainment reading. A classification model cannot truly predict romantic compatibility, so LensCrush blends it with quiz answers and presents everything as a playful experience.

Does LensCrush store your quiz answers? The experience is designed to stay lightweight and privacy-conscious. See the privacy page for the current browser-side setup.
What should you do after the result? Use the score as a playful prompt, then keep reading with the blog guides and daily tips for better next steps.
Is LensCrush meant to be scientific? No. LensCrush is an entertainment-focused relationship tool with clearer context, disclaimers, and follow-up guidance.