|
Not terribly sure where I'm going with this game, but it's turning into a point-and-click adventure game with platforming elements. Watch it in HD to see my pretty, pretty particle emitter  Short video, but I've only got about 2-3 minutes of content in the game so far.If you want to check it out, it's at http://beneath.dungeo.com/ https://www.youtube.com/watch?v=-KT1Tsyy8hw
|
#
?
Sep 5, 2014 23:04
|
|
|
|
| # ? Apr 27, 2024 03:15 |
|
|
 A project I started in Jan. 2013 is shipping on Tue 
|
|
#
?
Sep 5, 2014 23:14
|
|
|
 Worked on a glitch shader yesterday! Most of it was altering a mosaic shader to step down the pixel size with distance from a set point, as well as faking chromatic aberration using a modified edge detection shader (so you'll notice bits are magenta / green) I'm not totally sure what the additional artifacts are from -- they don't show up if the cube is static, regardless of the starting rotation of the cube. I want to blame some Processing but it's probably some kind of issue about sampling pixels that I'm not handling correctly.
|
|
#
?
Sep 6, 2014 00:55
|
|
|
I wanted to learn the twitter API so I made a stupid bot that will make word clouds out of text files you tweet at it (or if you just tweet at it, it will make one out of your tweet): Maybe it won't break, either way it was fun
|
|
#
?
Sep 11, 2014 00:45
|
|
|
Rides Naked posted:I wanted to learn the twitter API so I made a stupid bot that will make word clouds out of text files you tweet at it (or if you just tweet at it, it will make one out of your tweet): Make it find tweets based on a certain hashtag and make a wordcloud out of that!
|
|
#
?
Sep 11, 2014 23:17
|
|
|
 I'm writing a Yelp scraper in Python. It uses a single coordinate and math to generate multiple points and make multiple queries to the Yelp API instead of a single, lame 20-result query. Lots of work to be done, but I found getting maps to create was gratifying.
|
|
#
?
Sep 11, 2014 23:46
|
|
|
Literally Elvis posted:
Hi there Austinite.
|
|
#
?
Sep 12, 2014 00:20
|
|
|
You should make it add music tracks to each flag from the year that the establishments were founded. You know, synergy and all that.
|
|
#
?
Sep 12, 2014 00:28
|
|
|
kayakyakr posted:Hi there Austinite. Hi there, Austinite.
|
|
#
?
Sep 12, 2014 18:09
|
|
|
avidal posted:
Bonus points if either of you know what the origin point in my screenshot is.
|
|
#
?
Sep 12, 2014 20:11
|
|
|
Literally Elvis posted:Bonus points if either of you know what the origin point in my screenshot is. Tupelo, Mississippi?
|
|
#
?
Sep 12, 2014 20:22
|
|
|
Literally Elvis posted:Bonus points if either of you know what the origin point in my screenshot is. Center looks like 6th and Red River so... Esther's Follies? El Sol y La Luna? One of those random food trailers? *shrug*
|
|
#
?
Sep 12, 2014 21:01
|
|
|
 I turned the zipcodes of Florida into an equalizer. Uses PostGIS + Node.js / Express + Three.js + WebAudio API +  custom shader It's pretty sweet how little actual code this is. Also I get to play with this for work. Tres Burritos fucked around with this message at 02:50 on Sep 13, 2014 |
|
#
?
Sep 13, 2014 02:47
|
|
|
netcat posted:Make it find tweets based on a certain hashtag and make a wordcloud out of that! I threw that in over the weekend for the hell of it. You just tweet a hashtag at it, like the completely understandably popular #LiamBrokeHisArmOnHisAnaconda  Ah the insights big data can provide
|
|
#
?
