Softimage Nvidia Flex

Comments on work in progress
Pedrito
Posts: 27
Joined: 21 Feb 2018, 12:46
Location: Zaragoza,Spain

Re: Softimage Nvidia Flex

Post by Pedrito » 12 Mar 2018, 20:20

PD:: do you guys want a small tutorial on how I am creating the VDB vector fields? Or is that something already known here? Because I never used it before and one day ussing hudini my mind says, well you can do this in xsi thanks to oleg...lets try, and it works, but never seen anyone using it theese way so.... If there is interest I can create a small video and scene showing the process :)

User avatar
rray
Moderator
Posts: 1554
Joined: 26 Sep 2009, 15:51
Location: Bonn, Germany

Re: Softimage Nvidia Flex

Post by rray » 12 Mar 2018, 20:52

I was thinking about a gas/cloth interaction similar to what's in this video of your's



but with air instead of water. So the cloth would be "under" air particles and the external forces would act on the air instead of the cloth directly.

I'm not sure if it's possible to make that one simulate stable/fast enough, but I think that two-way interaction is what's missing from most cloth simulations I've seen so far --- I always had the impression that they simulate cloth in a vacuum and the "fake" forces only work one-way instead of back and forth between cloth and air.
I will try yo replicate that and add it to the bundle if you want :)
The plastic bag scene is probably the first thing I'd try to recreate when it's released. So I wouldn't peek how it's done. Depends on you & your time ofc! but would be interesting to compare the results later.
"as a maya discussion grows longer, the probability of a suggestion involving deleting the preferences approaches one"

softimage resources section (formerly known as rray.de/xsi) updated May 30 2018

kissb
Posts: 16
Joined: 18 Oct 2014, 17:25

Re: Softimage Nvidia Flex

Post by kissb » 12 Mar 2018, 21:12

these fluids made me wet

Pedrito
Posts: 27
Joined: 21 Feb 2018, 12:46
Location: Zaragoza,Spain

Re: Softimage Nvidia Flex

Post by Pedrito » 12 Mar 2018, 21:14

I dont think thats possible in the way you say, sorry :(
but you can offcourse "fake or create" dual interaction forces if you use a vdb field, first you create your rest air( rest in sense there is not cloth, but you can add some animated noise ) you add the cloth, and apply the field forces to the cloth, simulate, then you have new forces on the cloth caused by both the field and the solver, if you apply that new force derived from cloth to the field again, you are modifiying the field, so you have some interaction... Ofcourse you will need a fluid solver in case you want swirly air movenent i think, but you can use emFluid5 at low res to do the advection instead of a simple vdb field, and with that you can fake a very low dense ambient...

All i said now is pure thinking, dindt try anything yet but I hope that make some sense for you :)

About the cloth specific forces I talk before...From the flex manual : "Lift and drag - Triangles will generate drag forces and lift forces proportional to area and orientation. This can generate interesting fluttering motion on pieces of cloth, typically drag forces should be set higher than lift."

Pedrito
Posts: 27
Joined: 21 Feb 2018, 12:46
Location: Zaragoza,Spain

Re: Softimage Nvidia Flex

Post by Pedrito » 12 Mar 2018, 21:20

kissb wrote:
12 Mar 2018, 21:12
these fluids made me wet

;))

Pedrito
Posts: 27
Joined: 21 Feb 2018, 12:46
Location: Zaragoza,Spain

Re: Softimage Nvidia Flex

Post by Pedrito » 15 Mar 2018, 22:03

Hi there! I have finished and tested the Skin Setup and it is workig very good :)


User avatar
rray
Moderator
Posts: 1554
Joined: 26 Sep 2009, 15:51
Location: Bonn, Germany

Re: Softimage Nvidia Flex

Post by rray » 15 Mar 2018, 22:29

Haha I didn't think this creepy thing could be made any creepier, I couldn't have been more wrong :D Great stuff again!

Air+cloth 2-in-1 simulation would be a great research topic for a scientist. Sadly nothing to be found anywhere yet.

