Tuesday, October 1, 2013

Soon...

 

20 comments:

Unknown said...

Can't wait! The teaser has me all hot and bothered... :P

BuD said...

Oh god, can't wait as well. I expect to much from this project :3

Probably you answered this question before, still, can I ask what GPGPU API use Brigade for parallel processing?

Next gen gonna be awesome! (assuming next gen could support some kind of GPGPU API)

Retina said...

this car looks like it has more polys than a new Unrealengine4 game in total. haha

Retina said...

by the way , is that dark chrome?

Kevin said...

Sweet Baby Jesus!
I check this blog daily (sometimes twice) and there hasn't been a post in more than a month.
You really know how to tease us! :D

Seriously I have so many questions. Will there be an updated kernel, will Brigade have Cloud integration like it was demonstrated in octane, will there be a techdemo for us to try out, will there be a new video (that will hopefully be posted on vimeo this time, for less agressiv compression), will there be news about a potential mini game?

ARG I can not wait! :D

Retina said...

that is freakin crazy. i am watching this picture for nearly an hour now. lol
all these details and physical correctness of shading and materials. Incredible.

Anonymous said...

Still very noisy..is this in bridage3?
I'd love to see that old sci-fi interior on the newer builds.

Anonymous said...

Bring it on! Just few weeks ago i was thinking about brigade and some nice hipoly model of a car and here it is. Cant wait.

Unknown said...
This comment has been removed by the author.
Unknown said...

Brigade 3 powns Chuck Norris. But seriously, Brigade was the inspiration for my research direction! Thanks Sam, Jacco and ALL the other awesome people!

Anonymous said...

Hey, Monte Carlo, what have you done?
The noise in my image is a real problem, man!

https://www.youtube.com/watch?v=sTUhAfwzsGg

AlMantiq said...
This comment has been removed by the author.
Sam Lapere said...

BuD: Brigade supports both CUDA and OpenCL to get the most out of Nvidia and AMD GPUs (running OpenCL on Nvidia is not a good idea).

colocolo: indeed, the car has 4.3 million triangles and it runs buttersmooth :) we'll give more details about the new material system soon

Kevin: no tech demo yet, but we'll post a new video soon (with a really cool scene that we haven't shown yet)

anonymous: monte Carlo noise is going to be the new lens flare, so you better embrace it :) I personally prefer a noisy image over a perfectly clean image which looks too CG like, and it's proof that every pixel is calculated in real-time without needing any baking.

Mauro: glad this thing inspired you to do graphics research. Voor mij is dit ook reteboeiend.

Anonymous: cool paper, even cooler video, wavelets are efficient but have their shortcomings and we don't like to compromise on visual quality.

Fady: thanks, it should be at least sooner than half-life 3






Kamus said...

Looks good... can't wait for the day where nVidia and AMD both decide to make their products strictly for compute.
I'm wondering how much performance would increase on brigade if they did that.

Also, what are you thoughts on AMD's mantle, and would brigade be able to take advantage of it?

Anonymous said...

apropos noise.... Have you guys watched closely at Crysis 3. there is some black noise in the pixels if you watch carefully.
this hand.... and the background....
these must be robots! :)

Anonymous said...

you should try utilizing CogniMem CM1K in path-tracing experiments

Sam Lapere said...

Manuel: there are a few ways to massively increase the efficiency of path tracing on the GPU ;)

Mantle looks promising, and I applaud every effort to get rid of DirectX as soon as possible. I hope AMD will not treat Mantle the same way as they did with OpenCL but put some real effort in it.

Anonymous: Sounds like Crysis 3 is trying to fake the path tracing look then :)

Anonymous said...

The blog description says real-time but these are not real-time, are they?

ray_vrayz said...

wow..
very nice raytracing with cuda
how programming script that ??
please

c++ or c#

Anonymous said...

I'm really impressed with your demos. Can Brigade handle huge, dynamic, open worlds like GTA5? Not just in terms of processing but memory foot print. The Caustic cards had 8GB, but you're using GPUs with 4 or less.

From some of the videos I've seen, memory doesn't seem to be an issue with Brigade, but how? Doesn't ray tracing require huge amounts of memory?

Speaking of Caustic, do you think you'd get real time frame rates with no noise with any degree of scene complexity if somebody built huge, fixed function processor that handled most of what your engine is doing?