Sep 15, 2014 20:36
|
|
|
Alibaba day, the world gets to see stuff I worked on. Similar to a few pages back, all the screens showing prices and scrolling things: Big 8-panel wide ticker, each panel is actually orientated vertically and rotated using XRandR in order to bypass acceleration defects in nVidia's drivers:  Big board comprising 4x4 panels, ticker slowed down so not much of a performance problem.  Small board at the main entrance which I think Fox like to sit a camera permanently underneath.  Another variation of the small board, spot the classy HTML encoding error by editorial.  The floor is unsurprisingly busy, hiding behind CNBC all morning.  The first trade indicators as the institutions work on IPO price.  Really rear end looking LED panels showing a barely legible company name.  For some reason this half of the floor has less people, probably because no cameras.  CNBC presenters going blah, blah, blah. 
|
|
#
?
Sep 20, 2014 01:42
|
|
|
Well that's a lot cooler than what I'm gonna post (congrats!). I made TicTacToe in Swift (to start getting used to it) with an unbeatable AI:  https://github.com/BayPhillips/TicTacToe Doh004 fucked around with this message at 04:28 on Sep 22, 2014 |
|
#
?
Sep 22, 2014 04:24
|
|
|
MrMoo posted:Alibaba day, the world gets to see stuff I worked on. Similar to a few pages back, all the screens showing prices and scrolling things: Did anyone catch some poo poo for using that panel? With all the money floating around there, you'd think someone would have sprung for something a little nicer.
|
|
#
?
Sep 22, 2014 21:23
|
|
|
Two are broken on one end of the floor, no one can be assed to fix them as it takes hours to get up there and change it. Everything is run by different companies and contractors, NYSE itself was recently bought out by ICE and a lot of things needing attention. Everyone waiting for the new exchange name for example.
|
|
#
?
Sep 23, 2014 00:45
|
|
|
I had to take some time away from game development to work on my bread and butter - an Augmented Reality app for a nature reserve. I spent the last week completely rewriting the whole thing, and it all works about a million times better now. I couldn't help but get a screenshot (and video) of this while testing, though:  and the video: http://youtu.be/6HyrH1ErEFo
|
|
#
?
Sep 23, 2014 15:56
|
|
|
Pi Mu Rho posted:I had to take some time away from game development to work on my bread and butter - an Augmented Reality app for a nature reserve. I spent the last week completely rewriting the whole thing, and it all works about a million times better now. This is cool, but I'm confused as to what exactly I'm seeing. The butterfly is being superimposed onto what? Something like Google Glass? An iphone screen?
|
|
#
?
Sep 23, 2014 16:03
|
|
|
In this case, I did a screen recording from the tablet I was testing on. So yes, you basically use your phone/tablet screen as a viewer.
|
|
#
?
Sep 23, 2014 16:14
|
|
|
Still tinkering and slowly improving my Chip8 IDE. You can try it here or check out the thread.
|
|
#
?
Sep 23, 2014 16:27
|
|
|
Working on integrating CorePlot into my project that pulls engine data and displays it on a gauge/chart. Lots of work to do but the real-time functionality seems to be working ok. More info here.
|
|
#
?
Sep 23, 2014 16:59
|
|
|
Pi Mu Rho posted:I had to take some time away from game development to work on my bread and butter - an Augmented Reality app for a nature reserve. I spent the last week completely rewriting the whole thing, and it all works about a million times better now. I'm not sure how it works but it is pretty dang cool.
|
|
#
?
Sep 24, 2014 01:44
|
|
|
I have some more BeagleBone Black Android work just about wrapped up. I pulled documentation from half-a-dozen different places and pushed past the old Android build process for Android 4.2.2 (Jelly Bean) to move on to integrating the BBB's 3.8 kernel tree into AOSP's Android 4.4.4 (KitKat). It actually works on all of my various LCD cape boards and the built-in HDMI. Just plug them in and go without changing any configuration settings and they all "magically" work because the proper device tree overlays are inserted into the active device tree on-the-fly by the kernel's capemgr: 4D Systems capes:   CircuitCo capes:    Built-in HDMI (via an awful HDMI-to-NTSC video capture):  
|
|
#
?
