Inspiration: We were inspired by the silly filters we see on SnapChat and how my teammates love to prank eachother. While looking for a good filter to prank our friends, we realized snap didn't have such filter. Thats why we created one.

What it does: In action, this filter is very realistic! Point to a tv screen and the screen is replaced with an image of a cracked screen.

How we built it: We coded in javascript and imported some textures and materials from previous snap tenmplates. We also imported pretrained object detection model.

Challenges we ran into were when the object didnt detect properly. We realized that it was because we had to turn off a function.

Accomplishments that we're proud of is finally figuring out why the object didnt detect properly.

What we learned is to look at the simple things first before trying to solve the problem in difficult methods such as rewriting the code.

What's next for Cracked Screen: We plan to make it more realistic, allow users to choose different types of images, and make our own dataset that we trained as that would be more efficient because we would be able to train it to do different things.

Built With

Share this project: