oh no i cant link to my swag myspace profile anymore :(((( fuk u micro$h1t
[editline]12th January 2012[/editline]
[QUOTE=Darwin226;34182649]It's horrible.[/QUOTE]
I'm not a big fan of it, either.
[QUOTE=amcfaggot;34182940]oh no i cant link to my swag myspace profile anymore :(((( fuk u micro$h1t
[/QUOTE]
I don't know if that was a prod at me but I used to be able to share funny stuff I found on the interwebz and now I can't because everyone is too lazy to copy and paste :(
And I like MS.
[QUOTE=swift and shift;34182662]i haven't noticed a single thing Microsoft have done to ruin it[/QUOTE]
IE Control
[img]http://goo.gl/CdzX1[/img]
Plus the horrible memory leaks it has.
[img]http://i52.tinypic.com/fw5e9d.png[/img]
[img]http://dl.dropbox.com/u/29760911/ZScreen/2011-12/5n5ndndn7nKnnnqnQnnnenTnin.png[/img]
Didn't screenshot it but last night it was at 3GB of memory before crashing. Thought I already had a screenshot of its memory usage. I guess not.
Also some asshole designer for Skype decided to remove the volume control. So now you have to go into the settings to change it.
[img]http://goo.gl/KFgh6[/img]
Since Microsoft has taken over the only 1 thing they have done which I like is the automatic call reconnecting. But somehow they managed to even fuck that up. Now when I hangup calls it will sometimes try to automatically reconnect me.
Really Vent/TS3 are better at this point. How have they still not added a feature to allow you separately control each persons volume like TS3/Vent.
[QUOTE=esalaka;34181453]Probably the best case would be simply having a separate processing core for each and every ray. Two million cores might sound ludicrous but if there were that many, each core could be used to render the lighting caused by a single ray. Unless the rays interact with each other in manners that are unparallelizable. (I'm assuming this sort of raytracing is embarrasingly parallel.[/QUOTE]
I honestly think the 'future' of computing graphics is raytracing point cloud or voxel data in a hierarchical manner with dedicated hardware. I think Intel was researching this, and they definitely have the expertise and infrastructure to make it happen. They're probably in a better position than Nvidia even because CPU architecture is better suited for it than GPU architecture. (See below)
[QUOTE=Kamshak;34181518]isn't CUDA/OpenCL very well suited for these kinds of massively parallel operations? Now i'd really like to look into ray tracing and see if this can give a noticable performance boost[/QUOTE]
Graphics hardware has the advantage of being massively parallel, but graphics hardware sucks at branching (and raytracing involves lots of branching). IIRC the limitation (at least a few years ago) was that even though you could branch, every core in the GPU executes each instruction on each core synchronously so cores that branch ahead in code stall until all the other cores catch up. So while GPUs are certainly much faster today (just because they have [i]so many[/i] cores), CPU architecture would probably be a better starting point for designing dedicated raytracing hardware, since CPUs include stuff like branch prediction and cores operate independently.
[QUOTE=ROBO_DONUT;34183535][B]I honestly think the 'future' of computing graphics is raytracing point cloud or voxel data in a hierarchical manner with dedicated hardware. [/B] I think Intel was researching this, and they definitely have the expertise and infrastructure to make it happen. They're probably in a better position than Nvidia even because CPU architecture is better suited for it than GPU architecture. (See below)
Graphics hardware has the advantage of being massively parallel, but graphics hardware sucks at branching. IIRC the limitation (at least a few years ago) was that even though you could branch, every core in the GPU executes every instruction on each core in synchronously so cores that branch ahead in code stall until all the other cores catch up. So while GPUs are certainly much faster today, CPU architecture would probably be a better starting point for designing dedicated raytracing hardware, since CPUs include stuff like branch prediction and cores operate independently.[/QUOTE]
Wasn't that essentially what the "unlimited detail" thing was all about, or am I misunderstanding?