Sep 24, 2014 05:08
|
|
|
Hughmoris posted:I'm not sure how it works but it is pretty dang cool. The Idea of AR is that a symbol of some kind is detected by the AR software and on the device rather than seeing the symbol it is replaced with a 3D model placed into the world.
|
|
#
?
Sep 24, 2014 07:35
|
|
|
Playing around with OpenCV to see if I can automatically turn these hand-written probate registers into searchable text. It's a pretty limited vocabulary (names, job titles) which should help once I get to the OCR part. The techniques were found in random papers with google. Original  Cropped to the main section by using Hough transform to locate horizontal & vertical lines, then separated into text lines by summing gray values of each row of pixels. The rows with the least writing will have the highest sums (by being closer to white = 255). Smoothing the sums with a rolling average & getting local maxima gives the offsets for the space between each line, above in blue.       De-slanted by trying differently angled skew transforms & selecting the one that gives the most tall peaks in its summed columns (= most vertical pen strokes), then separated into letters as above but using columns instead of rows. I was actually attempting to just separate into whole words, but it's pretty close to getting the correct letter boundaries, probably just need to tune it a bit. Looks like I'll also need to pad the lines a bit more so less gets cut off. Next part is probably determining where each entry begins (where the lines start with big letters perhaps?), and then the OCR part. Carthag Tuek fucked around with this message at 23:25 on Sep 27, 2014 |
|
#
?
Sep 27, 2014 23:21
|
|
|
That's barely legible for humans, funny if a machine learning computer system could do better.
|
|
#
?
Sep 28, 2014 00:02
|
|
|
Snapchat A Titty posted:Playing around with OpenCV to see if I can automatically turn these hand-written probate registers into searchable text. It's a pretty limited vocabulary (names, job titles) which should help once I get to the OCR part. The techniques were found in random papers with google. Sup fellow OpenCV buddy, I've been working with trains instead of text. (Source video I was using for this older gif had dupe frames which was messing with my motion detector)  Your post has given me fodder to try a few different things. Are you planning on writing a new classifier or trying to feed that into an existing one (e.g. tesseract)?
|
|
#
?
Sep 28, 2014 00:06
|
|
|
MrMoo posted:That's barely legible for humans, funny if a machine learning computer system could do better. It's a late 18th century mix of latin cursive (the names) and gothic cursive (the rest), also it's in Danish. Luckily it's one of the more legible hands, some of them are atrocious. It took me a while to get familiar with gothic cursive but I can read it decently now. My transcription: pre:Andresen |: afgl: Christence Nielsdatter Juul, Enke af forrige Kok Morten Andresen Andersen |: Charlotte Marie :|, Oldermand for fær- gelauget her i Staden Jörgen Stephensens hustru Anne Margrethe, Enke af Under-Jæger Fran- çois Rousselle  smileys were how they wrote parentheses back then. The writer has forgotten to close the first oneTrabisnikof posted:Sup fellow OpenCV buddy, I've been working with trains instead of text. (Source video I was using for this older gif had dupe frames which was messing with my motion detector) I haven't decided how to do the OCR part yet. Looking at papers for popular methods, there seem to be a lot of mentions of Hidden Markov Models, not so much any existing OCR engines.
|
|
#
?
Sep 28, 2014 01:22
|
|
|
hendersa posted:I have some more BeagleBone Black Android work just about wrapped up. I pulled documentation from half-a-dozen different places and pushed past the old Android build process for Android 4.2.2 (Jelly Bean) to move on to integrating the BBB's 3.8 kernel tree into AOSP's Android 4.4.4 (KitKat). It actually works on all of my various LCD cape boards and the built-in HDMI. Just plug them in and go without changing any configuration settings and they all "magically" work because the proper device tree overlays are inserted into the active device tree on-the-fly by the kernel's capemgr:  I finally released this thing: http://www.bbbandroid.org I use repo to combine 361 different git repositories together to make the Android image. And, since we need a screenshot, here is the website that I just put up this morning for the project:  Last time that I did this, I received thousands of mails. Let's hope that the response is a touch less enthusiastic this time around.
|
|
#
?
