Hahaha wow, I've never seen that page.
at least that simplifies my search for why my server is down.
[QUOTE=AngryChairR;49610654]Hahaha wow, I've never seen that page.[/QUOTE]
First world problems.
So each time ive decided that the structure I used (programming really IS googling stackoverflow :v:) is inadequate and decided to rebuild. And each time the scope has increased but actual function has decrease, lol.
So I tore it all down and started over, but now I'm working on N-body stuff. So far, I have made a 1x10 9-d array structure. 3-dimensionality for distance, velocity, and acceleration.
So far the structure to dynamically generate the various functions is a bitch, but I've also added the ability to load from external files so I can import standard "systems" for simulation. Definitely going to have to find a better algorithm, I tested more last night after fixing my first demo and instabilities developed quickly.
Also part of the problem is that I keep forgetting to upload my work to my drive, so I have a version I right during the day on my surface, then evening version on desktop, then new version next day on surface again :v:
Balancing game rules is harder than it looks :cry:
[img]http://i.imgur.com/fJMOQsz.png[/img]
The big day is here! My first credit in a LEGO game (and indeed my first credit in a game with a retail/box release) :D
LEGO Marvel's Avengers is available right now to customers in the US, as well as [url=http://store.steampowered.com/app/405310/]on Steam[/url] globally. Europe will sadly have to wait until this Friday (29th) for the retail release.
My name in the credits :D
[t]https://i.imgur.com/xycCMX3h.jpg[/t]
I just wanted to try the Unity GI system... how did this happen?
[img]http://i.imgur.com/I1KSrbS.png[/img]
I made a game in only a weekend: [url]https://www.microsoft.com/da-dk/store/games/sequence-nine/9nblggh5gdgt[/url]
Coming to a private tracker near you! (soon-ish)
[img]http://i.imgur.com/MCxQUCO.png[/img]
[QUOTE=Mega1mpact;49613291]Coming to a private tracker near you! (soon-ish)
[img]http://i.imgur.com/MCxQUCO.png[/img][/QUOTE]
only flac and mp3 y you do this
mp3 is so bad, and I have such a hate boner for it. I think compression is super neat, but I have to be really careful when exporting a finished master since its really easy to get mp3 distortion due to its headroom weirdness. if you wanted to, you could check to make sure the audio isn't hitting more than -0.3dbfs (if it is, consider normalizing it lower), and make sure to dither before doing bit reduction. There's more, like making sure to roll off the stereo field for 128kbps exports but thats a bit more work.
regardless, thats a pretty gui
[QUOTE=paindoc;49613459]only flac and mp3 y you do this
mp3 is so bad, and I have such a hate boner for it. I think compression is super neat, but I have to be really careful when exporting a finished master since its really easy to get mp3 distortion due to its headroom weirdness. if you wanted to, you could check to make sure the audio isn't hitting more than -0.3dbfs (if it is, consider normalizing it lower), and make sure to dither before doing bit reduction. There's more, like making sure to roll off the stereo field for 128kbps exports but thats a bit more work.
regardless, thats a pretty gui[/QUOTE]
How would you suggest transcoding the flacs to mp3 then? Also this is meant to add missing formats to torrents on what.cd where flac is available but mp3 is missing.
I wrote a basic API in Node.js which converts YouTube videos to MP3s using youtube-dl and ffmpeg and streams you the response as it's still being converted, then caches the conversion for later requests.
I designed it mainly for use inside GMod to play YouTube videos through the BASS library.
[IMG]http://i.imgur.com/o507mmQ.png[/IMG]
[IMG]http://i.imgur.com/1tdL1Ok.png[/IMG]
[IMG]http://i.imgur.com/nUzwumZ.png[/IMG]
Also has a 10 minute max video length as it's designed for music and I don't want requests bogging down the server too much!
[t]http://i.imgur.com/Tn8o0Ei.png[/t]
Integrated it with my web interface too.
Feel free to try it out: [URL]https://abyss.mattjeanes.com:8090/[/URL][get/play]?id=[youtubeid] - you'll need to call 'get' on a video to request conversion then you can call 'play' as soon it returns success.
Would anyone be interested in the source? I'm thinking of putting it on GitHub.
Very nice, worked perfectly for me. I might have to start using this over all those sketchy converter sites :P
[QUOTE=Mega1mpact;49613481]How would you suggest transcoding the flacs to mp3 then? Also this is meant to add missing formats to torrents on what.cd where flac is available but mp3 is missing.[/QUOTE]
I'd say 128kbps is the lowest I'd go in bitrate, for one. Still a huge space savings compared to FLAC. Second, is there an option to perform dithering? You're converting from FLAC down to mp3 so dithering helps to avoid truncation errors caused when doing this. The check for peak volume is to avoid clipping due to the same truncation errors.
