Re: Realflow and the bottle neck of fluid sim

["adrian" <adrian.wyer(at)fluid-pictures.com>]

i think you nailed the whole 3d industry, with or without realflow.......

almost every job is like this, a battle of wills versus technology, versus 
budget versus time

and we (the artists) are the casualties of that war

man am I bitter this morning....!

a
----- Original Message ----- 
From: "VR XSI Forces" <forum(at)virtualrepublic.org>
To: <XSI(at)Softimage.COM>
Sent: Friday, July 07, 2006 12:06 PM
Subject: Realflow and the bottle neck of fluid sim


> A first simple >filling the water glass - of course< test with RF4 gives me
> the impression that RF4 is working better. No crashes yet, looks nice in 
> XSI
> ... simulation seems to be slower but more accurate?!
>
> But let's talk about the major bottle neck of fluid sims with RF.
>
> Filling a water glass is working fine but we have to understand that this 
> is
> just a simple simulation and far away from a more complex task. Why does 
> RF
> not have some perfect setups for loads of different fluid types which work
> physically perfect in small and big scaled situations [let's forget the 
> sim
> time in this case]? The only answer I could imagine is: the developer also
> doesn't know how to set them up.
>
> RF ist fascinating indeed and it's fun to play with. The problem begins if 
> a
> client needs to solve a fluid task, let's say liquid chocolate flowing 
> like
> highspeed filmed over a waffle. You get some real footage of perfectly
> filmed high-speed EUR 40.000 fluids in high aesthetic style. Due lack of
> fluid sim experience you start with high motivation beeing sure: I will
> solve that. After a few days trying hard you will come to the conclusion:
> not that easy but possible. One week later you feel running out of time 
> and
> budget and you ask your admin what a 4 proc number crunsher with 128 GB 
> ram
> will cost. 2 weeks later you will understand that the beauty of those high
> speed fluids is mainly ruled by perfect >infinite res< surfaces and a
> special complexity you can't work out with RF in a reasonably time plus
> those damn budget limits. The client expects that he will get the computer
> generated fluids cheaper and that he is able to have some post control 
> over
> it for >fast< changes or variations. At the end you have to tell them: 
> each
> change will take 2 more days [if they like the result].
>
> A first time killer is that you can't work in fast low res mode with short
> sim times and just set up the values to a higher res because this will
> affect the whole behaviour of the fluid. So you must start with high res 
> as
> possible to get sure that the result you present to the client is not much
> different to the next >perfect< step. The work flow is getting as low as 
> the
> sim times will explode exponential. A good but still not perfect looking 
> 150
> frames >filling HALF the water glass< will probably take 3 hours on a Dual
> XEON 3,0 Ghz machine. Doing more complex stuff end up easily in a whole 
> day
> of sim calculation you have to wait for till you can go on. There's no
> chance to get it faster with usual equipment because you are forced to 
> work
> with the higher settings till you hopefully get what the client likes.
>
> And at the end you are surely confronted with another unfair fact. Once 
> you
> have worked out hard something which looks close to a filmed reference the
> client will drive you mad with: this doesn't look real ... there's 
> something
> wrong ... the drops look wrong ... this drop should fly there and those 3
> drops shouldn't be there ... etc. etc. etc. ... even if the result looks
> [theroretically spoken] 95% like the filmed material strangely 95% of the
> clients would not really accept or approve it. I am totally sure. And this
> is the next big surprising killer for the whole production. Without a good
> fighting producer the chance is high that the whole job ends up in pain 
> and
> chaos?!
>
> Probably I am wrong ... due lack of intensive RF experience *:-)
>
> Michael
>
> ---
> Unsubscribe? Mail Majordomo(at)Softimage.COM with the following text in body:
> unsubscribe xsi 

---
Unsubscribe? Mail Majordomo(at)Softimage.COM with the following text in body:
unsubscribe xsi