Research Focus
Main Research Question
How can Generative AI be effectively utilized to create and enhance different types of assets for Mach8's projects while maintaining quality and efficiency?
This question drove my exploration across three key domains: image generation, 3D experiences, and technical integration. Each project contributed unique insights to answer this question.
Getting Started
Learn about my role at Mach8 and the goals of my internship.
Read Introduction →Review my initial project plan and objectives.
View Plan →See how the project evolved and was scoped.
Explore Scope →Image Generation Projects
Research Question Focus
How can Generative AI be used to generate professional-quality photos while preserving the integrity of the original inputs?
Product Photography: Pipeline + Front-end
An innovative AI pipeline that places real objects in any setting with realistic lighting, revolutionizing product photography workflows.
Socios Project
An innovative project focused on AI-powered image generation for social media content.
Research Question Answer
My research across both the product photography and Socios projects revealed that generating professional-quality photos while preserving input integrity requires a structured, multi-step approach. Through careful object segmentation and controlled image manipulation, I achieved consistent, high-quality results in both scenarios - whether placing products in new environments or swapping football kits. The key to success lies in breaking down complex transformations into discrete steps, each with its own quality controls. This methodology, supported by systematic testing and parameter documentation, enabled me to maintain crucial details like product features and kit designs while seamlessly adapting lighting and environmental context.
3D Generation Projects
Research Question Focus
How can we create interactive 3D experiences using AI?
Virtual Human
An ambitious project to create an interactive figure using AI and Unreal Engine 5.
Research Question Answer
The virtual human project revealed that while AI can enhance specific aspects of 3D experiences like real-time lip-sync and character interaction, success lies in combining AI capabilities with traditional game development techniques. The project's implementation of gRPC servers, Audio2Face and Unreal Engine 5 combined with OpenAI's Real-Time AI showcases this hybrid approach's effectiveness in creating engaging interactive experiences.
Project Feedback
Review detailed feedback from user testing sessions and expert reviews for each project. This section includes implementation status and insights gathered during development.
View Project Feedback →Portfolio Feedback
This page presents the third portfolio review, detailing original feedback and how each point was addressed. It covers aspects like research clarity, stakeholder context, content consistency, navigation, writing style, and project planning, showing the steps taken to improve the portfolio based on the feedback received.
View Portfolio Feedback →Key Findings
Main Research Question
How can Generative AI be effectively utilized to create and enhance different types of assets for Mach8's projects while maintaining quality and efficiency?
Research Question Answer
Through practical implementation across multiple projects, we found that AI's effectiveness varies significantly by asset type and use case. Image generation proved most mature and practical, while 3D experiences required a hybrid approach combining AI with traditional development methods.
Key Insights by Domain
- Commercial Viability: Image generation proved to be the most mature AI application, with immediate practical benefits for product photography and marketing content
- Workflow Integration: Successfully developed efficient pipelines that combine multiple AI technologies (GPT Vision, Stable Diffusion, LoRA) with traditional image processing
- Quality Control: Implemented robust systems for maintaining consistent quality across large batches of generated images through careful prompt engineering and parameter optimization
- User Experience: Created intuitive interfaces that allow both novice and expert users to leverage AI capabilities effectively
- Hybrid Development: Most effective approach combines AI capabilities with traditional 3D development tools, particularly in Unreal Engine 5
- Real-time Processing: Successfully overcame significant technical challenges in audio processing and facial animation through custom gRPC solutions and optimization techniques
- Character Creation: Achieved high-quality results by combining MetaHuman technology with AI-driven animation and interaction systems
- Performance Optimization: Developed sophisticated solutions for ray-tracing and physics calculations while maintaining visual quality and real-time responsiveness