Search next time. This was in the Programming forum.
[QUOTE=Agent766;21708807]Search next time. This was in the Programming forum.[/QUOTE]
Search is still disabled.
[url]http://www.facepunch.com/search.php[/url]
It's icon just isn't there.
[QUOTE=Agent766;21708989][url]http://www.facepunch.com/search.php[/url]
It's icon just isn't there.[/QUOTE]
I wish garry would stop hiding everything :argh:
I have NEVER seen anything that was software based being faster than work being done in hardware.
Also, point rendering is bullshit as it takes mad amounts of processing power to work.
They don't have a real artist working on the project, so guess what happens if they hire 2-4 artists to model, skin, etc.
Also, videos:
[media]http://www.youtube.com/watch?v=Q-ATtrImCx4[/media]
[media]http://www.youtube.com/watch?v=THaam5mwIR8[/media]
[media]http://www.youtube.com/watch?v=l3Sw3dnu8q8[/media]
[QUOTE=IMA SHAARK;21704238].
[editline]04:30PM[/editline]
That's why it isn't used.[/QUOTE]
I'm pretty sure they make models the same way they do in ZBrush.
[QUOTE=pentium;21709130]
Also, point rendering is bullshit as it takes mad amounts of processing power to work.[/QUOTE]
You're reading what I am right?
They should put their money where their mouth is then. Where can we try these demonstrations for ourselves?
I can't image how that works with animated models. Either their great searching algorithm is extremely slow (it's not from what they showed us) or it has to presort all the points which could take extremely long. That would make animations impossible.
Here's a bit more informative article on this technology, the OP didn't really tell much about it:
[url]http://www.newscientist.com/article/mg20627546.400-pointillist-style-could-bring-lifelike-graphics-to-pcs.html[/url]
I somehow doubt that this could work well. First, it takes a huge amount of data to store so many points, you need to store the position and color of every single one of those points. With simple objects consisting of millions of points, a basic scene could take gigabytes of storage space. And you probably can't compress this data very much, unless you use the same objects multiple times. But why would you want to do that, when extra detail won't affect rendering speed?
Then there's shading. All we have so far are simple points that always stay the same color. To properly shade them, you would also need to store a normal vector for each point, and possibly some other data, depending on what effects you want to achieve. This increases the already huge amount of data even higher. The shadows could be tricky to do as well.(you can see that in the videos)
And the biggest problem: aliasing. If you use a sample from only one point for each pixel, there's going to be horrible aliasing, but not just on the edges of the objects, it'll be everywhere in between as well. If you don't know what i'm talking about, try upscaling/downscaling an image with nearest-neighbour interpolation (aka no interoplation). Taking multiple samples isn't really an option, since the performance is already quite low. Some kind of a 3d analogy to mipmapping is pretty much a must here to make objects recognizable at a distance. Having a million points covering an area of a single pixel is a horrible idea, since only one of them will be picked to represent all of them. But then again, these "mipmaps" would increase the storage requirements even further.
[QUOTE=Robber;21714824]I can't image how that works with animated models. Either their great searching algorithm is extremely slow (it's not from what they showed us) or it has to presort all the points which could take extremely long. That would make animations impossible.[/QUOTE]
[QUOTE=Jimmg;21700441]
And to all the naysayers about animation
[media]http://www.youtube.com/watch?v=cF8A4bsfKH8[/media]
[/QUOTE]
All the people complaining about modelling, they said in their video you can model a polygon then collect the point cloud data from it.
[QUOTE=BAZ;21717442]All the people complaining about modelling, they said in their video you can model a polygon then collect the point cloud data from it.[/QUOTE]
Then what's the point? The models will look exactly the same.
[QUOTE=Roast Beast;21718314]Then what's the point? The models will look exactly the same.[/QUOTE]
Never heard of subd?
[QUOTE=johanz;21718356]Never heard of subd?[/QUOTE]
Doubt it yo.
I liked the idea of 3d scanning, you could have an exact replica of a sculpture there in your scene without the worry of polygons messing it up
Why is everyone so pessimistic? It isn't fake. Be excited for the possibility of real looking games!
