Raspberry Pi - A credit card sized, linux based mini-computer for $30

http://www.rasplex.com/index.html

when we move I'm thinking about doing this for our home wifi:

http://www.tripwire.com/state-of-security/security-data-protection/sweet-security-part-2-creating-a-defensible-raspberry-pi/

After a couple of months of not using mine it now refuses to sync with the TV over HDMI. Tried two different distros, neither of which has changed in the meantime but all I get is "no signal" on the TV.

Need to check the HDMI cable.

I tried Rasplex but it's a bitch to set up as a standalone HTPC. It can't do ANY streaming video services w/o connecting to yr main computer, and transferring files to it is insaaaaaaanely slow (and the file system makes it m/l impossible to just xfer files directly onto the SD card)

basically I wanted this to be a standalone Plex server and it's not working and idk what else I should do with it (I don't really play games all that much?)

Put the files on a USB stick?

just got my raspberry 3 in this morning. anyone else still fuck with a pi? my goal is to use an ultrasound sensor (or two or three) as input, determine if someone is there, then output different videos and sounds according to if someone is present and where they are positioned.

i'm sure this will be super easy and i won't run into any problems

it's kind of fun and silly connecting it to my living room tv as a display. tbh i wish i could just connect it directly to my macbook, like arduino. apparently there's a way to do it with RPi but it doesn't seem nearly as simple/plug-in-play as arduino in that respect.

I sort of do! I just put Kodi on it and installed Exodus

at this point i have no fucking clue what you're talking about! i love that feeling.

what does Kodi do? what is Exodus? i could google it but i'm too busy deciding what my secret pi password should be. gotta make sure no one can hack into my pi and discover my secret files.

kodi is xbmc, x-box media centre.

Exodus is Dark Netflix

that's cool! so you use it primarily as a media center, then? i got sidetracked early on when i found out that minecraft comes pre-installed in raspbian.

i don't even like minecraft that much, wtf

I have a couple of Arduinos gathering dust while I wait for my electronics skills to progress beyond 'man discovers fire and recoils in fear'. Hopefully, it will still be exciting to toy around with them when they're twenty years out of date.

yeah, the electronics part is the thing i'm worried about. if all goes well i'd like to output my video directly to a projector, and i'm terrified of overloading everything and accidentally burning the building down

there's insanely little chance of that

It's a watered down version of Minecraft but yeah it's kind of amazing to discover when you're not expecting it to be there at all.

Are you living in Brooklyn? There's a store called Tinkersphere downtown in Manhattan that carries all the components you'd need to rig up all that stuff.

nah, i moved to Chicago. but i bet we have something similar here as well.

i'm lucky to have a new friend who is super into all of this stuff and wants to help me with my project.

btw, re exodus: https://torrentfreak.com/popular-kodi-addon-exodus-turned-users-into-a-ddos-botnet-170203/

Don't know how I missed that you moved to Chicago, KM. It's always rad to learn that there are rad people in the metro area outside of my hermitsphere.

I'm trying to teach the kid I'm tutoring what little electronics stuff I know in the hopes that he'll quickly surpass me in that magical way kids do and can, in turn, teach me more than the rudiments.

oh, i didn't make much of a fuss about it! i'm in hyde park, which i believe is located in the heart of the hermitsphere.

It's easy to teach kids basic electronic principles using Minecraft, because they're generally receptive to it.

one odd thing off the bat - i opened up one of the pre-installed python games (Simulate) and almost immediately a thermometer icon appeared in the upper-right to let me know that the CPU temperature was too high. i used "vcgencmd measure_temp" to get the exact reading, which was 83.8 C - just below the max 85 C temp that's supposed to be safe. as soon as i closed the Simulate python game and checked the temp again, it was down to 75 C. after another few minutes of inactivity, it was down to 65.

i repeated all of this and the same thing happened - running the python game instantly raises the CPU temperature to near its maximum safe reading, and closing it immediately drops the temp back down to normal.

i guess i should take the case off? i really didn't expect to need to use the heatsink just for running a simple python game.

heyo new chicago-ers i am nearly among u likewise after moving to evanston for my partner's phd.

