Idea
Lines of Fate explores how simple, instinctive doodles can become gateways to symbolic meaning. Users think of a personal question and draw three spontaneous shapes. Each doodle is captured, first through basic input, later via hand-tracking and interpreted through a small neural network that assigns it to one of several symbolic shape categories. Each category corresponds to groups of tarot cards. From these, the system selects a card to represent the user’s past, present, or future, and a language model generates a brief summary and a detailed interpretation. The goal is to transform fleeting gestures into poetic digital divinations.
Abstract (Idea's complex version)
Lines of Fate is an interactive art project exploring gesture, symbolism, and divination. In this experience, users think of a personal question and draw three doodles, each recorded separately. The system can use a touchpad or, if feasible, ml5.js Handpose to capture the drawings. A neural network and image classifier then interpret each doodle, and each interpreted input triggers the selection of a tarot card from the full 78-card deck, representing aspects such as the past, present, and future of the user’s question. Once all three cards are selected, a language model generates both a concise one-line summary and a more detailed textual interpretation.
Users can save their brief daily reading, which contributes to a “personal energy map” visualizing patterns across multiple sessions. Optional enhancements include sound feedback reflecting the mood of each card, and, if time permits, physical elements such as LED lighting controlled by Arduino to enhance the ritualistic atmosphere. By combining doodle-based input, tarot symbolism, and AI-generated interpretation, Lines of Fate transforms ephemeral actions into meaningful, poetic, and personal digital divinations, bridging the interactive and mystical.
Inspiration
Mysticism + Everyday Gesture
I wanted to combine the quiet ritual of tarot with the immediacy of doodling, which is something everyone does unconsciously. Tarot is symbolic and interpretive, and doodles are equally expressive. Blending them felt natural and rich with possibility.
Human Gesture as Meaningful Input
Instead of designing a mechanical controller, I was fascinated by how small, unintentional motions that can carry emotional tone. The project explores how technology might read these gestures not literally, but symbolically.
Machine Learning as a “Modern Oracle”
I was curious how neural networks, hand-pose tracking, and LLMs could mimic the logic of divination. Not in a deterministic sense, but as a system that interprets signs, suggests patterns, and invites reflection.
Audience
People who enjoy symbolic or introspective experiences
Anyone who likes tarot, astrology, journaling, or reflective daily rituals will enjoy the poetic, personalized nature of the readings.
Users curious about interactive art and playful AI
The project speaks to students, creatives, and technologists who want to experience how doodles, gesture recognition, and machine learning can blend into a single expressive interface.
Challenges
Reliable Doodle Interpretation
Free-form drawings are messy and varied. Ensuring that the classifier assigns them to symbolic categories without feeling arbitrary is a core challenge.
Choosing a Stable, Expressive Input Method
I’m still deciding whether to rely on keyboard/touchpad drawing for stability or implement handpose tracking for a more performative, embodied interaction.
Meaningful Mapping Between Gestures and Tarot
The biggest design question is How do I pair abstract shapes with symbolic card groups in a way that feels intuitive, coherent, and not too literal.
Plan
Phase 1 — Build the pipeline
Prototype with simple keyboard/touchpad doodle input; use a small set of images and categories to map doodles → symbol groups → cards → text output.
Phase 2 — Tarot + LLM Integration
Add the complete tarot deck and generate personalized interpretations using a language model.
Phase 3 — UI/UX Refinement
Clean up the interface, make the drawing process smoother, and design the layout for the three-card spread.
Phase 4 — Saving + Audio System
Implement daily-reading saving and simple sound feedback tied to mood or card energy.
Phase 5 — Add Handpose Tracking
Replace or augment doodle input with gesture-based drawing. Test stability and adjust interaction design.
Phase 6 — Polishing (If time allows)
Add optional extras such as LED ritual lighting or more expressive animations.