So, the process could go Check Peaks, normalize down to -0.3dbfs if needed > Dither the audio > perform rest of compression. [URL="https://www.izotope.com/en/community/blog/tips-tutorials/2014/06/mastering-for-compressed-audio-formats/"]Here's some more info
[/URL]. Also, nice service. FLAC is nice but space-heavy. OGG is another good compression format to check out.
Also octrees are go, yay! I have also created an external function (or is it method? or what? I don't know muh words) that is able to generate a table for N-bodies, featuring the aforementioned 9 values but now it can also load from NASA data tables and someone elses data tables. Its a bit buggy, and not entirely what I'm testing atm. For now, I'm just testing a random cloud of points to see how the visualization works. It scales the size and varies color based on mass and distance from (0,0,0) for now, but the masses aren't large enough to really show. That and I'm not really shrinking the octrees down enough, I think.
[t]http://i.imgur.com/UQliG2Y.png[/t]
Next, I'm implementing [URL="http://arxiv.org/pdf/1506.01084v1.pdf"]this[/URL] algorithm for actually calculating the needed values. Octrees should really help optimize things though, and I've found numerous python examples of what I'm doing here. After the algorithm I'm going to attempt to get the data reader to format the data so it all works, then maybe try a new visualizer or controller or something.
[QUOTE=MattJeanes;49613581]I wrote a basic API in Node.js which converts YouTube videos to MP3s using youtube-dl and ffmpeg and streams you the response as it's still being converted, then caches the conversion for later requests.
I designed it mainly for use inside GMod to play YouTube videos through the BASS library.[/QUOTE]
Couldn't you just download the M4A dash audio and stream that, or better yet just forward the dash link to gmod to play?
Or does BASS not support m4a?
[QUOTE=Berkin;49613206]I just wanted to try the Unity GI system... how did this happen?
[img]http://i.imgur.com/I1KSrbS.png[/img][/QUOTE]
I Had this issue before with probuilder, Whenever a bake had finished and I entered play mode all the mesh for probuilder normals were either inverted or removed completely. A Restart fixed the issue but it keeps re-arising every now and again.
I wrote something like that in Lua w/ openresty, except for oggs #modern #future #hashtag #vorbis
[QUOTE=Map in a box;49614275]I wrote something like that in Lua w/ openresty, except for oggs #modern #future #hashtag #vorbis[/QUOTE]
Once you train your ear to hear differences in encoding and spend time mastering you can't unhear how bad mp3s can go. A lot of people like mp3 now for how it sounds, so it's become this eras vinyl.
[QUOTE=paindoc;49614318]Once you train your ear to hear differences in encoding and spend time mastering you can't unhear how bad mp3s can go. A lot of people like mp3 now for how it sounds, so it's become this eras vinyl.[/QUOTE]
Also my car speakers don't like mp3, you can hear the high frequency crunching if you just pinch your ear the right way and it starts to annoy me in sharp turns.
MP3 has definitely noticably aged.
Vorbis can output smaller files while still having better quality. Maybe one day it'll be the de facto lossy audio format.
Just spent ~3hrs configuring MS-DOS 6 to play nice in VirtualBox. Got cpu idling, EMM386, audio, graphics (1024x768, 256-color), and networking (in theory) working. Getting everything to fit in memory is a bitch, lots of DRIVERHIGH and LH, still end up with "Out of memory" half the time, even when running edit, let alone windows.
My eventual goal is to write Windows 3.1 software. We'll see how that goes.
Almost got Mosaic installed, but Win32s crapped out. I might have to settle for Win 3.11, but I can't find any installation media. :/
Also, it turns out that Creative still have drivers for SoundBlaster 16 on MS-Dos 6/Windows 3.1 on their support site. :D
yep i bet this algorithm isn't going to be a pain in the ass at all when this is just the block of constants I had to declare
[CODE]%{
IAS15 is an adaptive timestep integrator for solving problems involving
forces between gravitational bodies. Works best for non-conservative
forces. Slower than WHfast but doesn't break during close approach.
Uses adaptive stepping.
