so i have the same story i suspect that many folks do regarding fitness... i knew it was time to get my rear to work on it, but the gym was boring, running was too hard on my knees, cycling seemed a little dangerous and weather dependent, etc etc excuses excuses. for a long time i dreamed a solution would be having a room in my house to put an elliptical machine and be able to watch my own programs, both to pass the time and to sortof kill 2 birds with one stone. this was reinforced by the idea that visual stimulus really helps make workouts go by faster. the problem that it probably took me too long to acknowledge is living in san francisco (and reluctant to leave), i wasn't likely to have this this extra workout room in my house any time soon.
i finally came into the idea that instead of bringing an elliptical into my house, i should be able to take my videos to the gym. in fact i have a gym a block from my house, one of the nice tradeoffs of having little personal space due to living in a dense area. in 2009 when i did this project, this felt like this was a particularly opportune project idea, with lots of technologies finally coming available at what felt was an exciting breakthrough time.
the solution may seem a little more obvious to others than it does to me. some might say iphone, but i don't think the screen size cuts it, either for viewing or control from a bouncy situation. some might say ipad, but i doubt the storage or locked-down environment would work. in either case, i doubt real quality roaming streaming capability is there. also i wanted to be able to sync all my content in a way that made the experience integrate with my sofa watching experience... random access and delete after consuming both at the sofa and at the gym, and that would take decent storage.
i decided to build the ui based on a web browser, which i have a lot of experience coding to. but there were still a lot of things i felt i needed, and fortunate to have at the right time:
a capacitive touch interface with a decent screen size. it needed to work with linux also. i decided on the hp tx2 tablet laptop, which i am very happy with, and some smart folks in the ubuntu community had just finished the work of figuring out how to drive it in xwindows. the tablet mode is very nice, i can lean it up in the magazine holder on whichever machine i'm using, and adjust the viewing angle as needed depending the holder's position. but, the tx2 is not without it's issues. only one of the three bezel buttons seems to work under linux for me... better than none tho. the removable dvd drive catch broke, so i don't really have one anymore. why it was removeable in the first place is a mystery... i called hp and the dvd drive is the only option available for that bay. it doesn't want to power up without a power plug, but after that i can remove the plug and run on battery normally. the touchscreen is really very good, but does have some phantom jumping and sometimes clicking when the screen displays a lot of moving dark contrast... not usually a problem though.
good amount of storage. these days a 640G 2.5 inch drive to go in a laptop can hold a very respectable amount of video for less than $100.
a compositing window manager (i felt the right control design was a transparent one, which would accommodate large buttons for bouncy fat finger manipulations, without encumbering video display real estate). i switched from fvwm (my wm of 11 years) to compiz, and the switch was shockingly painless even with my extensive old school xwindows desktop customizations. it also had the right features i needed to programatically control the window opaqueness (which i did with compiz display rules based on window titles, and i was able to manipulate the window titles with vanilla html title fields).
file synchronization (i wanted the exact same content available on the elliptical as on my sofa, and deletable in either place). i happened on relatively new csync, which is a sortof efficient bidirectional rsync. it fit better than unison, a different synchronizer i love but is more suited to text, often doing full file scans which just did not work on the huge video or even audio files (i've cobbled together a similar portable podcast content platform based on a sansa clip for my work commute, but that's another story).
good content collecting tools (podcatcher, youtube downloader, compatible format computer dvr, various sniffy scraper techniques, bittorrent).
grab-and-drag firefox plugin - for scrolling ui lists from the touchscreen
a number of other technical tools that were either relatively new or i just hadn't learned how to use yet:
xautomation/wmctrl - for launching and positioning windows in the wm... fullscreening, foregrounding, switching desktops, sending keystrokes. xsetwacom for enabling/disabling the touchscreen with the bezel button, to avoid phantom clicks while the laptop was stowed between uses.
floating html layouts - i just learned how to do these from my coworker henry. just what i needed for building scrollable areas and large fat-finger touchscreen buttons and generally decent control layouts.
jquery - making the control code easier to write
xmlhttprequest - for asynchronous controlling of the video player
video player with an approachable control api - i've used mplayer for years, and delighted to find it had fine full featured named-socket controls, and all the playback features i could ask for (disableable onscreen display with elapsed/total time, volume level, skipping an arbitrary period forward or backward, a/v delay for bluetooth syncing (i eventually gave up on bluetooth audio and decided wired earbuds worked better... some footnotes below)).
this project, though admittedly somewhat humble (there's plenty of mobile video gadgetry out there), was really exciting from beginning to end. it's not the first time i've built something that i knew was going to make my life better. but it was uniquely remarkable when i think about how a dozen different pieces all came together at the right time in the right way. many of the pieces hadn't existed just a couple months or years before, and just as many pieces may have existed but were completely new to me. on top of all this, any engineer knows that the more moving parts, even if they're mature and familiar, the more complex and likely to fail an idea will be. but in this case, with a just a modest amount of engineering and research, i wound up with exactly what i had imagined i wanted.
now, a couple years later, and can say with certainness this project was a total success. i still get excited about not wasting time watching video at home (well, not AS much), and about having tons of interesting and current content available to consume while burning calories at the same time.
here's a youtube video of the rig in action:
http://www.youtube.com/watch?v=K2637RSIvLM
footnotes:
i put a good amount of effort into wireless bluetooth headphones, but eventually gave up on them. i went through 3 different pairs that would break or just not cut it in different ways. the plantronics voyager 855 was too quiet. the motorola rokr lost sync too easily and often. the rocketfish knockoff of the motorola rokr performed the best and was cheapest, but the power switch eventually broke. it was also tricky to keep any of the headphones paired. i found i had to send a silent track at all times, else the headphones would go to sleep after a very short time. it was also a pain to manage the sync by delaying the audio on the player... but i would have put up with this if not for all the other issues.
i use firefox for the interface currently. i tried to use chrome, but it had funky behavior when i was remote and not connected to the network and just connecting to the local web cgi. also it did not blank the pointer the way i wanted, especially since the tx2 had a phantom mouse movement which would be distracting in front of the video if not blanked well.
cost was relatively modest. more than an ipad, but what i have is much more capable for my needs also. $800 for the hp tx2, $100 for 640g hd, and maybe $30 for the earbuds.
