What are you working on? November 2011 Edition - Developers

[QUOTE=HeroicPillow;33096115]Both.[/QUOTE] Keep in mind that games last for a single round.

[QUOTE=ROBO_DONUT;33095551]Time-reversal? Maybe he's developed some kind of anticausal game.[/QUOTE] I was actually thinking of making a 2D timeshift clone due to me not having the game yet, but I doubt anyone would play it. Thats just for my fireworks thing, its supposed to say 1 but I figured 0 was more dramatic.

Spent today optimizing the database. Although db4o doesn't really support concurrency. So even though I got player querying down to 5ms from 150ms. Its meaningless as there is constant committing going on which blocks querying so they take 1000ms anyways. Also got adding notes, changing colors and gamemode/gametype names done. [img]http://goo.gl/aZ3sf[/img]

I'm liking these pixel effects. [img]http://ahb.me/4vtZ[/img]

-snip wrong thread- Have some half-working content: [img]http://i.imgur.com/bpfqT.png[/img]

[img_thumb]http://dl.dropbox.com/u/11093974/Junk/obb-aabb.png[/img_thumb] OBB-AABB collision works (the objects move toward each other and stop when they collide). Next is OBB-OBB, which should be very simple with the way I have it setup (separating axis theorem ftw). Progress: [img]http://dl.dropbox.com/u/11093974/Junk/progress%21.png[/img] If anyone is wondering, I'm using [url=http://realtimecollisiondetection.net/]this book[/url] as a guide. It's very helpful, but the author uses the worst variable names ever (mostly single-letter variables like d, u, e, etc...) Would anyone be interested in the source code once I'm done? In this class I have to make a full physics engine, and I don't plan on releasing the whole thing here. But I feel people could really benefit from the collision-detection code.

Thinking of livestreaming coding again, anyone interested?

Have nothing else to do.

Today I finished the final update for the anonymous chat application I worked on for a client. Pretty happy with how it turned out. [t]http://i.imgur.com/FrjLx.png[/t] [t]http://puu.sh/81AW[/t] For the final layout I was inspired by [url=http://www.chatmetvreemden.nl]the website[/url] itself. Unfortunately I can't show a screenshot of the chatting as almost nobody is online at this time, but it looks something like [url=https://www.gstatic.com/android/market/nl.overv.vreemden/ss-480-1-2]this[/url]. [i]Vreemde[/i] means "stranger" in Dutch and [i]Chat met Vreemden[/i] means "chat with strangers". It also has notifications for when you've connected with a stranger and when a new message was received while the app was in the background.

Love's Image class needs a setData function.

[QUOTE=ZenX2;33097986]Love's Image class needs a setData function.[/QUOTE] love.image.newImageData...? I'm not sure what you mean, but that sets the image data.

[QUOTE=jalb;33097837]It's very helpful, but the author uses the worst variable names ever (mostly single-letter variables like d, u, e, etc...)[/QUOTE] That's how I do it. Also, mathematics in general uses single-letter variable names. If you understand the problem and have some established conventions and informative comments, you don't need full words for variable names. Granted, it depends on the scope of the variable. You probably don't want to have an 'x' in the global namespace... but it's fine for a single throw-away value in the middle of a short function.

[QUOTE=ROBO_DONUT;33098056]To clarify, I don't use single-letter variables everywhere[/QUOTE] Oh. He uses them [b]everywhere[/b]. :v: This is his OBB class: [cpp]struct OBB { Point c; Vector u[3]; Vector e; };[/cpp] [sp]c is the center point, u are the local x,y,z coordinates, e is the half extents[/sp] For someone learning this stuff, it's very inconvenient to have to remember what c, u and e are. I constantly have to go back to the beginning of the book to look at how the structs are defined.

[QUOTE=BlkDucky;33098024]love.image.newImageData...? I'm not sure what you mean, but that sets the image data.[/QUOTE] I have ImageData, which cannot be drawn on it's own, it needs be used with an image. The thing is, you can't reload an image with different data because the load function is not dealt with in the wrapper.

Cant you use a display buffer

I'm going to try that next.

[QUOTE=ZenX2;33098221]I have ImageData, which cannot be drawn on it's own, it needs be used with an image. The thing is, you can't reload an image with different data because the load function is not dealt with in the wrapper.[/QUOTE] Just overwrite the old image variable with the new image created from the ImageData Use love.graphics.newImage(imagedata) to create an Image from the ImageData

[QUOTE=amcfaggot;33093120]There needs to be a way to read real data like that. :v: Some sort of language or methodology of reading it.[/QUOTE] It wouldn't actually make sense, though. The information density is much too low. Suppose your "matrix code" symbols are displayed in a 16x16-pixel font, and assume a rudimentary 256-color (8-bit) display. Each 16x16-pixel block contains 2048 bits (16*16*8) of information, so it can display 2^2048, or about 10^617, distinct pictures. For a 24-bit display, it's much higher (6144 bits, 10^1850 possibilities). Unless your code alphabet has googols and googols of symbols, it can never convey anywhere near as much information as a picture would in the same amount of space. The idea that people can look at a screenful of matrix code and "see" a picture, like in the movies, is complete nonsense.