%}
%
hv=[0.0; 0.0562625605369221464656521910; 0.1802406917368923649875799428; 0.3526247171131696373739077702;
0.5471536263305553830014485577; 0.7342101772154105410531523211; 0.8853209468390957680903597629;
0.9775206135612875018911745004];
% Other constants
rrv=[0.0562625605369221464656522; 0.1802406917368923649875799; 0.1239781311999702185219278;
0.3526247171131696373739078; 0.2963621565762474909082556; 0.1723840253762772723863278;
0.5471536263305553830014486; 0.4908910657936332365357964; 0.3669129345936630180138686;
0.1945289092173857456275408; 0.7342101772154105410531523; 0.6779476166784883945875001;
0.5539694854785181760655724; 0.3815854601022409036792446; 0.1870565508848551580517038;
0.8853209468390957680903598; 0.8290583863021736216247076; 0.7050802551022034031027798;
0.5326962297259261307164520; 0.3381673205085403850889112; 0.1511107696236852270372074;
0.9775206135612875018911745; 0.9212580530243653554255223; 0.7972799218243951369035946;
0.6248958964481178645172667; 0.4303669872307321188897259; 0.2433104363458769608380222;
0.0921996667221917338008147];
cv=[-0.0562625605369221464656522; 0.0101408028300636299864818; -0.2365032522738145114532321;
-0.0035758977292516175949345; 0.0935376952594620658957485; -0.5891279693869841488271399;
0.0019565654099472210769006; -0.0547553868890686864408084; 0.4158812000823068616886219;
-1.1362815957175395318285885; -0.0014365302363708915610919; 0.0421585277212687082291130;
-0.3600995965020568162530901; 1.2501507118406910366792415; -1.8704917729329500728817408;
0.0012717903090268677658020; -0.0387603579159067708505249; 0.3609622434528459872559689;
-1.4668842084004269779203515; 2.9061362593084293206895457; -2.7558127197720458409721005];
dv=[0.0562625605369221464656522; 0.0031654757181708292499905; 0.2365032522738145114532321;
0.0001780977692217433881125; 0.0457929855060279188954539; 0.5891279693869841488271399;
0.0000100202365223291272096; 0.0084318571535257015445000; 0.2535340690545692665214616;
1.1362815957175395318285885; 0.0000005637641639318207610; 0.0015297840025004658189490;
0.0978342365324440053653648; 0.8752546646840910912297246; 1.8704917729329500728817408;
0.0000000317188154017613665; 0.0002762930909826476593130; 0.0360285539837364596003871;
0.5767330002770787313544596; 2.2485887607691598182153473; 2.7558127197720458409721005];
%}
% Integration weights for 8th order DE
w= [0.03125; 0.185358154802979278540728972807180754479812609; 0.304130620646785128975743291458180383736715043;
0.376517545389118556572129261157225608762708603; 0.391572167452493593082499533303669362149363727;
0.347014795634501068709955597003528601733139176; 0.249647901329864963257869294715235590174262844;
0.114508814744257199342353731044292225247093225];
[/CODE]
:dig:
You should probably have a look at various Particle-in-cell type simulations/optimizations if you haven't already. Because afaik those integration methods won't be applicable in any meaningful way if you go down that path.
Attending an A.I. course at the University of Aalto.
Did some A* implementation in C#
[code]
public class A_Pathfinder
{
public PriorityQueue<Node> pqueue = new PriorityQueue<Node>();
public Node FindPath(Tuple<int, int> root, Tuple<int, int> goal, Maze maze, out int ExpandedNodesAmount)
{
Node ExpandedNode = null;
Node GoalNode = null;
Dictionary<Tuple<int, int>, Node> exploredNodes = new Dictionary<Tuple<int, int>, Node>();
maze.Nodes.TryGetValue(root, out ExpandedNode);
maze.Nodes.TryGetValue(goal, out GoalNode);
ExpandedNode.Priority = 0;
List<Node> ExpandedNodes = new List<Node>();
pqueue.Enqueue(ExpandedNode);
//Preprocessing done
while (ExpandedNode != GoalNode)
{
pqueue.Dequeue();
int CurrentCost = ExpandedNode.Priority;
List<Node> neighbours = maze.GetNeighbours(ExpandedNode.Position);
foreach (var n in neighbours)
{
if (n.Walkable && !exploredNodes.ContainsKey(n.Position))
{
exploredNodes.Add(n.Position, n);
n.Priority = CurrentCost + 1 + n.Heuristic; //f(n) = g(n) + h(n) ... +1 because all adjacent steps cost 1.
n.LastNode = ExpandedNode;
pqueue.Enqueue(n);
}
}
ExpandedNodes.Add(ExpandedNode);
ExpandedNode = pqueue.Peek();
ExpandedNode.Priority = CurrentCost + 1;
}
ExpandedNodesAmount = ExpandedNodes.Count;
return GoalNode; //Return the goalnode and use the 'linked list' of LastNodes as a path.
}
}[/code]
It's pretty fast.