Speaking of which, have they disappeared again?
edit: uh, minus the dedicated hardware part, I think.
[vid]http://j.mp/zQRsxi[/vid]
guys the bible is a conspiracy, god created SHADOW after light
[QUOTE=Ziks;34178751]It's probably a good idea to decide that now. Let's have a semi-formal vote:
[img]http://www.facepunch.com/fp/ratings/box.png[/img] - Minecraft style first person, multiple players per colony controlling one character each
[img]http://www.facepunch.com/fp/ratings/cross.png[/img] - Isometric Transport Tycoon / Roller Coaster Tycoon style viewpoint, multiple players per colony controlling one character each
[img]http://www.facepunch.com/fp/ratings/rainbow.png[/img] - Isometric Transport Tycoon / Roller Coaster Tycoon style viewpoint, Dwarf Fortress style gameplay where you give orders to NPCs
[img]http://www.facepunch.com/fp/ratings/clock.png[/img] - Top down grid based, multiple players per colony controlling one character each
[img]http://www.facepunch.com/fp/ratings/funny2.png[/img] - Top down grid based, Dwarf Fortress style gameplay where you give orders to NPCs[/QUOTE]
It's your project, so feel free to tell me to fuck off :v:, but I'd like to point out that this is YOUR project. Designing a game by group vote is not only a bad way to make something worth playing, but it's also probably the main reason all those other group projects failed.
Projects need a decisive, creative leader. You can take advice and input, but in the end, it should be your decision.
Just my two cents.
[QUOTE=BlkDucky;34183562]Wasn't that essentially what the "unlimited detail" thing was all about, or am I misunderstanding?
Speaking of which, have they disappeared again?
edit: uh, minus the dedicated hardware part, I think.[/QUOTE]
Yeah, probably, but they spewed a lot of chaff in their videos, so who really knows what they've actually done :\
I mean they've certainly done something neat. Before I thought they were working with SVOs, because that's what I knew, but while I was researching hierarchical collision detection structures I found a lot of material about rendering polygon soup w/ BVHs. If you replace the polygons with points, you get something that sounds a lot like what they describe in their videos.
[QUOTE=Naelstrom;34183614][vid]http://j.mp/zQRsxi[/vid]
guys the bible is a conspiracy, god created SHADOW after light[/QUOTE]
What article did you use for this? If any.
[QUOTE=RyanDv3;34183636]It's your project, so feel free to tell me to fuck off :v:, but I'd like to point out that this is YOUR project. Designing a game by group vote is not only a bad way to make something worth playing, but it's also probably the main reason all those other group projects failed.
Projects need a decisive, creative leader. You can take advice and input, but in the end, it should be your decision.
Just my two cents.[/QUOTE]
I regretted the vote as soon as I saw almost no-one wanted an isometric game! It won't matter for a few more days, because the 3D stuff that's being done now can apply for any view point. We could even leave it up to the player which viewpoint they choose.
[img]http://anyhub.net/file/6WZK-octrees2.png[/img]
Octrees! That's enough for tonight, I'll get the thread up tomorrow morning so we can plan the project.
Thought I'd take a stab at Chip-8:
[IMG]http://i.imgur.com/lsAiY.png[/IMG]
It can also disassemble, although it's mostly guessing on where the code ends since Chip8 ROMs can just throw data wherever the fuck they want (right now it assumes it's reached EOC if it reaches a JP that jumps to its own address or if it finds an unrecognized opcode)
[IMG]http://i.imgur.com/ZUHMU.png[/IMG]
Next up: Assembling!
[QUOTE=NorthernGate;34183784]What article did you use for this? If any.[/QUOTE]
[url]http://www.gamedev.net/page/resources/_/technical/graphics-programming-and-theory/dynamic-2d-soft-shadows-r2032[/url]
I wouldn't mind explaining a thing or two if you add me on steam. The article is pretty bad at explaining how to implement it in standard opengl.