Sep 28, 2014 18:15
|
|
|
I've been playing around in Unreal Engine 4 for a while. I've also started to learn blender.
|
|
#
?
Sep 28, 2014 21:05
|
|
|
Valtis posted:Baby's first compiler What's up compiler buddy? I just started working on one too. I haven't built the parser for it yet (but the syntax will probably end up looking Lisp-ish) but I've just finished up implementing first-class functions and closures. Nothing special yet but it's pretty fun to write, I recommend making one to others. 
|
|
#
?
Oct 2, 2014 18:17
|
|
|
I am putting together a sample Android HAL for a variety of hardware, and I have a sample up and running: The sample hardware uses some GPIOs (button input and LED output), A SPI-based temp/pressure sensor, and an I2C FRAM. Press the button and the LED comes on, the sensor data is collected, the data is written to the FRAM, and then the data is retrieved from the FRAM to ensure it wrote properly. 
|
|
#
?
Oct 5, 2014 21:36
|
|
|
Snapchat A Titty posted:Playing around with OpenCV to see if I can automatically turn these hand-written probate registers into searchable text. It's a pretty limited vocabulary (names, job titles) which should help once I get to the OCR part. The techniques were found in random papers with google. I wonder if it's worth trying to use m-turk/crowdflower to do this. Edit: In combination with your work too, I mean. I think m-turk is US only, but crowd flower gave me $100 free credit when I signed up (and I haven't used it yet) EAT THE EGGS RICOLA fucked around with this message at 22:35 on Oct 9, 2014 |
|
#
?
Oct 9, 2014 21:58
|
|
|
EAT THE EGGS RICOLA posted:I wonder if it's worth trying to use m-turk/crowdflower to do this. What i want to see is the Spell checker that RTM (internet worm) 's dad wrote... It apparently was a spell checker that got 99% of mis spellings and had no dictionary. It was supposed to use pattern matching to work out the spelling rules.
|
|
#
?
Oct 10, 2014 07:09
|
|
|
I've owned this thing for 4 years and kept it through 3 moves and I finally got around to doing something with it. https://www.youtube.com/watch?v=8mHXzukItVk Now to figure out what exactly to utilize it for.
|
|
#
?
Oct 10, 2014 11:45
|
|
|
Trying to make some trees. At first I thought I could model them on my own but ultimately I made a python script to have them procedurally generated in blender. 
|
|
#
?
Oct 10, 2014 15:11
|
|
|
EAT THE EGGS RICOLA posted:I wonder if it's worth trying to use m-turk/crowdflower to do this. I don't think it'd be necessary to use paid transcribers, actually. There's been a lot of volunteer-driven projects in Danish genealogy circles that have done quite a lot of work. ~480,000 card catalog of Jutlanders indexed by name, title, place, year over a 15 month period: http://nygaards-sedler.dk/ ~17,000,000 census entries from 1787 onwards, project has been going for over 20 years with several censuses completely transcribed: http://ddd.dda.dk/ddd_en.htm ~1,400,000 police register cards with addresses of Copenhagen residents 1890-1920: http://www.politietsregisterblade.dk/en/om-projektet I currently envision my thing as a sort of transcription helper that continually learns a given writer's characteristics until it can supply suggestions & maybe eventually let the transcriber just hit "correct" on each word, occasionally correcting them. BTW, for the interested, look up the ICDAR (odd years) and ICFHR (even years) competition papers, lot's of interesting algorithms on offline handwritten cursive recognition.
|
|
#
?
Oct 10, 2014 17:42
|
|
|
|
| # ? Apr 27, 2024 03:15 |
|
|
ijustam posted:Now to figure out what exactly to utilize it for. Evil. Failing that you could make some sort of system that responds to sound levels/FFT data and use it at parties.
|
|
#
?
Oct 10, 2014 23:22
|
|
