• What Are You Working On? - April 2014
    1,313 replies, posted
Well, it's a binary file, just use [CODE]sh BoxCopter.sh[/CODE] :P
[QUOTE=Lumaio;44669084]Well, it's a binary file, just use [CODE]sh BoxCopter.sh[/CODE] :P[/QUOTE] [code] ~/D/Linux> sh BoxCopter.sh Error: Unable to access jarfile /home/shodan/Downloads/Linux/BoxCopter.sh : numeric argument required 0[/code] How about just giving us a jar. Because from what I can see it's just a jar encoded into a messy shell script which is silly.
[QUOTE=Mega1mpact;44669127] How about just giving us a jar. Because from what I can see it's just a jar encoded into a messy shell script which is silly.[/QUOTE] Okie. I'll also hand over the source to everyone here [URL="http://patrick.kecklers.com/Storage/BCSource.rar"]Source[/URL] Gimme 2 minutes to grab the jar and put it up. [B]Edit:[/B] Having minor difficulties installing java8 on my linux machine. [B]Edit2:[/B] [URL="http://patrick.kecklers.com/Jar.zip"]Here it is[/URL]
That feel when you optimize an interpreter so much the terminal can't keep up with the resource-heavy test apps. :v: [sp]I'm soon getting 60fps before the xbox, suck it down microsoft[/sp]
[QUOTE=MatheusMCardoso;44650449]Maybe there's a Ballmer peak for Marijuana.[/QUOTE] There is for me!
[QUOTE=Darwin226;44668889]If I understood correctly, it looks for a certain pixel pattern in an image right? What can it be used for?[/QUOTE] Well, the first thing that comes to mind is text recognition in book scans, Google uses stuff like this for maps, recognizing licence plates, just about anything that involves "computer vision" in this sense. In my case I use it for something entirely different. The idea is that for each pixel in an image, I'll look for matching pixels in a region around it. Of course, comparing two single pixels is meaningless in a statistical sense, so it actually compares blocks, and then the center pixels of these blocks are assumed to be similar if their respective blocks are similar too. This is all nice, but it won't work if two blocks are identical but rotated wrt. one another. So you either need a similarity measure that is rotationally invariant (which is almost non-existent), or you make the similarity measure rotationally invariant by making sure that before you compare two blocks, you are sure they are aligned. But if you account for that rotation, your algorithm becomes extremely slow and just useless for any reasonably sized image. Unless you implement it in CUDA, which can greatly accelerate element-wise algorithms, which is what I did. With these matching pixels I will estimate the noise for every pixel and use that estimation to calculate a new pixel value (with fancy statistics), thus removing the noise. Roughly. If a photographer uses a longer exposure to average every pixel in time (thus increasing S/N ratio with time), then I use in a sense a single image to average it with itself by finding similar pixels in the image for every pixel in that image (clusterfuck). This is relevant for MRI, because acquiring a single image might take quite a while and if there's a child or a mouse in there, then they won't enjoy being in there for a long time and they'll start moving and blurring the image.
Back on my commercial contract again. Finally got the client to work without a hitch after fixing a [URL="https://bitbucket.org/devonit/qjsonrpc/issue/10/invokeremotemethodblocking-invokes-slots"]bug[/URL] in QJSONRPC. (well, sort of) [img]https://dl.dropboxusercontent.com/u/27714141/netforeman_splash.png[/img] [img]https://dl.dropboxusercontent.com/u/27714141/netforeman_server_client.png[/img] God, I hate database work.. so tedious and boring. But that was somewhat of a challenge. It works like this: RPC calls are made to populate the item model, which is closely tied with the table view. The RPC calls are bridged over a SSL/TLS connection to a remote server, which maintains instances of QSqlRelationalTableModel internally, those are populated by a MYSQL connection to a SQL server. The server also handles user credentials and permissions through the same RPC mechanism and the same database. [sp]I'm just not giving you the full phone numbers you jerks[/sp] [editline]29th April 2014[/editline] Oh fuck I forgot about my own phone number [editline]29th April 2014[/editline] Problem solved.
