PDA

View Full Version : Nvidia Flex



seghier
09-08-2015, 01:53 PM
interresting : Nvidia Flex
Key Features: https://developer.nvidia.com/flex
Artist-focused tools to ensure turnkey solutions
Unified solver for effects
Rigid/deformable bodies
Phase transition
Particles
Fluids
Cloth
Rope
Adhesion
Gases

https://www.youtube.com/watch?v=1o0Nuq71gI4

MSherak
09-08-2015, 02:33 PM
interresting : Nvidia Flex
Key Features: https://developer.nvidia.com/flex
Artist-focused tools to ensure turnkey solutions
Unified solver for effects
Rigid/deformable bodies
Phase transition
Particles
Fluids
Cloth
Rope
Adhesion
Gases

https://www.youtube.com/watch?v=1o0Nuq71gI4

And since it's CUDA based it pretty much useless. Made for real-time more like games than to be applied to a 3D rendering package.

seghier
09-08-2015, 02:37 PM
a plugin for 3dsmax based on nvidia flex already created

creacon
09-08-2015, 03:48 PM
And that's an assumption based on what exactly?


And since it's CUDA based it pretty much useless. Made for real-time more like games than to be applied to a 3D rendering package.

erikals
09-08-2015, 04:02 PM
Looks neat! http://forums.cgsociety.org/images/smilies/arteest.gif

Dodgy
09-08-2015, 09:41 PM
I think it looks fantastic. All those elements interacting, gfx card accelerated, what's not to like?

creacon
09-08-2015, 11:52 PM
The demo is part of the sdk download, you can try it out for yourself if you are a registered developer.

creacon


I think it looks fantastic. All those elements interacting, gfx card accelerated, what's not to like?

lightscape
09-09-2015, 04:02 AM
And since it's CUDA based it pretty much useless. Made for real-time more like games than to be applied to a 3D rendering package.

Useless for you. But useful for others. I could definitely use that liquid fx on some beer commercials that come along once in a while.
That would mean I wouldn't have to use blender and stay with lightwave...if it had liquids.

Dodgy
09-09-2015, 05:18 AM
I second that, liquids in LightWave are something I really want. I look forward to seeing Creacon's work in that area.

prometheus
09-09-2015, 05:25 AM
And since it's CUDA based it pretty much useless. Made for real-time more like games than to be applied to a 3D rendering package.

strange assumption, especially since a vfx studio relied a lot on cuda cards for the last airbender, and if you look at what turbulence does with cuda enabled.

creacon
09-11-2015, 01:54 AM
I already have hooked up the flex liquids (just emitter and groundplane) and the simulation looks different than PhysX and it's easier to get thick fluids. for high particle counts it's slower than PhysX. But it looks promising.

creacon

Marander
09-11-2015, 07:18 AM
strange assumption, especially since a vfx studio relied a lot on cuda cards for the last airbender, and if you look at what turbulence does with cuda enabled.

...exactly, and Resolve, 3D-Coat, AutoCAD and hundreds of other applications as well.

Flex looks amazing, this engine in LW combined with volumetric rendering in Octane 3... that would be great.

Dodgy
09-11-2015, 07:18 AM
I already have hooked up the flex liquids (just emitter and groundplane) and the simulation looks different than PhysX and it's easier to get thick fluids. for high particle counts it's slower than PhysX. But it looks promising.

creacon

Cool stuff!

tyrot
09-11-2015, 10:12 AM
eagerly waiting

MSherak
09-11-2015, 10:42 AM
a plugin for 3dsmax based on nvidia flex already created

And that's an assumption based on what exactly?

I think it looks fantastic. All those elements interacting, gfx card accelerated, what's not to like?

strange assumption, especially since a vfx studio relied a lot on cuda cards for the last airbender, and if you look at what turbulence does with cuda enabled.

Nvidia. Nvidia and the love of themselves. NVidia only of course. Again Nvidia only

It's not the tech per say but how it is delivered. They could make this run on any GPU, yet ends up CUDA only. This is not a CUDA only functional math problem and all forms of computation (GPU, CPU, DSP, etc.) should be able to be utilized for the consumer. I get it, that's all they make so sell more for the shareholders and turn all video cards into NVidia. Kinda silly because in the future you will see more of the hybrid CPU/GPU processors and stuff like this will never run on it in it's current state. So Nvidia needs either to jump into the CPU market or changes the way this CUDA only stuff goes. This is my opinion alone and why I think it's useless. Anyway, enjoy the tech.

erikals
09-11-2015, 11:44 AM
nah... Intel used to love themselves too...

not saying i'm a fan of the Nvidia strategy, but certainly it's bearable, unlike two other companies who's name start on the letter "A"

squarewulf
09-11-2015, 11:55 AM
And since it's CUDA based it pretty much useless. Made for real-time more like games than to be applied to a 3D rendering package.


CUDA is awesome. Octane is CUDA based and i use it religiously. I don't understand how this is a bad thing?

spherical
09-11-2015, 12:00 PM
Some people prefer other video cards. CUDA is proprietary and unique to nVidia. If technologies are developed that run on it, those running other cards can't take advantage of them. It's kind of a Windows/Mac/*NIX thing.

ActionBob
09-11-2015, 03:36 PM
Not to step on any toes, but it seems to me that a company who makes something cool that runs on their proprietary hardware / libraries has every right to NOT port it to other hardware. Afterall, it is part of their marketing strategy to stay relevent.

Seems that users of other hardware should be complaining to their respective hardware manufacturers to implement something similar. We all make choices and sometimes you make the wrong choice because you are allied to a certain company or brand.... For years, I always chose hardware based on what I could afford, which had the best support with regards to Lightwave (lightwave was the deciding factor even though I was / am a big gamer). The choice was always easy for me - an Intel based machine (and believe me, I was a resistant and hard core Amiga user / evagelist back in the day).

