Thorne Brandt

I'm often doing tedious work such as animation. I usually use some kind of nonfiction spoken work in order to keep me focused. Sometimes I run out of material and what I really want is just to browse discussions of a recent topic on online forums such as reddit, but this can be a tedious distracting process. The goal of this project was to turn online forums into spoken word for ease in multi-tasking.

I thought I could take this opportunity to implement javascript's Web Speech API which includes a TTS SpeechSynthesis If I made a chrome extension, it would only be a matter of parsing the text correctly.

You can install the unpacked version of the chrome extension by cloning the repo

Had the privilege of being interviewed by Erisa Apantaku of SouthSideWeekly a few weeks ago and the audio and are located on their site

I was shamefully late in appreciating tests. I saw them as monotonous labor, more complex, cumbersome, and twice as time consuming as the actual code they were supposed to be testing. When I was getting started with front-end unit testing ( phantom, mocha, jasmine ), most of the time it was the tests themselves that were broken. I would spend hours stubbing methods with near-copies, which gave me the frustrating sense that I wasn't actually testing anything. Then when the code changes, you have to basically start all over again. This post is meant for anyone that is feeling similarly discouraged by the apparent lack of value compared to the effort and time-sink required for testing.

I've stumbled upon down the first step for developing face swap feature, and it's quite entertaining by itself. Facial tracking, poisson blending, and model creation. For this example it can be manually triggered, on things which aren't necessarily faces, which means you can now finally safely faceswap with your cat!

You can try out the experiment yourself here , considering you have a webcam, are in a well lit area, and are using Chrome.

I'm trying to create a faceswap feature.

In trying to debug the facetracking problem in which the alignment was off and my drapes were consistently becoming the face, I am now capturing the image data for the mask on every frame. Because of the difference in scale, it creates almost a liquid displacement effect which I thoroughly enjoy.

TODO - create a tool that exagerrates this.