Originally I wanted to deal with Australian Indie bands but the website I was trying to use made it so no easy way was possible. So Instead I went Indie Video Games.
The result is a visualization of 777 indie video games’ names in relation to the game’s rating and word occurrence. In classic word cloud style, words that occur most often are largest. Words are also colored according to the average rating for games who it was a title for (Red being negative, Green being positive).
Most annoying bug: Python & Javascript split differently on spaces and I assumed they behaved similarly. It took me a good 2-3 hours to figure it out.
I use Jason Davies D3 Word Cloud Library as the base for my visualization.
For this project, I created an app which takes a person’s face, and generates a sine wave based on a few characteristics. Here’s a picture of me testing it out. There’s no way to look graceful with this thing:
Here’s a video of our esteemed classmates trying it out:
My biggest takeaway from this project was that sometimes it’s easier to work things out on your own, rather than fumbling around with an addon that promises it will work. I had been working with ofxTonic, but memory issues in the addon download made my app crash soon after it was launched. I tried other libraries, but none worked well enough for my purposes.
Thankfully, we have some electrical engineering classmates, who helped me tremendously by teaching me about the math of sine waves. With their help, we populated the sound buffer manually, rather than trying to get ofxTonic up to speed. It works really well!
In which Paul Peng makes a map of approximately 500,000 hotels using their price and hash number to determine their location. I also used their geographic locations to determine the area of the ground they take up and their height. I also made all of them white pyramid-like structures that kind of look like the Pyongyang Hotel in North Korea.
My original plan was going to include actual shading, a real ground, a color scheme that is both considerably less stark and considerably more alienating, rectangular prisms instead of pyramids, a slight fog to create a sense of atmospheric perspective, soft buzzy atonal (figuratively) pads that fade in and out very periodically but in a constant beat and whose tone and harshness depends on the nearby hotel substitute structures, and first person perspective and movement that allows one to travel through and become immersed in this strange world of alienated quantification. I would include sketches and mock-ups of this vision but unfortunately and in a move very uncharacteristic of me I have not made any of these sketches or mock-ups. I also would have implemented all of this if I didn’t begin implementation on the midnight before the assignment was due, which is also why I don’t have the ofxAddons assignment or the Web Scraping and Display Assignment done either.
I used Kimono Labs and Excel for this project. I decided to grab the top 1000 movies of all time as claimed by the NY Times, and plot the number of movies on the list per year.
The curve was relatively as I expected. I expected relatively low numbers very early on, a middle point in history where we claim the “best films” come from, and then a drop as time declined slightly. I was surprised, though it makes sense, that there were so many spikes, some years where tons of great, noteworthy movies were released, and years where there are none of any real note. Kimono Labs was a great tool to work with, and made the whole process painless and quick. After some data cleanup and reorganization, I managed to plot the number of movies in the top 1000 list over time.
Given the database of all hotels in the world, I decided to run a TFIDF (Term Frequency Inverse Document Frequency) algorithm upon the hotel names, dividing the hotels up by countries, treating the countries like documents. In other words, given any country in the world (ie Germany), I can return a list of the most German ways to name a hotel, by comparing all the ways Germans name hotels to the entire corpus of countries, on a term by term basis.
The two addons I used were ofxCsv and ofxUI. I compiled the two of them together, but didn’t really end up utilizing ofxUI for anything practically speaking, due to the minimalist aesthetic I decided to go for, creating my own scrollbar rather than one from ofxUI.
The final interaction allows you to scroll through a text list of all of the countries in alphabetical order, alongside a few terms that are how those countries distinctly name hotels. Unfortunately, the algorithm I’ve written for the TFIDF is extremely granular, much moreso than what would realistically be required for an accurate data visualization. For this reason, my program has been running for a couple of hours and I’ll sleep on it and hopefully the numbers will have crunched fully. After the numbers of crunched, the program should export the data as a csv so that next time I open it up it should just load in the data and you can interact with it right away.
[UPDATE] After realizing my algorithm wasn’t optimized after running for 48 hours, I updated the code (from a horrendous O(n^3) to a manageable O(n^2)) and within a couple of hours, the algorithm finished. In a lot of ways, the results weren’t totally surprising. What I noticed were a few (unsurprising) trends: smaller countries tended to have their largest cities as disproportionately popular terms; countries tended to be divided among language lines; the term “Hotel” occurred much more often than I had expected considering the TFIDF process. What I suspect is that many smaller countries tended to use the term “Hotel,” while larger countries did not. For this reason, it could still be considered a disproportionately large portion of the entire corpus of appearances for those small countries.
Next steps? I’m very pleased with the general outcome of this project. My TFIDF is functional and fairly optimized. I plan to thread it as to increase speed up further. Top results also included the term “&” and a few other things I had to clean up in post, so I now know how to preprocess some of the data to ensure my results are better next time I use it.
I plan to use this algorithm for my data-selfie assignment, in which I will be using the entire corpus of my Facebook and text messaging, and very the inputs slightly to get a wide range of results (comparing how I talk to each user, comparing how I talked to a given user over time, using n-grams rather than individual terms, comparing how I talk to men vs women, comparing how women talk to me vs how men talk to me, etc.).
This project was definitely a learning experience for me…
I don’t usually work with this much pure data, so it was quite the curve of figuring out how to use it in an efficient way. I’m still currently trying to figure out the most effective way to use this amount of data to map the coordinates.