[QUOTE=Map in a box;44668767]Is that visual studio 05?[/QUOTE] 09
[QUOTE=voodooattack;44670927]Back on my commercial contract again. Finally got the client to work without a hitch after fixing a [URL="https://bitbucket.org/devonit/qjsonrpc/issue/10/invokeremotemethodblocking-invokes-slots"]bug[/URL] in QJSONRPC. (well, sort of) [img]https://dl.dropboxusercontent.com/u/27714141/netforeman_splash.png[/img] [img]https://dl.dropboxusercontent.com/u/27714141/netforeman_server_client.png[/img] God, I hate database work.. so tedious and boring. But that was somewhat of a challenge. It works like this: RPC calls are made to populate the item model, which is closely tied with the table view. The RPC calls are bridged over a SSL/TLS connection to a remote server, which maintains instances of QSqlRelationalTableModel internally, those are populated by a MYSQL connection to a SQL server. The server also handles user credentials and permissions through the same RPC mechanism and the same database. [sp]I'm just not giving you the full phone numbers you jerks[/sp] [editline]29th April 2014[/editline] Oh fuck I forgot about my own phone number [editline]29th April 2014[/editline] Problem solved.[/QUOTE] Looks fancy and fun :D
[QUOTE=Eric95;44662037]What the fuck is our Ludum Dare game even [vid]http://dl.dropboxusercontent.com/u/679615/video/whoa.webm[/vid][/QUOTE] Are you making any games outside of Ludum Dares? Whoever is doing your art is really good (the previous game about the dog also had really good models). It'd be a waste if they aren't using their talents in making some bigger/real games.
[QUOTE=Duskling;44671321]Looks fancy and fun :D[/QUOTE] Thanks, I'm trying to make it really easy to use.
going to put my minecraft classic server on a VPS tamarrow to see how much yall will flood it [editline]28th April 2014[/editline] In other news, drmemory is useful
Spending a lot of hours refactoring all of the [url=https://i.imgur.com/f6dBuwx.png]mobile game [/url] code :v: but it's being so worth it. After almost a year of making drastic changes and changing stuff around it had become one hell of a mess - what would have been an easy task like adding a new monster becomes a daunting one when there's unpredictable behavior going on on the background. I'm still refactoring, but it's a great feeling when you start noticing how much better the code base is. Everything works as it should and it's very easy to modify things without breaking others. Can't recommend enough to start thinking about the code structure early on, especially on mid-to-big sized projects. It can avoid many headaches in the future. Maybe it has been posted before, but this is a good place to start: [url]http://gameprogrammingpatterns.com/[/url]
It has been a long night, but it has totally payed off: 14x performance gains without going overboard with RAM usage! (still under 1.5mb!) I cached: - tokens (pre-lexing step) - endif/else/case positions - variable reference table index
[QUOTE=Map in a box;44673032]going to put my minecraft classic server on a VPS tamarrow to see how much yall will flood it [editline]28th April 2014[/editline] In other news, drmemory is useful[/QUOTE] I call dibs on finding the first exploit :v:
Man I can't say how inspiring these threads are. I can't do much beyond hacking around in existing software, what you guys do is incredible. e: work instead of whining -- the work continues! prop functions, writing more stuff in Lua than in C, and cross-compiling on Linux! [t]http://puu.sh/8s0kM.png[/t]
Do any of you guys know if I can get OpenAL through homebrew? My configure script can't seem to find it
TCP Backdoor tool tester thing [IMG]http://cra0kalo.com/upload/staging/store/push/AttackMe.png[/IMG] Download here and src code [url]http://dev.cra0kalo.com/?p=90[/url]
[QUOTE=adnzzzzZ;44671361]Are you making any games outside of Ludum Dares? Whoever is doing your art is really good (the previous game about the dog also had really good models). It'd be a waste if they aren't using their talents in making some bigger/real games.[/QUOTE] I work together with a friend as a duo, the same girl I make Senpai Club (an anime parody series thing) with. We call ourselves makebabi.es (that's the URL to our website), and she does character design, concept/promo art, and texture work while I do all of the 3d work besides texturing (modeling, UV mapping, rigging, animation), programming, music, sound effects, and putting it all together. Right now we're not making any games outside of Ludum Dare, since we're focusing on Senpai Club, but we might do something bigger game-related in the future.