I was already guessing it's a better fit for a volumetric instead of particle approach. Thanks for the workaround-workflow tips, I'll surely try that out.
"as a maya discussion grows longer, the probability of a suggestion involving deleting the preferences approaches one"

softimage resources section (formerly known as rray.de/xsi) updated May 30 2018

Boiler
Posts: 19
Joined: 17 Feb 2015, 13:15

Re: Softimage Nvidia Flex

Post by Boiler » 08 May 2018, 11:08

Is there any news ?
;)
Long time no new videos on vimeo channel ((

anhungxadieu
Posts: 175
Joined: 17 Apr 2014, 10:39
Skype: nguyenvuducthuy

Re: Softimage Nvidia Flex

Post by anhungxadieu » 08 May 2018, 14:52

Still waiting :(

Pedrito
Posts: 27
Joined: 21 Feb 2018, 12:46
Location: Zaragoza,Spain

Re: Softimage Nvidia Flex

Post by Pedrito » 10 May 2018, 10:30

Hi everyone!

Sorry for the long time no news anywhere,
in fact I had this project a bit abandoned last month because of lot of work at the office and some personal issues that doesn't allowed me to work in house.
I'm trying to get some time to finish it, It could be released now but there are a lot of small thing I will like to fix before releasing it...sorry for tat :S

But if I encounter my self again without any time to finish it....well I will probably release it for everyone as it is in not to much.

Hope you guys understand it, and I promise it will not be to long from now!

PD:: the only thing extra from the lasts videos ( including some fixes ) is a scene with hair made with strands and FLEX (very simple one ), and a faster version of the CONSTRAINED_GRAINS_SETUP of the first video.

Land-Y
Posts: 2
Joined: 04 Dec 2009, 02:58
Skype: land-yaore

Re: Softimage Nvidia Flex

Post by Land-Y » 18 May 2018, 08:15

I am very much looking forward to it !

User avatar
rray
Moderator
Posts: 1554
Joined: 26 Sep 2009, 15:51
Location: Bonn, Germany

Re: Softimage Nvidia Flex

Post by rray » 25 May 2018, 12:16

Great to hear - hoping everything works out well for this project (and you!!)
"as a maya discussion grows longer, the probability of a suggestion involving deleting the preferences approaches one"

softimage resources section (formerly known as rray.de/xsi) updated May 30 2018

Pedrito
Posts: 27
Joined: 21 Feb 2018, 12:46
Location: Zaragoza,Spain

Re: Softimage Nvidia Flex

Post by Pedrito » 14 Jun 2018, 11:18

Hi there! Look at this ;)



You will also have it for xsi sonn! I promise :)

User avatar
rray
Moderator
Posts: 1554
Joined: 26 Sep 2009, 15:51
Location: Bonn, Germany

Re: Softimage Nvidia Flex

Post by rray » 17 Jun 2018, 19:45

>>You will also have it for xsi sonn! I promise :)

Yay!

>>unscalable, limited particle count and nondeterminsitic, but pretty darn fast

Alright! See that's great.. seeing what to expect in 2 seconds, rather than sifting through PR swamp for 30 minutes!
"as a maya discussion grows longer, the probability of a suggestion involving deleting the preferences approaches one"

softimage resources section (formerly known as rray.de/xsi) updated May 30 2018

jonmoore
Posts: 108
Joined: 30 Jul 2016, 18:18

Re: Softimage Nvidia Flex

Post by jonmoore » 20 Jun 2018, 17:10

I've been testing varying FLEX facilitated options over the last couple of years in Max, Lightwave and C4D (plus Houdini thanks to Pedohorse's generous efforts). I put together benchmark reports for a large advertising group (Publicis Groupe) each year of the major players.

The interesting this is that each plugin offering has exhibited different strengths. Lightwaves's Unified Physics plugin is rather nifty at course liquid sims. Used in the right way it's capable as well as fast. C4D has a plugin called Uniflex 4 (by Remotion, a single developer vendor). The plugin is flakey in general but it's softbody (inclusive if cloth) and rope simulations are stable, reasonably 'physical', performanant and fiercely fast on a latest generation GPU (I tested with a 1080 Ti). There are a couple of Max plugins available that make use of Flex for scattering (anything like rocks, pebbles, breakfast cereals - you get the picture).

Overall I'm left with the impression that as long is keep to the central tenet that Flex is:
"unscalable, limited particle count and non-deterministic, but pretty darn fast"
It's useful for so many use cases, especially motion design stuff where beautiful motion doesn't necessarily mean it has to be physically correct. You're also using far corser particle resolutions but that still leaves the door open to lots of creative exploration. With cloth the trick is hiding the underlying coarseness of the simulation but Remotion have managed to achieve this brilliantly with Uniflex.

I'm really excited to see what Pedrito has achieved with his Soft implementation of Flex. I'd imagine it will sit alongside Mr Mootz's wonderful suite of tools beautifully. :)