[img]http://dl.dropbox.com/u/13824718/ZScreen/2011-10/61_FPS-2011-11-03_03.39.56.jpg[/img] Challenge mode: 1-bit graphics. Because why not? [URL="http://dl.dropbox.com/u/13824718/ZScreen/2011-10/test.png"]Complete map.[/URL]

[QUOTE=thelinx;33098382]Just overwrite the old image variable with the new image created from the ImageData Use love.graphics.newImage(imagedata) to create an Image from the ImageData[/QUOTE] I've tried that, it's pretty slow. An easy way to set individual pixels on the screen would be quite helpful. That functionality is basically the only reason I use processing, at all. [editline]2nd November 2011[/editline] I was just wondering about it. No real need to do anything. Now that I think about it, what I need is basically PixelEffect.

[QUOTE=ZenX2;33098598]I've tried that, it's pretty slow. An easy way to set individual pixels on the screen would be quite helpful. That functionality is basically the only reason I use processing, at all.[/QUOTE] Your two choices are basically: horrible, horrible speed or using a shader.

[QUOTE=BlkDucky;33098513][img]http://dl.dropbox.com/u/13824718/ZScreen/2011-10/61_FPS-2011-11-03_03.39.56.jpg[/img] Challenge mode: 1-bit graphics. Because why not? [URL="http://dl.dropbox.com/u/13824718/ZScreen/2011-10/test.png"]Complete map.[/URL][/QUOTE] As simple as those graphics are, they really still portray great feeling with the context of their location, and the lack of colors.

I made some changes to the path finding algorithm. It now works by checking every tile in every direction until it can no longer move. It's a real brute force method that's really inefficient. On smaller cases, the impact is minimal; in big cases, like the fighter, the game literally hangs for a couple seconds. Regardless, it works: [img]http://img510.imageshack.us/img510/6247/awscreenshot07a.png[/img] [img]http://img440.imageshack.us/img440/8576/awscreenshot07b.png[/img] [img]http://img46.imageshack.us/img46/1046/awscreenshot07c.png[/img] Above (not seen): the game freezing for a couple seconds while it loads [b]every possible solution[/b] [editline]3rd November 2011[/editline] Any suggestions on optimizing the algorithm? And before you suggest adding a check before a tile is added, doing that causes problems. One tile may be the end of the line of one path, but just another node in another path.

Okay, this is just ridiculous. If I just render the ImageData to an image, it's slow. If I want to use a pixel effect, I'd need to send it an Image to actually use the data.

[img]http://puu.sh/81W7[/img] I severely improved the online multiplayer. Clearly my game is superior.

[QUOTE=ZenX2;33099355]Okay, this is just ridiculous. If I just render the ImageData to an image, it's slow. If I want to use a pixel effect, I'd need to send it an Image to actually use the data.[/QUOTE] What are you trying to do exactly? In my experience LÖVE handles most situations as well as possible. Because it's a framework rather than an engine it doesn't abstract as many concepts as many other tools (like Processing), so coming from something that abstracts nearly everything might be confusing because the way computers - rendering APIs specifically - work can be different from what you expect.

I'm trying to figure out a way to manipulate individual pixels on the screen.

[QUOTE=ZenX2;33099496]I'm trying to figure out a way to manipulate individual pixels on the screen.[/QUOTE] Wouldn't shaders work? What exactly are you trying to do?

[QUOTE=Felheart;33099635]Wouldn't shaders work? What exactly are you trying to do?[/QUOTE] Passing color values for individual pixels, based on x,y coordinates, to love's shader system requires converting the tables or whatever you are using to imagedata, creating an image based on the imagedata, passing the image to the shader, and then setting pixel colors based on the pixels of the image. Basically, ~data~ -> ImageData (using setPixel or mapPixel, whatever) -> Image -> Shader There's no faster way besides just rendering the Image, as far as I can tell.

[QUOTE=BlkDucky;33098513][img]http://dl.dropbox.com/u/13824718/ZScreen/2011-10/61_FPS-2011-11-03_03.39.56.jpg[/img] Challenge mode: 1-bit graphics. Because why not? [URL="http://dl.dropbox.com/u/13824718/ZScreen/2011-10/test.png"]Complete map.[/URL][/QUOTE] I like how the houses have those cute little faces on them!