i rigged up my pi recently to start up a vpn, via an open source package called streisand, on an amazon web svc machine. i steered into the data collection paranoia skid earlier this yr and started trying to secure all my web traffic and encrypt things in advance of anticipated loosening of isp privacy protections of my web traffic. (i do not wish to become a data point for sale to marketing firms.) i am still in early days of developing working understanding for the tech and tbh it is a pretty pedestrian use for the lil pi but it's my only linux capable machine presently. i am considering conversion to a kodi box for more utility. Or maybe i will make a weather station. the primary limitation still seems to be motivation & time but who knows

> i'd like to output my video directly to a projector,

wouldn't you just use the hdmi port?

my experience with pygame is very similar - very intensive. it's trying to squeeze high fps values out of an interpreted language.

once python's up and going it should be fine unless python for arm sucks

just got my raspberry 3 in this morning. anyone else still fuck with a pi?

Once I went through all the "here's how to make fire" intro stuff I stuck mine on the shelf next to my arudino until I discovered that folks were using them for SDR and then got back into it again
https://learn.adafruit.com/freq-show-raspberry-pi-rtl-sdr-scanner/overview

i'm thinking i got a faulty one (overheating). even at idle it's around 52 C, which is apparently hotter than it should ever be, even when all 4 cores are maxed out with intense use. and when i run simple programs like VLC the CPU temp shoots up to 80 within 30 seconds.

I don't have a Pi 3, but apparently others are having overheat issues too.

I've had a Pi 3 for about a year now and I've never had a problem with overheating. Mind you I mainly use mine as a basic LAMP dev box.

i put the heatsinks on late last night and it dropped the idle temp down to 45 or so, increasing to 70 with intense CPU activity. still higher than i'd prefer, but at least it's workable.

i am learning so, so much about linux, RPi and python. i always hear that the best way to learn is to have a project in mind as you do it, and it's absolutely true in this case. i have run into about one trillion problems, large and small, and have managed to figure them all out.

i think i also solved a common problem that doesn't have a solution posted online anywhere? when playing multiple videos using omxplayer, there is a roughly 1-second pause between each video, often accompanied by the sound cutting out for the first bit of the clip. i came up with a band-aid workaround but fuck, i think it works! i have a loop going where it opens up a new instance of omxplayer just before the previous clip ends, and positions it one render layer higher each time around so that it overlays on top of the previous instance, which closes as soon as it's done playing a moment later. the effect is a seamless transition, and it only uses about 10% of the CPU! i'm super happy because before i started on this i didn't know how to use bash or the terminal and instances and all of that.

i mentioned it a bit upthread, but i'm getting closer to completing this project, and it's changed quite a bit since i started as i learn more about what i can do with python and the rpi. i plan on creating a ton of very short clips that show a face tracking an object - a dozen clips of the face looking straight forward, a dozen to the center-center-left, a dozen to the center-left, etc etc, and then putting them in different buckets. then i'll play back the clips on a small display according to sensor/GPIO readings, so if you're standing directly in front of the display, you'll see clips of the face looking straight at you, but if you walk off to the side, the face and eyes will follow. i have other fun surprises in store as well (or i'm trying to, at least), but that's the gist of it.

i pretty much have everything working on the tech side except for the sensor stuff. but i'm hopeful that it won't be hard, and that i'll be able to easily feed the sensor readings to my python script so that it can determine which bucket of video clips should be playing at any moment. but the rest of it, the hard part - learning how to install python and packages and modules and use the command line via a python script, getting seamless video playback - is pretty much over. the rest is just fun stuff!

anyway sorry if i'm a little giddy but i feel like i just completed a basic, but real, irl conversation in a foreign language for the first time and it's exciting!

also i might have to finally register a stackexchange account to share the video playback workaround, because i've seen so many different people ask about it across multiple years and all the suggested solutions i ran across seem inadequate.