[QUOTE=Deco Da Man;34178844][img]http://www.facepunch.com/fp/ratings/winner.png[/img] - First-person "Omnipotent-being" (free roam) view where you can give orders to NPCs in a top-down manner, but you can also take control of individual characters.
(Lego Rock Raiders-esque... if you're familiar with that game)[/QUOTE]
I began programming in the hopes to one day make a game exactly like that.
That'll teach me to act instead of waiting next time :v:
[img]http://puu.sh/ddcT[/img]
That's supposed to be perlin noise..
:pwn:
[QUOTE=calzoneman;34183895]Thought I'd take a stab at Chip-8:
[IMG]http://i.imgur.com/lsAiY.png[/IMG]
It can also disassemble, although it's mostly guessing on where the code ends since Chip8 ROMs can just throw data wherever the fuck they want (right now it assumes it's reached EOC if it reaches a JP that jumps to its own address or if it finds an unrecognized opcode)
[IMG]http://i.imgur.com/ZUHMU.png[/IMG]
Next up: Assembling![/QUOTE]
Coincidence, or did I inspire you? :D
[QUOTE=Chris220;34184103]Coincidence, or did I inspire you? :D[/QUOTE]
Both, in a way. I was thinking it might be fun to make an ASM interpreter and then I saw your post with that fantastic documentation and decided I'd hop on the bandwagon.
[QUOTE=Chris220;34184103]Coincidence, or did I inspire you? :D[/QUOTE]
I do have a question actually, do you happen to know the typical CHIP-8 execution speed? (or how many instructions I should process each second?)
[QUOTE=Naelstrom;34183924][url]http://www.gamedev.net/page/resources/_/technical/graphics-programming-and-theory/dynamic-2d-soft-shadows-r2032[/url]
I wouldn't mind explaining a thing or two if you add me on steam. The article is pretty bad at explaining how to implement it in standard opengl.[/QUOTE]
Thanks, I'll try my best at implementing it, but if I have any trouble I'll give you a shout ;D
[QUOTE=synthiac;34184848]why do people leave audio in their videos? it's not worth the extra bandwidth to hear some cruddy prog house tune.[/QUOTE]
Because it takes too much effort to press XF86AudioPause to care for you
[img]http://puu.sh/ddSH[/img]
Perlin noise generation and collisions! (thanks boomer678 for collision help)
[QUOTE=synthiac;34184848]why do people leave audio in their videos? it's not worth the extra bandwidth to hear some cruddy prog house tune.[/QUOTE]
Because most people don't care about whether or not synthiac has a good browsing experience.
[QUOTE=calzoneman;34185066]I do have a question actually, do you happen to know the typical CHIP-8 execution speed? (or how many instructions I should process each second?)[/QUOTE]
CHIP-8 itself doesn't actually define a clock speed, so you should use whatever feels right for you! 1 MHz seems to be a fairly common default speed, with options to raise or lower that to the user's preferences.
On the topic of CHIP-8, I added some support for a bunch more opcodes, most notably keyboard input. I was striving towards getting KALEID working (A kaleidoscope program, by the looks of it)
[t]http://i.imgur.com/2I7q9.png[/t]
It looks a lot nicer in motion, I'll record a video some time to show you all what it does.
After I had that working, I decided to have a quick look at some of the other ROMs to see if any would work with my currently supported opcode set, and look what I found:
[t]http://i.imgur.com/536et.png[/t]
That's right, it's Connect 4!
This is a really fun project, I'm enjoying myself immensely getting this stuff to work!
[QUOTE=Chris220;34186188]CHIP-8 itself doesn't actually define a clock speed, so you should use whatever feels right for you! 1 MHz seems to be a fairly common default speed, with options to raise or lower that to the user's preferences.