Kotaku did an article on my mod:
[url]http://kotaku.com/fans-are-turning-fallout-new-vegas-into-a-multiplayer-1755424654[/url]
I made a Reddit thread about it and it got a lot of attention too:
[url]https://www.reddit.com/r/Fallout/comments/42u947/new_vegas_multiplayer_help_us_test/[/url]
As someone who's never really had a lot of "success" (if you can call it that), I'm feeling warm and fuzzy reading some of these messages I'm getting from people. I sacrificed a lot putting work into this over the past 7 months and to finally see some response to my work makes me so happy.
Hello people. I had a question.
What do you think about DirectPlay for game networking?
Can it still be used in a reliable manner in current days?
I did try using it for my game. If it works, then it's decent.
But half of the time I just can't detect other games session to join/network with. What's up with that?
[QUOTE=paindoc;49609873]Not exactly, Verlet perfectly conserves angular momentum and generally conserves energies. Its best for usage when the only force on an object is due to a potential field, i.e gravity. Runge-Kutta develops exponential instabilities if the error bounds aren't set just right, and as far as I can tell Verlet is my best bet (instability develops linearly). I've found a number of modified guassian algorithms as well, including the one the AF uses (+NORAD), and found a highly improved verlet algorithm that is 3 orders of magnitude more accurate (but has 4 physics calls per iteration, so it is slower).
I think Verlet stuff, heavily modified, is what is used in plasmas too. Mostly due to the fact that magnetohydrodynamics is all about fluids being moved by potential fields (magnetic fields).
also.
FUCKING. FINALLY.
Currently using a stiff ODE but even that breaks less than halfway into my iteration process- I'm trying to convert to verlet now, but this at least has gotten my graphs working and I've now define a custom function that holds all the variables for whatever integrator I choose.[/QUOTE]
When I tried doing orbital simulations, i used RK4 with conservation of specific orbital energy, which does not change unless any other external force is introduced. Works okay in both scenarios: when you have an object affected only by the gravitational force, or you can introduce external forces as you increase the specific orbital energy.
where's the god damn d conetnt
[media]http://www.youtube.com/watch?v=Egv3dOxcz9E[/media]
final boss, turn planet
[QUOTE=hakimhakim;49619246]Hello people. I had a question.
What do you think about DirectPlay for game networking?
Can it still be used in a reliable manner in current days?
I did try using it for my game. If it works, then it's decent.
But half of the time I just can't detect other games session to join/network with. What's up with that?[/QUOTE]
[URL="https://facepunch.com/showthread.php?t=1446371"]You'll get more help here.[/URL] By the way, you don't want to use DirectPlay because it's been deprecated by Microsoft (that means it must be [I]really old[/I]). It was deprecated in favor of Games for Windows Live but that died, it's pretty tragic :v:
I've only done networking in high-level engines like Unity, so the only networking libraries I know of is RakNet. Just don't use DirectPlay, there is much better alternatives that are actively developed (and open-source).
[QUOTE=alexaz;49619249]When I tried doing orbital simulations, i used RK4 with conservation of specific orbital energy, which does not change unless any other external force is introduced. Works okay in both scenarios: when you have an object affected only by the gravitational force, or you can introduce external forces as you increase the specific orbital energy.[/QUOTE]
The IAS15 is much like an adaptive timestepping Runge-Kutta scheme, but tends to be much slower. It's 15th order, and is ludicrously accurate because of that.
The other algorithm I'm using uses some very clever transformations of the Hamiltionian allowing me to pretty much iterate based on the octree, so that I find the center of mass and position + velocity vector for a node in the tree than use this to solve the node lower in the hierarchy etc. This greatly speeds up computation time, since now I'm not calling my physics function for each body. The computation becomes O(logN) instead of O(N*Nbod) so that helps loads.
My main goal is to certainly be able to simulations of systems with only a few bodies N, but I'd like to be able to do huge systems too. So the plan is to create a switch case that changes the integrator I use - IAS15 for close bodies or nonconservative interactions, symplectic for conservative or group interactions. I think this schema was used to do tests of protoplanetary disks, actually.
My octree thingy is working but I'm trying to figure out the best way to index it - I want it to be sorted in the index array by distance from the origin. I'm wondering how this will affect computation time, and how id modify the indices that changed each time step without rebuilding the whole tree. I also need to add some more functions to it still.
[editline]27th January 2016[/editline]
I'm going to ask my astronomy professor what he thinks tomorrow too, he's done tons of simulations. Most of the stuff I'm doing exists, but I want to make my own just because I can. And it's a good way to learn about the computing I want to do
Iso pixelart shader thing. Scales so nicely
[IMG]http://i.imgur.com/ZLlczBC.gif[/IMG]
And here is the technique in action but with other stuff
[url]https://www.shadertoy.com/view/ld33DX[/url]
Sorry, you need to Log In to post a reply to this thread.