I did use ATI cards for a brief time, but couldn't stand the software (even though I liked the hardware better). Now, I see the writing on the wall - nVidia has good stuff and much better marketing - you go with what will suit you best and what brings you the most benefit for your situation. As I am a tinkerer and like building and custominzing my machines, I am a PC/nVidia guy. You get more bang for your buck no matter how you spin it - because you have the option to choose.

Choose which best works for you. If you see that somebody else has something you like, get it or don't. But thinking that everything should be on a even playing field and faulting a company for utilizing a tech / marketing edge to their favor is evil seems pretty wrong to me.

Ernest
09-11-2015, 05:47 PM
Yes, it's their right to keep it proprietary. What people usually hate about that is how it affects gaming. Since there is a large group of potential customers who do not have access to those features, gaming companies do not develop games that use those features in a meaningful way. A chief example of that is Physx, which could change games in wonderful ways but it gets used only for cosmetic things, like flying debris (that can be turned off) and not for anything central to the game dynamics, because then only nVidia users could run the game. So, in a gaming context, the proprietary features are hated because they are cool, but they don't get used, so we don't ever get access to them. In a DCC context it's different, because people will buy hardware tailored specifically to run certain software so it's much more irrelevant if something is proprietary or not. In DCC, someone will use the feature, even if it runs on only on certain hardware. And almost everyone already uses Nvidia for DCC stuff.

To be fair, the other "respective hardware manufacturer" AMD also tried to implement something similar and they made it open. It's called Bullet and it's in LW. Nvidia has just been much faster and more ambitious at developing more advanced effects.

MSherak
09-11-2015, 06:22 PM
Not to step on any toes, but it seems to me that a company who makes something cool that runs on their proprietary hardware / libraries has every right to NOT port it to other hardware. Afterall, it is part of their marketing strategy to stay relevent.

Seems that users of other hardware should be complaining to their respective hardware manufacturers to implement something similar. We all make choices and sometimes you make the wrong choice because you are allied to a certain company or brand.... For years, I always chose hardware based on what I could afford, which had the best support with regards to Lightwave (lightwave was the deciding factor even though I was / am a big gamer). The choice was always easy for me - an Intel based machine (and believe me, I was a resistant and hard core Amiga user / evagelist back in the day).

I did use ATI cards for a brief time, but couldn't stand the software (even though I liked the hardware better). Now, I see the writing on the wall - nVidia has good stuff and much better marketing - you go with what will suit you best and what brings you the most benefit for your situation. As I am a tinkerer and like building and custominzing my machines, I am a PC/nVidia guy. You get more bang for your buck no matter how you spin it - because you have the option to choose.

Choose which best works for you. If you see that somebody else has something you like, get it or don't. But thinking that everything should be on a even playing field and faulting a company for utilizing a tech / marketing edge to their favor is evil seems pretty wrong to me.

Sure.. They can do all they want. But my SurfacePro3 is not going to be able to run their stuff due to CUDA. Not all of us like to sit in front of a desktop unit and use a specific card for functionality. Would be better if they made their tech work with any GPU (OpenCL). But you are right the shareholders would not like that. So I have to buy their card if I want to use it.. Fair enough. I'll just wait..

Example: I have a license of Octane and it has been sitting around. Otoy is finally moving to OpenCL in 3.0. This means that even my SP3 will be able to use Octane. Otoy gets it and Arnold is not far behind. Hoping this is the route that Lightwave also takes in the future also.

creacon
09-11-2015, 07:02 PM
So you want to run my plugin that makes my dual Xeon workstation with 40Gb Ram run hot and out of memory, on your surface pro.
Good luck with that :-)
I don't get the problem, if something does what you want/need or it solves your problem, buy the software/hardware that you need and stop whining.

I am testing on a GTX 960 too, that will costs you $200 and you can dedicate it to PhysX/Flex and keep running your AMD card.
If that is a problem you don't "need" fluids in LW.

creacon

lightscape
09-11-2015, 08:47 PM
You want to run fluid sims on your surface pro 3? :ohmy:

seghier
09-12-2015, 03:23 AM
surface 3 ? who use it to create games ; animations and renders ?
and surface pro 3 cost :
Surface Pro 3 - 128 Go / i5 / 4Go RAM $899
Surface Pro 3 - 512GB / Intel Core i7 $1,799.00
with these prices you can buy : at least " 2 GeForce GTX 980 " or " GeForce GTX TITAN Z + GeForce GTX 970 "

tyrot
09-12-2015, 01:23 PM
creacon ... your reply made my day ... any timeframe for plugin...?

spherical
09-12-2015, 03:50 PM
surface 3 ? who use it to create games ; animations and renders ?
and surface pro 3 cost :
Surface Pro 3 - 128 Go / i5 / 4Go RAM $899
Surface Pro 3 - 512GB / Intel Core i7 $1,799.00
with these prices you can buy : at least " 2 GeForce GTX 980 " or " GeForce GTX TITAN Z + GeForce GTX 970 "

Agreed. Other than being able to do a client demo or work in you local espresso bar, a laptop or tablet to do 3D work just doesn't compute... to make a pun.

MSherak
09-12-2015, 06:11 PM
SP3 is an example, looks like the everyone missing the point.. But hey at least you will have fluids on that desktop unit for some extra cash.. Go CUDA!!!!

lightscape
09-12-2015, 09:23 PM
Vray was tested and Cuda on nvidia cards is fast. Opencl on amd cards is slow. Amd cards and cpu in general suck. They need to do better to compete.

tyrot
09-13-2015, 10:42 AM
cuda saved our business... editing composition and render