On the topic of CHIP-8, I added some support for a bunch more opcodes, most notably keyboard input. I was striving towards getting KALEID working (A kaleidoscope program, by the looks of it)
[t]http://i.imgur.com/2I7q9.png[/t]
It looks a lot nicer in motion, I'll record a video some time to show you all what it does.
After I had that working, I decided to have a quick look at some of the other ROMs to see if any would work with my currently supported opcode set, and look what I found:
[t]http://i.imgur.com/536et.png[/t]
That's right, it's Connect 4!
This is a really fun project, I'm enjoying myself immensely getting this stuff to work![/QUOTE]
1 MHz I assume is the clock speed of the chip, what value are you using for instructions per second?
[QUOTE=calzoneman;34186245]1 MHz I assume is the clock speed of the chip, what value are you using for instructions per second?[/QUOTE]
I'm not actually limiting mine at the moment, I'm having too much fun implementing the opcodes! :P
My first time with EventListener in Java.
[img]http://i.imgur.com/ELght.png[/img]
[QUOTE=benji2015;34186159][img]http://puu.sh/ddSH[/img]
Perlin noise generation and collisions! (thanks boomer678 for collision help)[/QUOTE]
It's like a 3D version of my LD21 game. :sniff:
[QUOTE=dvondrake;34186096]Theory-wise, I think that's pretty much the same as what I'm doing. I gave your code a try, but it gave me the same result.
Unfortunately I'm not using light volumes, I had an issue where if I made the radius of a light volume too big it would just stop lighting, plus I like the freedom lighting as a full-screen quad would give me. (If it worked.) Though maybe I have the completely wrong idea. How have light volumes been working out for everyone else? Also, wouldn't I have to do the same world->view space transformation for directional lights since they're done on a full-screen quad? Really my issue is that since I'm resetting the GL projection and modelview matrices to render the full-screen quad, I can't use them in the shader to rotate the light (as you would with light volumes).
[img]http://i.imgur.com/LXu2b.png[/img]
[url=http://www.facepunch.com/threads/1155127]Update on my problem, if anyone feels so kind[/url][/QUOTE]
The light volume technique is really worth it for deferred rendering since you only have to render the light withing a specific range (radius) meaning that you don't have to render each pixels like a fullscreen quad. Basically you can either use a sphere/cube for point light, a cone/pyramid for spot light and finally you can use a fullscreen quad for directional light, SSAO and final composing. If you're having hard time with this concept I would recommend building and understanding this demo first:
[URL="http://encelo.netsons.org/_download/glsl_deferred.tar.gz"]http://encelo.netsons.org/_download/glsl_deferred.tar.gz[/URL]
[QUOTE=HiredK;34187440]The light volume technique is really worth it for deferred rendering since you only have to render the light withing a specific range (radius) meaning that you don't have to render each pixels like a fullscreen quad. Basically you can either use a sphere/cube for point light, a cone/pyramid for spot light and finally you can use a fullscreen quad for directional light, SSAO and final composing. If you're having hard time with this concept I would recommend building and understanding this demo first:
[URL="http://encelo.netsons.org/_download/glsl_deferred.tar.gz"]http://encelo.netsons.org/_download/glsl_deferred.tar.gz[/URL][/QUOTE]
Don't forget about tiled deferred [URL="http://publications.dice.se/attachments/GDC11_DX11inBF3_Public.pdf"][1][/URL] [URL="http://bps10.idav.ucdavis.edu/talks/12-lauritzen_DeferredShading_BPS_SIGGRAPH2010.pdf"][2][/URL]
Got my assembler/disassembler working well
[IMG]http://i.imgur.com/4JX4c.png[/IMG]
(left = original source, right = assembled then disassembled)
[media]http://www.youtube.com/watch?v=N8plefUTbrM&feature=youtu.be[/media]
i finished my 3d cam node and implemented directinput8 in a "kopinput" class
so far its working really well so i think ill move on to obj loader
Sorry, you need to Log In to post a reply to this thread.