Good on ya, man. I barely got the thing to run RetroPi

Sounds pretty sweet!

that's a pretty cool solution!

programmer mind kicking in here -- if the render layers have any sort of sequence -- i.e. you're literally incrementing a variable by one, it might be worthwhile to have a check if there's an upper limit

the fix there would probably be to loop back to zero and you'd end up with a little pause again, but it'll be a little better than having someone restart your video art in the MoMA some day.

yep - i did set the render layers to increment by one with each cycle. the (scanty) documentation doesn't mention anything about an upper limit, but i reset it back to 1 whenever it reaches 999, so hopefully it won't cause any issues.

right now the biggest issue is performance. there's a tradeoff between the length of the videos, the dimensions of the video - if the length gets too short or the dimensions too big, it sometimes crashes. right now i have it cycling through 1-second long, 500x500 videos, and it's smooth! i want to make the videos as short as possible so that as the sensor tracks the human(s) observing it, they update as close to real-time as possible. the biggest constraint will probably be the sensor and how quickly it can send its new data to python via GPIO. i haven't even really looked into that yet, but i'm hoping that it can do it in near-real-time. i flailed my way through an afternoon arduino workshop a few years ago, and the sensor readings made it to the Serial almost instantaneously, so i'm hopeful that will be the case with this as well.

another interesting dilemma will be what to do if the sensor picks up more than one object on the physical stage. i guess i should just stop fucking around and buy a sensor and give it a shot.

is the ultrasonic beam collimated or focused somehow? how do you plan to arrange the sensors?

i...don't know yet.

a kind stranger in chicago contacted me a few months ago and offered to help me out with general programming stuff, and he went ahead and bought an ultrasonic sensor (i forget which one - an inexpensive one for sure) and started playing around with it. we met up a few weeks ago to talk about our progress and he said that the sensor was able to pick up humans in the room, up to about 10-12 feet away. i asked him if it was able to detect more than one at once, and he said that it did. but i haven't met with him since, and i haven't tried it myself.

so i'm not sure at the moment. ideally i would only need one sensor, but i might end up using two (one from the front, one from the side) in tandem? do you have any thoughts? i should probably have held off on mentioning all that until i had done even 10 minutes of research.

(that last sentence applies to everything i post on ilx btw)

eventually i may end up shifting to one master RPi equipped with the sensor(s) which sends out instructions to a set of subsidiary RPis with displays the videos accordingly. the ultimate evil plan is to go with a modular approach, so i can deploy individual RPIs with displays wherever i want in the room and they'll still be able to track the position of the observer.

i'm thinking each individual unit would only cost ~ $30 for the Rpi + another $30-60 for a small HDMI display. obviously i want to get one unit working well before buying any more, but having a set of 3 would be nice.

ok, after ~5 minutes of looking into i see that ultrasonic sensors only measure the distance to the closest object, so tracking more than one object would be tricky. that's fine though, i'm ok with it only identifying the closest object. that's almost better, in fact, because i'm planning on creating video clips that involve a human in a different dimension trying to reach across the void and make contact with the observer. almost like a frightened animal, so it would make sense that the video faces would only track the closest object, because the idea is that it's a struggle to clearly connect across the void in the first place. i have been watching a lot of unsolved mysteries lately.

as far as the positioning goes, i think i would place the sensor above the RPi + display, so that it's aiming down onto the stage at about a 45 degree angle. If I put it lower and aim it out horizontally, it could pick up objects that are 10-20 meters away, and i don't want that. i only want it to activate if someone actually approaches to take a closer look. (otherwise, the default/inactive state is looping videos of the faces going about their business, bored, or maybe trying and failing to establish contact across the void). if i place the sensor a little higher though, and aim it downward, i'll be able to define the sensor's field of vision.

does that sound right? i am making this up as i go but with the confidence that it should be possible.

So does the sensor return a reflection vs. distance or just a single distance value? The signal that the sensor emits will have some angular cone shape. The received signal will also have a cone-shape dependence away from the receiving aperture. "closest object" likely just means "largest reflection". It seems like it could get more complicated if you have multiple objects of different scattering cross-sections within these sensor cones. But it may also work out fine, as you said. You can probably do more with the signal processing if the sensor return something more than just a single distance value, but maybe that'd be overly complicated. Also, any signal processing might be too slow for your purpose.

I would also think that you'd direct the sensor along the axis you wish to detect movement. Otherwise it will be harder to distinguish movement along the axis you care about versus the orthogonal axis.