[QUOTE=cra0kalo;44674580]TCP Backdoor tool tester thing [IMG]http://cra0kalo.com/upload/staging/store/push/AttackMe.png[/IMG] Download here and src code [url]http://dev.cra0kalo.com/?p=90[/url][/QUOTE] [t]http://i.imgur.com/djT3x7A.png[/t] So I'm safe? :v:
[QUOTE=voodooattack;44670927][...] [sp]I'm just not giving you the full phone numbers you jerks[/sp] [editline]29th April 2014[/editline] Oh fuck I forgot about my own phone number [editline]29th April 2014[/editline] Problem solved.[/QUOTE] Please tell me you're developing against realistic but not real customer data... The program looks really nice though, how well does it integrate with other OS UX? As a side note: If you type Arabic into a text box, does the OS/whatever insert RTL and/or LTR codepoints before/after the text or do you have to set the text box to RTL in code?
[QUOTE=MattJeanes;44675698][t]http://i.imgur.com/djT3x7A.png[/t] So I'm safe? :v:[/QUOTE] Yep
[QUOTE=Number-41;44670726]Well, the first thing that comes to mind is text recognition in book scans, Google uses stuff like this for maps, recognizing licence plates, just about anything that involves "computer vision" in this sense. In my case I use it for something entirely different. The idea is that for each pixel in an image, I'll look for matching pixels in a region around it. Of course, comparing two single pixels is meaningless in a statistical sense, so it actually compares blocks, and then the center pixels of these blocks are assumed to be similar if their respective blocks are similar too. This is all nice, but it won't work if two blocks are identical but rotated wrt. one another. So you either need a similarity measure that is rotationally invariant (which is almost non-existent), or you make the similarity measure rotationally invariant by making sure that before you compare two blocks, you are sure they are aligned. But if you account for that rotation, your algorithm becomes extremely slow and just useless for any reasonably sized image. Unless you implement it in CUDA, which can greatly accelerate element-wise algorithms, which is what I did. With these matching pixels I will estimate the noise for every pixel and use that estimation to calculate a new pixel value (with fancy statistics), thus removing the noise. Roughly. If a photographer uses a longer exposure to average every pixel in time (thus increasing S/N ratio with time), then I use in a sense a single image to average it with itself by finding similar pixels in the image for every pixel in that image (clusterfuck). This is relevant for MRI, because acquiring a single image might take quite a while and if there's a child or a mouse in there, then they won't enjoy being in there for a long time and they'll start moving and blurring the image.[/QUOTE] Wouldn't it be a problem if you use disjunct parts of the image to denoise if an important feature is smaller than the sample size? [URL="http://www.theregister.co.uk/2013/08/06/xerox_copier_flaw_means_dodgy_numbers_and_dangerous_designs/"]Xerox had this problem in some of their copiers, they would change numbers imperceptibly[/URL], under certain circumstances.
[QUOTE=Tamschi;44675844]Please tell me you're developing against realistic but not real customer data...[/QUOTE] Uhh.. well.. it's not like you can scrape anything vital from that screenshot, I made sure of it! [QUOTE=Tamschi;44675844]The program looks really nice though, how well does it integrate with other OS UX?[/QUOTE] Thanks. It's all cross-platform and uses JSON to communicate so I'd say pretty well. The final product will run as a Linux server with Windows clients, so that's the end goal. [QUOTE=Tamschi;44675844]As a side note: If you type Arabic into a text box, does the OS/whatever insert RTL and/or LTR codepoints before/after the text or do you have to set the text box to RTL in code?[/QUOTE] Qt handles this, it automatically switches to RTL layout if you enter any text in Arabic.
