Unlimited Detail

yeop .. (didn't watch the vid, so maybe there's something there, but from the comments i have developed enough of an alternatively rendered picture to get the detail on the situation.)

there's quite a few interviews in the past few years where Tim Sweeney (the uber nerd mega genius architect behind all the unreal engines) discusses that in the very near future he can see games being rendered in software and not graphics API's like open GL or Direct X and thinks GPUs and CPU's will wage in an evolutionary war of what architecture is more necessary or optimized for handling all mediums in processing.

  On 3/11/2010 at 9:46 PM, æsterlingman said:

i like how that guy is attempting to demolish the graphics card industry. he should hire a bodyguard.

graphics cards are already pretty flexible - nvidia even released source code for voxel raytracing. Also, I don't think there's much interest in rendering animated characters with voxels or points instead of polygons

  On 3/12/2010 at 2:26 AM, Ego said:

Voxels are a total game-changer. You can't just put a shader on something, it needs to be defined in the voxel data.

wouldn't shading just work like a deferred renderer?

( for example: http://http.developer.nvidia.com/GPUGems2/gpugems2_chapter09.html )

  Quote

Look at the most realistic renders today. All polygons.

renders can have as much geometric complexity as you want; point/voxel representation is a way for game artists to not have that kind of constraint. Of course displacement mapping is another approach

maybe there's potential in this but those screens look like ass

he would say something like:

anyone who believes this is an idiot blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah I've worked in this industry for several years and blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah as an expert i think that blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah when i designed an entirely new rendering system for pixar blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah.

i am sure jswift has already developed a real-time photoreal rendering engine able to be run on an old pentium with 64mb ram, but in the end thought it as a waste of time when he could be making a cure for cancer.

Hey, I had the idea of something like that search algorhytm a while ago... but I thought it was impossible and would have been done long ago if it wasn't, so I forgot about it. It sounds and looks promising (despite the samples being quite poor designwise but hey...), but... if these point cloud graphics have that many individual points, what would happen if there was even simple animations like - in an fps - arms and a weapon with just as many points moving around? despite being not forced to render them, it would still need to constantly calculate each of the points with their new coordinates wouldn't it, even if they were grouped or summed up to a single object?

not an expert ...

  On 3/12/2010 at 7:46 PM, Terpentintollwut said:

Hey, I had the idea of something like that search algorhytm a while ago... but I thought it was impossible and would have been done long ago if it wasn't, so I forgot about it. It sounds and looks promising (despite the samples being quite poor designwise but hey...), but... if these point cloud graphics have that many individual points, what would happen if there was even simple animations like - in an fps - arms and a weapon with just as many points moving around? despite being not forced to render them, it would still need to constantly calculate each of the points with their new coordinates wouldn't it, even if they were grouped or summed up to a single object?

 

not an expert ...

not only that, i'm pretty sure the data structure that enables that awesome fast search algorithm is highly optimized for static geometry constellations. like, it probably takes ages to re-arrange those search trees or what have you, once some positions have changed.

I am going to marry the man who narrated that.

i think its pretty BS, until he shows something animated and properly textured

well that's something we will probably never see, this tech is highly unlikely to work well with even sub-molasses speed animation and obviously its not gonna make textures any easier on your ram.

then what the F is the point

  On 3/13/2010 at 10:01 PM, Blanket Fort Collapse said:

this tech is highly unlikely to work well with even sub-molasses speed animation

Ignorant english man curios as to what the heck molasses is///

  On 3/13/2010 at 11:36 PM, mcbpete said:

  On 3/13/2010 at 10:01 PM, Blanket Fort Collapse said:

this tech is highly unlikely to work well with even sub-molasses speed animation

Ignorant english man curios as to what the heck molasses is///

why dont they just make the fucking polygons smaller so we can't detect them.

duh.

i haven't watched the vid (China, sigh) but what they really need to do is develop some sort of artificially intelligent dynamic LOD-ing.

i am a computer scientist, and i can tell you that this guy is full of shit.

  On 3/14/2010 at 6:42 AM, vodor said:

i am a computer scientist, and i can tell you that this guy is full of shit.

oh yeah well im a psychiatrist, and i can tell you that you're full of shit.

well i am a shiteyeitrest and i can tell you that your shit is full of eyes.

  On 3/14/2010 at 6:43 AM, 42Orange said:

  On 3/14/2010 at 6:42 AM, vodor said:

i am a computer scientist, and i can tell you that this guy is full of shit.

 

oh yeah well im a psychiatrist, and i can tell you that you're full of shit.

oh yeah well im a proctologist, and i can tell you that you're both full of shit.

here at the st. petersburg institute of poopology we study turds, poop, feces, and shit in all its forms from liquid to 'afterwards, goatse' (AG).

you're all full of shit. especially that snake-oil salesman in the video.

I was never impressed with polygons (in video games), ever since Star Fox came out I thought they looked worse than 2D IMO, of coarse today they're close but it looks like this will seal the deal

so that pretty much blew my mind. the only question is though; lighting, shadows, how would they work? the objects in those videos look self-illuminated and pre-shadow mapped... if ray-tracing is too slow and obsolete... how will they raytrace?

if you had just invented a technology that shifts the entire paradigm of 3D graphics and could destroy ATI/Nvidia, would the first thing you do be make a smug youtube video with a mozart midi file playing in the background?