by admin on April 24, 2013
Sterogum the awesome music site debuted the MNDR/Faster Horses music video I directed. Spin also had a great write up!
by admin on March 4, 2013
DayWalker attempts to literally record your whole day and condense it into a music video at night and upload it to YouTube. Imagine smashing your ennui with your mundane existence enhanced with soundtrack – and instant Kafka-esque Die Verwandlung of your journey through this world. DayWalker explores your environment both online and off and mashes it up into a short film that you can save for posterity or share with friends. You can even analyze it for connections and patterns you didn’t even know existed, like Albert Brooks in “Defending Your Life“.
So things are progressing rapidly thanks to the Raspberry Pi! Let’s take a look at the hardware spec first:
So DayWalker hardware consists of a Raspberry Pi as it’s brain – outfitted with a shiteload of things to interact with the environment around your. GPS – to grab your position on the planet. Bluetooth and WiFi to scan all devices and networks that are near you. A microphone to catch and analyze all the audio around you and break it down into words and relate that to your day as well. Let’s wrap the package up nicely with a touch screen that let’s you know the status of the device, and on-off switch and a rechargeable lithium-ion battery all wrapped up in a custom 3D printed case with two arms straps.
Now let’s drill down into the meaty software spec:
WTF? Okay the short of it is this unwieldy Plesiosaurs of a diagram. The right side of the diagram represents the real-time data collection from the sensory devices attached to the Raspberry Pi which is stored in the DayWalker‘s Database. This includes WiFi network names, Bluetooth phone names – possibly phone calls from your own phone, music playing near the device, conversations that you have and of course the very section on earth you are standing on.
At the end of a set time period (midnight!) , or at any time the user of DayWalker can trigger it to generate a video of what it has collected. Then the Middle of the diagram kicks in collecting additional data from the last period from various social media sources of the user : I.E: Instagram, Twitter, Google calendar, Facebook, Flickr, and other social networks to generate the video.
All this goes to the render engine that begins to pull photos for all the processed audio it has heard during the day, and match photos and positions to WiFi networks, map locations, calendar events, etc. It then beat matches a nice piece of music, or several pieces depending on the length of the video. After compiling it – it uploads it to a youtube account of your choice!
Additional software processes run a web interface for real -time monitoring and input and output to the touch screen for setup and real time updates.
Now that the idea has been laid out clearly – Let’s shake the caboose and see what comes loose.
by admin on February 14, 2013
Everybody Watching Me is an experimental automated low-fi mashup engine in the pursuit of the soi-disant hacker-art.
Everybody Watching Me takes live feeds from open web cameras around the earth in real time and mixes it with cool audio tracks to effectively create a running 24/7 days a week music video out of the mundane minutiae of people’s lives to satisfy the techno-voyeur in this global self-surveillance society.
I tweeted out the idea, and then i thought, why not flip the phodeo engine that powers TweetCinema to do this. So I set to the task of coding this up. That’s the high concept, let’s hold the bolts and look at the nuts of this operation:
Here is a flow chart of the original concept before vi ever opened a text file on this project:
The engine grabs 30 frames from each camera around the world, filters them by several given criteria (The most crucial one is timezone which is coded in but not turned on yet) and then assembles them into a music video with random music tracks & uploads it to the Everybody Watching Me Youtube Channel.
The engine also pushes out a realtime live stream from the website that you can view – that can run pretty good if you have a fast connection – and you can also watch it on video delay which is a loopback archive in mp4 format that is easier to digest across platforms.
I cobbled together an extra function to do some light motion tracking. Right now I kept it to the first two frames, but it could easily be let loose to track all frames. It’s a simple threshold compare to find out where there is motion. Here is a example below. The top box is a TV monitor, the bottom box is a persons arm that is moving, cleaning a table:
Once if finds motion, at the moment, it selects only one, and the biggest motion it can find and attaches a SMS style text speak to who or what is ever moving at their loci, effectively hypertagging them in real time. The text is animated just for even more music video immersion. Below is an example of a lady reading a paper the engine tagged with WTF!
I think clearly the time zone bit needs to be activated to avoid looking at cameras that are pitch dark because it’s nighttime. I also dont know yet if 30 frames is the sweet spot. It currently takes 2 or 3 hours to run a circuit of cameras and right now the database is polling only 160 cameras out of maybe 3-4 thousand available.