[QUOTE=voodooattack;44676061]Uhh.. well.. it's not like you can scrape anything vital from that screenshot, I made sure of it![/QUOTE] You should still consider getting some dummy data, since it prevents any liability issues in case something leaks. It's true that I can't get any important information from here, but that might just be because I can't type much of it into Google to see if there's a connection. [URL="http://www.wired.com/2012/08/apple-amazon-mat-honan-hacking/all/"]Partial data can also be a huge issue, if some service uses it for identification.[/URL] In this case you only display the "common part", so it should be not too serious, but there might be conclusions that can be drawn from this data when connected to other things. For example, I might be able to complete the addresses by cross-checking with a phone book even if it's not in there directly. [URL="https://en.wikipedia.org/wiki/Telephone_numbers_in_Egypt"]The land line gives me the area in which to check street names and the mobile number tells me something about the carrier[/URL], which I might use to pose as a technician (since someone's carrier choice is usually not considered "public" info by laypeople). From the dates I can tell that these subscriptions(?) here are probably not current, but I might be able to find out who you're making this for once it's deployed, if the name of the software doesn't change (or just by calling people, if I knew the language). I could use this to make ex-customers believe there was a serious data breach or send this as "proof" of hacking to a news outlet (which probably wouldn't work though, since normally the data is not redacted as much in such a case). In conclusion: Even insignificant looking pieces of data can tell you a whole lot about someone if you have an internet connection nearby. Please consider creating some fakes. [QUOTE]Thanks. It's all cross-platform and uses JSON to communicate so I'd say pretty well. The final product will run as a Linux server with Windows clients, so that's the end goal.[/QUOTE] I actually meant in regards to native look-and-feel, but since you use QT that should handle it automatically. [QUOTE]Qt handles this, it automatically switches to RTL layout if you enter any text in Arabic.[/QUOTE] I assume it inserts the control characters then, and it should work similarly with other frameworks. At least that part of globalization is nothing to worry about it seems.
[QUOTE=Luni;44674369]Man I can't say how inspiring these threads are. I can't do much beyond hacking around in existing software, what you guys do is incredible. e: work instead of whining -- the work continues! prop functions, writing more stuff in Lua than in C, and cross-compiling on Linux! [t]http://puu.sh/8s0kM.png[/t][/QUOTE] Just a note I would use require, and not dofile. If you call dofile twice you execute the same file twice, if you call require twice it recognizes that you have already required that file and does nothing, among a myriad of other small differences.
[QUOTE=Tamschi;44676059]Wouldn't it be a problem if you use disjunct parts of the image to denoise if an important feature is smaller than the sample size? [URL="http://www.theregister.co.uk/2013/08/06/xerox_copier_flaw_means_dodgy_numbers_and_dangerous_designs/"]Xerox had this problem in some of their copiers, they would change numbers imperceptibly[/URL], under certain circumstances.[/QUOTE] There are ways to deal with this by using a weighing, for example based on the euclidean distance to the reference block, or by limiting your search window. You actually provide a very valid point that might explain the trouble I'm having now. I thought taking the mean/median would already give some nice results, but it tends kill detail. In the end it's always a tradeoff, you can't create a perfectly noiseless image because you always lose information as soon as you introduce noise. Only if you make the right assumptions you can try to restore it a little.
New version of my guide to SSAO, now with angle-based SSAO (used in UE4): [url]https://dl.dropboxusercontent.com/u/17984511/SSAO/110157%20-%20SSAO%20Draft%203.pdf[/url]
[QUOTE=Natrox;44678096]New version of my guide to SSAO, now with angle-based SSAO (used in UE4): [url]https://dl.dropboxusercontent.com/u/17984511/SSAO/110157%20-%20SSAO%20Draft%203.pdf[/url][/QUOTE] Would you mind if I downloaded it for further reference? I like the explanations and the code samples.
[QUOTE=commander204;44678214]Would you mind if I downloaded it for further reference? I like the explanations and the code samples.[/QUOTE] That's not a problem, but be aware that it's not finished yet. I've yet to add HBAO / HDAO and scalability analysis.
Sorry, you need to Log In to post a reply to this thread.