We've made custom shirts before, and we always cross our fingers and hope the design comes out like we made it.
What it does
We created a procedure to identify a grid pattern in an image and calculate a transformation to project a chosen design onto the grid.
We created a web app where you can upload your shirt design and see what it looks like on real models wearing the grid pattern.
How we built it
We used several features of OpenCV, including adaptive thresholding, finding contours, polygonal smoothing, our own heuristic for finding and several perspective transforms to remap the image onto a deformed grid.
Challenges we ran into
Running into problems with OpenCV compatibility.
Accomplishments that we're proud of
Our algorithm is fairly robust and can detect all the squares in an image if they are visible.
What we learned
We used some OpenCV functions we had used before, but we had to explore several new ideas to efficiently calculate our transformations.
What's next for Augmented Design
Our computations are fast enough to be applied to real time videos. This grid idea can be applied to several use cases besides the custom t-shirts, and can be used for all different types of Augmented Reality.