Faster server? Less Frames? Code tweaks? When pulling data from foreign websites you are sometimes at the mercy of the uplink and the timeout. A faster circuit run would really be awesome – any ideas on that are welcome.
I already rewrote the motion tracker to handle multiple objects with a tighter center target. I think it would be cool if it could put animal heads on people who are moving or little cute icons. I do plan to cut on the full tracker across all 30 frames and let the words follow them where they go, that seems like that would be awesome.
I think maybe a cute little flag of what country the camera is at might be a nice ideas as well. I also think a frame rate pull based on the beat of the music would also lead to cuts that land on the beat of the music.
Another idea is that instead of a linear poll, perhaps it should poll cameras randomly. Another idea, is of course threading the engine for multiple pulls which would speed it up considerably.
Yet another idea would be to analyze the words in songs and pull camera footage that the engine thinks matches the lyrics, or conversely switch the music based on the images!
Finally I think I’ll add in a end slug with the logo identity for the project – just for that professional touch! everybody watching me – directed by The World!
Let me know what you think, and don’t forget, someone is always, always, always, watching….
by admin on February 5, 2013
What it does simply is take the top trends off twitter and match them to photos using a simple multi-stage algorithmic process to find the right photo from primarily public domain images. It’ll eat any kinda of photo as well and it stores it in a database for later future projects (big data makes apps smarter!)
Now instead of trying to figure out what that cryptic hashtag that everyone is talking about – you can quickly see who, what, where and occasionally how and why people are talking about that subject on twitter! I find it useful myself from where an geopolitical events are taking place – to sports players names who are being traded, to celebrities pulled over for DWI and actually a lot more – all in real time. Big Data, Game Theory, Predictive AI in baby steps.
So the extra special touch of a hybrid programmer/director – every day at night the server also automatically generates a music video out of the top trends of the day!
It assembles it all into a dupstep track that even says the name of the trends and then uploads it to YouTube with a description and hashtags. Here is an example of a completed one:
The idea is to generate one daily, monthly and a year in special all automatically from the server. Set it and forget it. The whole channel is here @: [ tweetcinematrends ]
I’m gonna extend these auto-generate videos a bit more with an extra special live component. I just have to cast for the perfect humans. More on that shortly – but will come to the table in the next 30.
I also have a bit of SEO magic going on the web interface. Each trend that enters the grasp of tweetcinema it generates a static page and directory with the image associated with it. A an example is a trend #BongoTakeover that ranked today. The server generates this: http://www.tweetcinema.com/trends/Bongo_Takeover/ which is all SEO friendly in everyway. The only thing holding it back at the moment is that they aren’t technically linked anywhere to enter the zeitgeist. That’ll change shortly.
I found out while just running tests, things were getting weirdly meta! TweetCinema tagged photos were winding up high up in search engine image results since they were tagged properly. Even stills from the YouTube videos were ending up in searches. Not to mention YouTube video SEO is one of the least exploited high priority search results on the whole internet. Yet another dimension to the experiment.
Right now it’s international focused, which means certain times of the day it can be dominated by various regions and languages. I find it interesting that the primary movers on top international trends tends to be USA, Latin America and Turkey. Every once in a while South Korea, France, England and Russia sneaks in.
The next steps for TweetCinema is to enable on the web interface for weekly, monthly and yearly buttons so one can review longer tail trending information – and then of course a companion mobile app for Android and iOS so you can take it on the go. Maybe – and just maybe – a XBMC version too! (What about a Hewlett Packard Printer app? is that too far?)
I’m finishing up post on Monsters so hopefully I can coded both those apps in the next 30 days. If you have any questions just ask!
by admin on January 23, 2013
So I’ve got these ideas of bridging AI, Game theory and Data Visualization into a new level of awareness in-betwixt directing for a living. It sounds grand on the macro scale but let’s start small. With the advent of the internet all the data we longed to shove into a computer for processing is here. This will have profound effects on game theory, AI, visual recognition (there goes privacy!). A lot of researchers see it. The singularity is near – but I’m getting ahead of myself.
