In the new video, when he upped the sampling to 64 (8²), it was a lot more impressive. They should’ve done that from the start. It doesnt stutter ‘that’ much considering the amount of samples and it does indeed seem to be more efficient at tracing more rays.
If this is just the start I can imagine its only going to get better.
That Raytrace depth of field is really impressive.
I’d like to see some highres stills, I hope this card can render out really high res stuff.
The card just calculates, if your scene would break due large resolution in max for example, it would be due restrictions to max and memory, OS etc I belive.
So the Caustics card doesn’t load up the scene then ?
These days I hardly run into problems rendering large datasets, proxies with 64bit Mental Ray.
It quasi loads the scene. It loads the geometry for the ray intersections (they claim 500 million vertices (about .75 to a billion polys). That’s the only memory limitation. But who has billion poly scenes and if so I’m sure they could make an virtual planet version or smart cacheing.
Another great advantage of an accellerated workflow is that you spend a lot less time optimizing and more time rendering. I probably spend 50% of my time making it render fast enough to get out the door in time. I just saved not only all my render time but a ton of setup time.
Yep not much chance of me having scenes of a billion polys
I wonder if any of the unbiased renders will be able to benefit from this card ?
It will be interesting to see where this Caustic card goes and compared to Larabee where all software renderers will benefit.
yeah i’ve heared some good things from ppl testing larrabee, but the nda forbids them to tell more
Yummy…real-time shader evaluation as you code it. See the latest video here.
Again, this only adds to the pile of interactive enhancements, another director/production unit gab has been cut away.
any new news?
you can check out the twitter page, they put stuff there.
For anyone that doesnt know it, Caustics James McCombe and brazil coder Cuneyt Ozdas made a presentation at Siggraph 2009 some hours ago. It was a realy solid presentation I think.
They showed a massivly sped up Brazil in max 2010 in RT environment, the DOF had almoste no impact on the Frame rate.
you should check out the presentation they do tomorrow over at http://area.autodesk.com/siggraph2009
Check your local time and the scheldule on the Day 2 tab.
( I did )
This looks interesting. If they can make DOF so fast, that probably means they can do all glossy effects as fast. If the development will continue successfully, I bet in a few yers we will see a much more robust version of Caustics. This must keep the competition though, in the end we artists win. They claimed they will have s 200x solution next year? :surprised
Check it here if you missed the last set that is, or want to see it again
ya, the developer card runs on the FPGA at a very low frequency. the consumer model will be 12 times or around faster than what they preview in the video.
This is simply jaw-dropping, definitely the most impressive stuff I’ve seen in the whole my career regarding 3d hardware.
I think those guys will become rich very fast.
The price is very reasonable (under 2500$).
yeah man…im so excited. I knew this was happening and it does happen. What a great great piece of tech we can look forward too.
I think the people that still thinks Brazil is slow has to pitch up on the news
did someone say the fastest around? ow…thought so
So true… it will be in a blink of an eye. Lighting will be soo much easy and pleasant. :drool:
Just curious, will we need Renderman still with all its clever hacks?
There is always a market if there is sales, but for sure those who wont or dont step on the wagon will naturally compromise them self as alternative solutions are much more flexible and has interactivity on the level you can get with the CausticTwo.
Also dont forget you can stack multiple cards pr machine and they scale in linear stenght.
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.