Project Updates

Kin Lane

I Am Not Well

I am not well. I have something systemic wrong with me. I am not responsible for its origins, but I am responsible for perpetuating it. It is a condition that is difficult to see despite being ubiquitous throughout our lives.
Kin Lane

Seeing the Machine

I wish I could paint the pictures I have in my head from my work as the API Evangelist. The closest I can come is projecting light and transforming images as part of my Algorotoscope Work. Some day my painting skills may get me closer to what I see emerging around us, but for right now my Algorotoscope Work is the closest I got. I am just turning the knobs and dials on the Algorotoscope lenses I have created, until the reception gets better...
Kin Lane

Why I Do Algorotoscope

I started Algorotoscope as an escape from the 2016 election. I wanted to learn more about machine learning and TensorFlow while simultaneously wanting to escape the world. Once I established a working model I wanted to get out in the world more to take photos and videos. I lost the Amazon Machine Image (AMI) for my Algorotoscope process as part of switching accounts, so I haven’t produced anything new in a while. I am just working from images I have already applied my models. Now I have my...
Kin Lane

Rebuilding the Texture Transfer ML Process

I had an AWS machine image that had my texture transfer process all setup. I had my AWS account compromised and someone spun up servers across almost every AWS region—my bill was pushing $25K. Luckily it was due to anything I had done and AWS didn’t charge me for any of it. However, in the shuffle from that account to a new API Evangelist specific account, my AWS machine image got lost—forcing me to have to rebuild from scratch. It is always a daunting thing to dive back into the world of ...
Kin Lane

Rebuilding the Texture Transfer ML Process

Historically the ML models I have trained for Algorotoscope have been fairly dark in nature. I had the training wheels of the models that came with the ML process that I adopted, but then I quickly began training on Salvador Dali and playing around with a distortion in time, but then found myself drowning in Nazi and Russian propaganda. I would take these models and apply them to normal everyday photos ...
Kin Lane

Landing on a Stable Instance of Tensorflow on Amazon to Train Models

I have bounced around quite a bit from the original Tensorflow model I was using when I started this work. Algorithmia had done a lot of the heavy lifting for me, but that original Amazon Web Services machine image eventually fell into disrepair and I couldn’t operate it anymore. For the last couple of years I bounced around trying different text transfer services, and kicking the tires on a variety of Tensorflow models that were cheaper to operate, but none of them produced the same resul...
Kin Lane

Adding Some Collections of My Algorotoscope Work

After organizing a couple of the latest batches of images I produced as part of my algorotoscope work I wanted to step back and see what I had made. I wanted to get better at organizing some of the more interesting images I had produced using specific Tensorflow machine learning models. Much of my work is pretty random without much connection between the image I trained my models on and the images I applied my models to–they are just making for some interesting colors and textures. However...
Kin Lane

Showing What Algorithmic Influence On Markets Leaves Out

I’ve been playing with different ways of visualizing the impact that algorithms are making on our lives. How they are being used to distort the immigration debate, and how the current administration is being influenced and p0wned by Russian propaganda. I find shedding light o...
Kin Lane

Highlighting Algorithmic Transparency Using My Algorotoscope Work

I started doing my algorotoscope work to better understand machine learning. I needed a hands-on project that would allow me to play with two overlapping slices of the machine learning pie–working with images and video. I wanted to understand what people meant when they said texture transfer or object recognition, and quantify the differences between machine learning providers, pulling the curtain back a...
Kin Lane

Why Would People Want Fine Art Trained Machine Learning Models

I'm spending time on my algorithmic rotoscope work, and thinking about how the machine learning style textures I've been marking can be put to use. I'm trying to see things from different vantage points and develop a better understanding of how texture styles can be put to use in the regular world. I am enjoying using image style filters in my writing. It gives me kind of a gamified layer to my photography and drone hobby that allows ...
Kin Lane