[QUOTE=Kabstract;21723872]Why is everyone so pessimistic? It isn't fake. Be excited for the possibility of real looking games![/QUOTE]
They are probably butthurt that they might not be able to brag about their 1337 computers and graphics cards.
I doubt it's true, you can't have unlimited detail and run it at reasonable fps.
How will lights or physics react?
[QUOTE=johanz;21723964]I doubt it's true, you can't have unlimited detail and run it at reasonable fps.
How will lights or physics react?[/QUOTE]
oh man. computing lighting for every vertex on the screen makes my stomach churn.
[QUOTE=pebkac;21714981]First, it takes a huge amount of data to store so many points, you need to store the position and color of every single one of those points. [b]With simple objects consisting of millions of points, a basic scene could take gigabytes of storage space.[/b] And you probably can't compress this data very much, unless you use the same objects multiple times[/QUOTE]
Soon we will have games that are 1TB in size and we have to buy them on SSD drives. USB will transfer 4TB a second wirelessly.
And no use for screens, we will all have glasses that are interactive with our world and our brains will control the glasses.
Fuck yeah, future.
[QUOTE=Unreliable;21728229]Soon we will have games that are 1TB in size and we have to buy them on SSD drives. USB will transfer 4TB a second wirelessly.
And no use for screens, we will all have glasses that are interactive with our world and our brains will control the glasses.
Fuck yeah, future.[/QUOTE]
But we still don't have our flying cars.
[QUOTE=surfur;21728589]But we still don't have our flying cars.[/QUOTE]
I dont want to be hit by a drunk flying car 300 meters up in the air. I can live driving on the ground for now.
[QUOTE=Unreliable;21728229]Soon we will have games that are 1TB in size and we have to buy them on SSD drives. USB will transfer 4TB a second wirelessly.
And no use for screens, we will all have glasses that are interactive with our world and our brains will control the glasses.
Fuck yeah, future.[/QUOTE]
Fuck yeah imaginative future no one cares about.
Think today, not tomorrow.
[QUOTE=Xyrec;21713529]They don't have a real artist working on the project, so guess what happens if they hire 2-4 artists to model, skin, etc.
Also, videos:
[media]http://www.youtube.com/watch?v=Q-ATtrImCx4[/media][/QUOTE]
Why does he sound like he's sitting alone in a gym?
[QUOTE=Unreliable;21728660]I dont want to be hit by a drunk flying car 300 meters up in the air. I can live driving on the ground for now.[/QUOTE]
That's a good point.
[QUOTE=Karmah;21703168]It isn't taxing, it's only renders what you can see, and cuts everything else out.[/QUOTE]
Yes because this is an entirely new concept and isn't used in practically every video games ever created. Ever tried mapping for source? Heard of "visleafs"?
[editline]11:03PM[/editline]
I love how he references how Google's searching algorithm is instant, but if you actually do a Google search it displays how long it took for it to search. Now this is on the order of milliseconds, but that's a massive overhead if you're working on large 3D scenes that need to be rendered at over 40 fps, and not just searching for a single string in a database like what you do with Google.
[editline]11:05PM[/editline]
[quote]Dell says he has got around this by finding a smart way to compress the points to cram a huge number into the RAM. However, until he has been awarded the patents covering the technology, he says he's not prepared to reveal more.[/quote]
Bullshit; there's a limit as to how much you can compress data.
[editline]11:06PM[/editline]
And the time spent uncompressing it would take a long time too, making another massive overhead.
[QUOTE=DainBramageStudios;21761286]
I love how he references how Google's searching algorithm is instant, but if you actually do a Google search it displays how long it took for it to search. Now this is on the order of milliseconds, but that's a massive overhead if you're working on large 3D scenes that need to be rendered at over 40 fps, and not just searching for a single string in a database like what you do with Google.
[/QUOTE]
Not to mention the amount of processing power google has
I guarantee you guys that in the last thread about "unlimited detail" it was confirmed to be fake.
I believe this is real :saddowns:
Sorry, you need to Log In to post a reply to this thread.