TalkSnap AI

InspirationThe inspiration came from:

The rise of virtual influencers and digital humans

The popularity of tools like Ready Player Me and Metahuman

The desire to build an AI system that combines vision + 3D + voice + conversation

Exploring how AI can make digital avatars instantly accessible to everyone, not just professional studios

I believed there should be an easy, intuitive experience where anyone can upload a simple photo and immediately interact with a 3D version of themselves.

What it doesThe final result is a full interactive AI experience:

Rotate, view, customise avatar

Speak or chat with your AI assistant

Download your avatar image

It’s a great demonstration of what’s possible when combining 3D graphics, AI, and modern web experiences.

How we built itReal-Time AI Conversations

The chat system uses an AI model to generate natural responses

The assistant can reply in text or simulated voice

The UI provides both typing and microphone-style interaction

Modern Web Frontend

Built with HTML, CSS, and Vanilla JavaScript

Three.js is used for rendering and animation

Smooth UI with styled components, prompts, loading states, and previews

Backend API (Node.js / Express)

Handles file uploads

Sends photos to the avatar API

Returns a ready-to-load 3D model URL

Sends chat messages to an AI processing endpoint

Challenges we ran intoUsers may upload:

Large images

Unsupported formats

Blurry or low-light selfies

Pictures with multiple faces

The issue

Bad input → avatar API fails → confusing error.

Fix

Limited uploads to < 5MB

Added file-type validation

Displayed clearer instructions and error messages

Accomplishments that we're proud of

What we learnedWorking with multipart image uploads

Polling asynchronous AI API tasks

Handling real-time user interactions in a conversational UI

Designing smooth UX for AI workflows

What's next for TalkSnap AI

Built With

• apikey
• openai