Machine Learning Style Transfer For Museums, Libraries, and Collections

I putting some thought into some next steps for my algorithmic rotoscope work, which is about the training and applying of image style transfer machine learning models. I'm talking with Jason Toy (@jtoy) over at  Read more →
Kin Lane

The Russian Propaganda Distortion Field Around The White House

I am having a difficult time reconciling what is going on with the White House right now. The distortion field around the administration right now feels like some bad acid trip from the 1980s, before I learned how to find the good LSD. After losing their shit over her emails and Benghazi, they are willing to overlook Russi...
Kin Lane

Algorithmic Reflections On The Immigration Debate

We are increasingly looking through an algorithmic lens when it comes to politics in our everyday lives. I spend a significant portion of my days trying to understand how algorithms are being used to shift how we view and discuss politics. One of the ongoing themes in my research is focused on machine learning, which is an aspect...
Kin Lane

When The Companies Who Have All Your Digital Bits Promise Not To Recreate You

I'm thinking about my digital bits a lot lately. Thinking about the digital bits that I create, the bits I generate automatically, the bits I own, the bits I do not own, and how I can make a living with just a handful of my bits. I have an inbox full of people who want me to put my bits on their websites, and people who want to put their bits on my platform so that they are associated...
Kin Lane

Algorithmia's Multi-Platform Data Storage Solution For Machine Learning Workflows

I've been working with Algorithmia to manage a large number of images as part of my algorithmic rotoscope side project, and they have a really nice omni-platform approach to allowing me to manage my images and other files I am using in my machine learning workflows. Images, files, and the input and output of heavy object is an essential part of almost any machine learning task, and Algorithmia makes easy to...
Kin Lane

Finding The Right Dystopian Filter To Represent The World Unfolding Around Us

I got sucked into a project over the holidays, partly because it was an interesting technical challenge, but mostly because it provided me with a creative distraction after the election. I started playing with image filters from Algorithmia, using their Deep Filter service, which some may recogni...
Kin Lane

Exploring The Economics of Wholesale and Retail Algorithmic APIs

I got sucked into a month long project applying machine learning filters to video over the holidays. The project began with me doing the research on the economics behind Algorithmia's machine learning...
Kin Lane

Learning About Machine Learning APIs With My Algorithmic Rotoscope Work

I was playing around with Algorithmia for a story about their business model back in December, when I got sucked into playing with their DeepFilter ser...
Kin Lane

Make It An API Driven Publishing Solution

Once I had established a sort of proof of concept for my algorithmic rotoscope process, and was able to manually execute each step of the process from separating a video, and applying filters, to reassembling the video, I quickly refactored my prototype code to be API-first. I did this even before I built any sort of interface for executing and managing the process, as this would allow me to not just execute the process, it would also allow me to manage, extend, and scale as many algorithm...
Kin Lane

Unique Algorithmic Filters Is Where It Is At

I am having a blast with the image texture filters that Algorithmia includes as part of their DeepLearning service. I've been playing with how each filter will behave when applied to different images. Some work better for the desert images, while others work best for water, and I'll apply to my waterfalls, lakes, and rivers. The process has been a welcome distraction, but where I really get lost thinking about the pos...
Kin Lane

Opportunity In Breaking Up Videos Into Separate Images

I have been working my way through about 300 GB of drone and GoPro videos from this summer. One of the lingering thoughts I was having throughout this process centered around the concept of breaking videos up into individual slides. Once I stumbled across Algorithmia's Deep Learning, I found myself thinking about how I could break up videos into separate images, and apply algorithmic texture filters to each individual...
Kin Lane

Algorithmic Rotoscope

I had come across Texture Networks: Feed-forward Synthesis of Textures and Stylized Image from Cornell University a while back in my regular monitoring of the API space, so I was pleased to see Algorithmia building on this work with their Deep Filter project. The artistic styles and textures you can apply to images is fun to play with, and even more fun when you apply at...