Enter Tweetcinema: The idea is simple. Take the top ten trends on twitter and match them to relevant images. The view of tweetcinema will no longer be faces with a cryptic hash tag, but rather an image to quickly absorb in 10 frames what the zeitgeist is talking about. Doesn’t matter if it’s Justin Beiber or an earthquake in the pacific ocean – tweetcinema will show it to you quickly and you can digest it in seconds without having to cut and paste a hash-tag term. Here is a flow chart of the scope of the project:
If you are a decision-maker – especially let’s say on youth entertainment – you could scan this and readily find out whats going on and what kids are talking about instead of waiting for your intern to decipher hash tags and bring it to you. If you tuned into global events you could rapidly see what is developing in a foreign country even if you don’t have the names and places down, the images would effectively give you a leg up on processing this.
The key is developing an algorithm that sorts metadata on the net from various image resources and match the right kind of photo. The upshot is that this also helps in turn categorize more photos with metadata that can be used in other projects across a wide scope.
I actually had this idea two years ago and started coding it – but did some directing – but back to coding! I plan to use the engine for two or three other ideas rattling across my notebook like Daywalker and Atlas Eleven. More soon!
by admin on January 23, 2013
I did a wide ranging interview with Marc Raco who has appeared in both “Hackateer” and in the upcoming tentatively titled “Monsters”. The interview covers some computer hacking, film directing, and a few other things that I have to legally censor. What a world!
Check it out here:
Monkey Radio Interview of John Threat
Or even just listen to it here:
Looks like it became a staff pick as well on 12/10/12 at www.blogtalkradio.com! Kick arse Mr. Marc Raco, Mr. Ben Rose and Ms. Elsie
by admin on April 30, 2012
Here is a little interview I did on the fabulous Tallulah Bankhead’s new blog. Hopefully i didnt make *more* enemies by doing it. She is a firecracker and wicked smart. Follow her for major laughs and some insights into this crazy world.
by admin on January 3, 2012
Makeout released their video today for “You Can’t Be Friend with Everyone”
A low budget high energy romp thru New York City I directed that saw us almost get locked up and get into fights. We shot this on a Red thanks to Shane Sigler. I love Makeout – thanks Ian, Leah and Jesper!
The video also features QR codes – codes you can scan with your mobile for some extra treats! Check it out and let me know what you think.
by admin on January 2, 2012
by admin on February 4, 2010
In certain types of information warfare campaigns a powerful component is the false website. It’s versatile tool that can be used to establish credibility for an operation, or as a way to disseminate memes into the zeitgeist or even as a sophisticated honey-pot to lure a targeted group in. However setting up such websites can leave quite a paper trail for investigators to follow which can lead to detection of the origin of the campaign and it’s perpetrators.
A lot of false websites are really just small organized crime with phishing scams. These websites are setup from other countries with stolen credit cards. The cards can usually be tracked down and help become a fingerprint that identifies the culprit – but most law enforcement agencies don’t have the resources to take the fight that far except for the largest cases.
Enter the Gift Credit Card – offered by the 3 giants of the credit card industry – Visa, Amex and Mastercard. You can walk into a Duane Reade with no ID and purchase a $5000 gift credit card that is registered to no address or name.
Using these cards, one could acquire the resources for websites, hosting and email and other online services ( phone, fax, voice mail, etc )that require credit cards relatively anonymously of course being careful where one originates one’s data calls to the network from.
They are small and can be used to move cash to other countries rapidly, where as being caught travelling internationally with 10k plus in cash could be tricky – 5 or 10 gift credit cards purchased at 5k a piece will not be noticed for the moment by customs – so that one could purchase services from a remote location exterior to the states with US dollars via a credit card instrument that these services require.
These cards represent a cash cow [ estimates are it is approaching a $100 Billion dollar business and growing ] and the banking/credit card industry is a powerful lobby so they aren’t going away any time soon. Much to the chagrin of the one or two Senators raising the alarm – there is no political will currently to modify the system.
While the secret is out about these cards [ Multi-agency report on fiscal instruments including gift cards ] , accountability for their usage is quite far off and for the present time remain a powerful tool for initiating clean information warfare campaigns that include fictitious websites as a component of the strategy and they also make great gifts!