In this short 3 week R&D project, Myself and three other artists were tasked with using this project to find practical ways to use AI tools for fast prototyping. Our end of project deliverables were just a video playthrough (vertical slice) of the environment and documentation of what we learned. We had no engineers on the project.

The four of us came up with the basic concept for this underwater side scrolling game. I was responsible for the environment assets, the level construction, and the lighting. I also did some AI generation work for the character model.

The AI mesh generation served as a fast proof-of-concept, but reaching production quality required targeted manual iteration. The effort spent prompting complex 3D meshes, particularly the main character, emphasized the balance needed between rapid AI prototyping and traditional 3D pipelines

Takeaways:
1. The assets were good enough for this quick prototype, but not ideal for final game assets. AI image generation or mesh generation was more time consuming than expected but I am interested to see how this tech will continue to develop and streamline the process in the future.
2. I found it better to use AI images or mesh generation as reference and not try to force it farther into my workflow.
3. I should have quickly given up on using AI for your asset generation when it couldn't give me exactly what I wanted.

Software Used: Unreal Engine 5, Blender, Substance Painter, Tripo 3D, Meshy