My journey started by taking the tilde separated file of the Hotel Data and going through and cleaning it to make it usable. My first attempt caused massive errors when converting it to a CSV file using the ofxCSV addon for openFrameworks. After much debating, I found out a major problem was that there were commas in the tilde file before replacing the tilde with commas! These extra comma screwed my charting of the data from the very beginning… it was such a sad moment when I found out this simple error. So before converting the tildes to commas, I first converted the prior commas of the files to a period. This was the conversion didn’t get confused. This fixed everything, and laid out my data to create a clean CSV file.
The part that gave me even more trouble was plotting these points. I decided to use TileMill… I regret this decision now, due to the fact that no computer I’ve found can plot the mass amount of data through tileMill in an effective way. Every time I try to zoom to another layer or export, the entire program crashes my entire computer.
To try to allow my computer to breath a little (still to little avail) I went through the CSV file and deleted all non essential data. Thus, leaving only the Latitude and Longitude data. This still didn’t provide adequate breathing room for TileMill so I trimmed it a bit more to only plot above the equator. This was the last image I could get out before my computer crashes.
Next time, I’ll stick to openFrameworks, where it seems much more likely that it can handle this much data in a more interactive way. As a whole this experience taught me a lot about large quantities of data, and how I need to learn how to use them more efficiently.
Description this assignment has been much more frustrating than what it could have been. Installing visual studio on windows and running the project generator were the biggest challenge. Another challenge was finding openFrameworks addons compatible with visual studio and free of bugs. After several trials and failures, I decided to switch to Mac and run xcode. At the end of a long day of testing ofxAddons and trying to find creative ways to put two together, I had a dream about a project idea where OpenCv and Box2d would work together to create an interactive software where shapes would rotate around real-time moving objects. And I turned my dream into reality…
In this project, I thought it would prove to be a good idea to use it as a dry test run/proof of concept towards my final project idea. Right now, for my final project I want to create a fully interactive stage. This will be centered around a pepper’s ghost stage, which will help augment reality even further creating a hologram-like experience for the viewer. The two Addons I used for this project where ofxLeapMotion and ofxSyphon.
ofxLeapmotion was a fun addon to play around with. It is also very powerful and full of potential. The one thing I did find out though is that it is very sensitive. Sometimes it thinks it sees a hand when there shouldn’t be one in its view, and sometimes it just creates a ghost for no reason. BUT besides all of that, It proved to very fun and powerful once I learned how to use it more in a performance like test.
The second addon, Syphon, was a must-learn for me. This is due to our projection mapping software of choice in the Media Department relies heavily on Syphon inputs. I figured if I was going to code anything for the theatre, it would eventually need to be able to output to other programs in an efficient way. So I thought this was the best time to go ahead an learn it!
This project consisted of me using the LEAP to control a custom 3D object (I made a 3D model of this year’s School of Drama Centennial 100 Logo) in 3D space while outputting through Syphon to be projection mapped using Millumin (our mapping software of choice in the department). This was then sent out to be projected onto the ground, to be caught into the reflection of the Pepper’s Ghost stage to create a hologram like object floating around my body. In the end, I decided to make a fun tango for the occasion using the 3D object as my dancing partner! I had a lot of fun (maybe a little to much… nah, never to much) with this project and learned a ton while doing it!
Looking through the list awhile ago I found three that jumped out at me. These where also due to me looking ahead to the future and what I would want to possibly use in my final project to create a fully interactive stage environment, centered around augmenting the world around the actor’s characterizations/movements.
ofxLeapMotion
This add-on caught my attention due to it’s fairly accurate and responsive nature. You can get very subtle details in the hand, that you just can’t get with a standard Kinect. This could be interesting if it where embedded into a set piece and allowed for actor to manipulate something in their environment with a simple gesture or swipe.
ofxSyphon
This one jumped out at me simply because we use syphon already so much in the Media department in the School of Drama. If I want to code anything in openFrameworks, I will definitely need to learn how to use this addon to allow us to use it with our mapping softwares of choice, or even other interfaces to allow further manipulation later in the pipeline.
ofxBox2d
This one seems interesting to me due to the formation of a mesh the acts as a boundary box for other meshes. This could prove useful to project things around the actor and make it so that they can still move/have control and the projected environment will manipulate to them.
As I said earlier, all of these are purely idea based, with me looking ahead to the final project….
I’ve only touched the tip of the iceberg when it comes to openFrameworks and looking to other’s for inspirations always gets me excited at the possibilities that can come with it to benefit my design process when creating/designing for the theatrical environment. The three below caught my attention because they all have ideas/concepts that I would possibly want to explore in the future (maybe for my final project?).
Puppet Parade:
This one catches my eye at the idea of virtual puppetry. It has long been an interest of mine to allow the actor to virtually control the puppet and have it augment out through projection mapping into a new environment. Thus allowing them to take a new form and personality altogether. This is a major step into that direction. The design of it also helps this project out. Since it is clean and stylized, they can get away with a simpler 3D model to manipulate live. This was smart planning on their part.
Mortal Engine: By Chunky Move:
I can’t confirm if this was done completely in openFrameworks, but I’ve seen many openFramework sketched that do something similar. This one is just a prime example of how well this concept can be accomplished. This also leads to my goal of having a fully interactive stage environment, where the actor is in complete control so they are not confined to the limits of the program.
Shutter Spot:
This one is a very interesting idea. To have the actor control the lights attributes with their actions would be a very interesting/cool concept to look into in the future. I also love what they have done with the sound responsiveness in this installation as well.
