System Requirements Document (SRD) for Mango-Torch

1. Introduction

This document outlines the system requirements for the Mango-Torch project, an AI-powered image editing tool designed for users in the UAE (AE) and beyond. Mango-Torch leverages advanced machine learning models to provide a seamless and intuitive interface for editing images based on user prompts. The tool is built with a focus on adaptability, scalability, and user-friendliness.

2. System Overview

Mango-Torch is an interactive AI-driven image editing tool that utilizes the Stable Diffusion model to transform user-uploaded images based on textual prompts. The system is designed to operate efficiently on both GPU and CPU environments, ensuring accessibility for a wide range of users. The tool provides a clean, intuitive interface powered by Gradio, enabling users to upload images, input prompts, and adjust parameters such as strength and guidance scale for precise control over edits.

Key features include:

• Real-time image editing using AI.
• Adjustable parameters for customization.
• Support for high-quality image outputs.
• Device adaptability for optimal performance.

The system is targeted at creative professionals, hobbyists, and anyone looking to enhance or transform images with minimal effort.

3. Functional Requirements

• As a User, I should be able to upload an image for editing.
• As a User, I should be able to input a textual prompt describing the desired edit.
• As a User, I should be able to adjust the strength of the edit using a slider.
• As a User, I should be able to adjust the guidance scale using a slider.
• As a User, I should be able to view the edited image output.
• As a User, I should be able to download the edited image.
• As a User, I should be able to compare the original and edited images side-by-side.
• As a User, I should be able to process multiple images in a batch.
• As a User, I should be able to toggle advanced features like inpainting or masking specific regions.

4. User Personas

4.1 Creative Professional

• Description: Graphic designers, photographers, and digital artists looking for advanced tools to enhance their work.
• Needs: High-quality outputs, customizable parameters, and batch processing capabilities.

4.2 Hobbyist

• Description: Casual users exploring AI-powered tools for personal projects or social media content.
• Needs: Simple interface, quick results, and easy-to-use features.

4.3 Developer/Tech Enthusiast

• Description: Developers and AI enthusiasts experimenting with image editing technologies.
• Needs: Access to advanced features, model variants, and detailed documentation.

5. Visuals Colors and Theme

Color Palette

The Mango-Torch project adopts a vibrant and modern color scheme to reflect creativity and innovation:

• Background: #1E1E2F (Deep Midnight Blue)
• Surface: #2A2A3C (Charcoal Gray)
• Text: #F5F5F7 (Soft White)
• Accent: #FF9800 (Vivid Mango Orange)
• Muted: #757575 (Neutral Gray)

This palette ensures high contrast for readability while maintaining a visually appealing aesthetic.

6. Signature Design Concept

Interactive "Torchlight" Landing Page

The Mango-Torch homepage will feature a dynamic, interactive design inspired by the concept of a "torchlight." When users hover their cursor over the page, it will act as a virtual torchlight, illuminating sections of the homepage in a circular spotlight effect.

• Background: A dark, textured canvas resembling a starry night sky.
• Animation: Subtle glowing particles that react to the torchlight's movement, creating a sense of depth and interactivity.
• Transitions: Smooth fade-ins and slide-ins for content as it enters the illuminated area.
• Micro-interactions: Buttons and links will "sparkle" briefly when hovered over, reinforcing the torchlight theme.
• Call-to-action: A central, glowing button labeled "Start Editing" that pulses gently to draw attention.

This design will create an unforgettable first impression, immersing users in the creative and innovative spirit of Mango-Torch.

7. Non-Functional Requirements

• The system must support GPU acceleration for optimal performance.
• The system should load the Stable Diffusion model within 5 seconds on a GPU-enabled device.
• The UI must be responsive and accessible on both desktop and mobile devices.
• The system should handle up to 50 concurrent users without performance degradation.
• The system must comply with UAE data privacy regulations.

8. Tech Stack

Frontend

• React for Web
• React Native for Mobile App

Backend

• Python
• FastAPI

Database

• MySQL (with Alembic for migrations)

AI Models

• Stable Diffusion (via Diffusers library)
• GPT 5.4 for user-friendly responses

AI Tools

• Litellm for LLM Routing
• Langchain

Orchestration

• Docker
• Kubernetes

9. Assumptions and Constraints

• The system assumes users have access to devices capable of running modern web browsers.
• The system assumes GPU availability for optimal performance but will fall back to CPU if necessary.
• The system is constrained by the computational limits of the hosting environment.

10. Glossary

• Stable Diffusion: A deep learning model for generating and editing images based on textual prompts.
• Gradio: A Python library for building interactive user interfaces.
• CUDA: A parallel computing platform and application programming interface model created by Nvidia.
• Torchlight Effect: A visual design concept where the cursor acts as a spotlight, illuminating parts of the page.