Dinos

AI assisted randomly generated dino-fighting simulator

Comment

Inspiration

Games like pokemon and axie infinity were inspirations for our project. We wanted to incorporate generative AI and made sure we included a dinosaur based theme.

What it does

The game randomly generates Dinos and allows you to battle them. The generation is done using a combination of AI and randomization to form the game. Each Dino is a randomly selected color, head type, body type, eye, tail, and hat. All these categories make a total of 7500 different combinations available. AI Generates the names and attack moves. When you defeat a Dino in battle, it is added to your collection and you get to battle with it.

How we built it

JavaScript with Express for backend, Java for the front end and MongoDB for the database. ClouldFlare Workers AI was used for generation of names and moves. Randomization is used to create the Dino textures. Our project allows for 7,500 unique Dinos textures, as there are several body/head/eye/color/tail combinations that are possible. The game engine was built from scratch, which took a lot of time, but allowed for complete control of the whole project. All body parts were made from scratch, using Adobe Photoshop. All Dinos are saved in MongoDB.

Challenges we ran into

The AI initially did not generate unique outputs each time, so we used a string of random words as a seed to allow for more uniqueness for names and moves. Fine tuning the prompts to get the correct JSON output was definitely the most difficult part of the project. Creating a game engine from scratch and a RESTful backend within the given time constraints was also very challenging.

Accomplishments that we're proud of

Completely automated generation unique Dinos is something we're very proud of. Being able to incorporate CloudFlare's AI models into our system and fine tuning the prompts to get the desired output is not as easy as it would seem. We're also proud of the fact that we were able to create a game engine with features such as particle effects and task schedulers in the given amount of time.

What we learned

AI can act in unexpected ways sometimes, and you may need to have extremely specific rules for it to follow. Also, graphic design can take a lot longer than anticipated.

What's next for Dinos

The next step for the project is to turn it into a multiplayer game, this is why we made the choice to have a backend instead of keeping all the code in one application. We would also expand on the combat system to make it more interesting.

Share this project:

Updates

posted an update —

We have just about completed the backend. We have it consisting of a total of 7,500 possible dinosaur combinations that are randomly generated. The AI is Cloudflares worker. It manually creates the name, and a is of possible moves based on different types. Each one is randomly generated, based on a set of given criteria. To create the dinosaur, I created a generic size for the Dino. Based on the original design, I split the dinosaur into several parts. Head, tail, eyes, Hat, and body. Each of these are made to all fit together no matter what combination. The different body parts are all separate, and add a since of creativity to our design. All body parts were manually creating using adobe cloud. We used a default color for the specific areas. Within the backend, we have the colors overridden to the random color chosen by the loop.

Log in or sign up for Devpost to join the conversation.