Logistical Nightmare!! Breaking up the storm

[FutureMan]

Thanks to Dhin I've revived an idea for a scene of rain hitting a windshield. Now for this I'm using Clothreyes to simulate the drops deforming as they hit the windshield and rolling down. Here's where my problem comes in, Clothreyes needs to calculate and create keyframes for each drop. Now one drop isn't too bad only takes a couple minutes at max to calculate, but a whole rainstorm would take far to long to simulate altogether for my computer; propably weeks, at least days and I can't do that.
However Clothreyes allows you to merge in the calculated keyframes(deformations in this case), but it needs to merge in the objects as well, in this situation the windshield and the small group of drops, and you can not delete any of the objects. So this is leaving me with many copies of the windshield I don't need that I simply make unrenderable and hide, they don't show up in the renders.

Now here is the real problem due to the fact that windshields aren't flat and curve I simply can't use the same calculations over and over, I need to calculate for each section. So I'm going to break it up but this is going to get confusing very fast! I'll of course be saving many times. As of now I'm not exactly sure what arrangement I'll use but I'll have it all planed out before I start.

Anybody got ideas how I should proceed? Beside very carefully that is

[David Ciprian]

If i were you, i'd use a texture for the windshield so the splashing drops arn't calculated.A deflector for the particle system will do the cliping for the windshield.It would me hardly noticeable in a animation

[kuman]

might i suggest another pathway.

i assume you're not doing this for real time since you're considering cloth reyes as a soft body solution for rain drops.

anyhow i was thinking about how you could change your shots to accomdate a cleaner solution. why not start with a shot where you see through from the inside of the car in a small section so you only need a small simulation and then cut to a scene from the outside fo the car moving through the rain and then transfer back to the a farther shot in the interior.

my idea would accomdate 3 things.

1-small simulation at first, that you could possibly get a animated texture from in terms of alpha

2-zoom out to a simpler solution of partilce systems

3-return to a farther inside who where you could fake it with textures, possibly from a reflection/refraction mapping of that animated alpha you derived from 1 to use on the windshied.

i guess what i'm saying is that your method you're describing sounds immensely tedious and slow, perhaps a reconsidering of shots you do can help you circumvent that logistical nightmare.

good luck
-k

[FutureMan]

You make some good points about limiting the scope of the shots to better accomodate my processing power.

This project has alot of meaning for me, so I'm not too worried about the amount of work\time needed to complete it. I'd want to get it done exactly right rather than rush it or trade quality for speed. Also this is going to be in my free time so I'm able to work under my own deadlines.

This scene is going to be at an important point in the story so it'll be valuable to me if this comes out perfect, no matter how many months it takes.

I'd prefer to do it the way I discribed before simply because I think it would give me better freedom in the end. However it was my intention to always be inside the car looking out through the windshield. So I think I might do a merge of ideas and only really simulate for the beginning of the storm and then it would reduce the amount of rain needed.

But I'll go ahead and finish storyboarding it out so I can get a better plan and start moving ahead.

Also I forgot to mention that the car is stationary and not in motion so that's not a factor at all.

[FutureMan]

Here's a quick test of what kind of look I'm trying to get.

Just one of max's standard sky maps to get a feel for it.

Edit:Oh and I forgot to mention I'm hoping on intergrating this with some as yet unfilmed live action, won't really be a cg movie. I want that sureal extreme detail that I couldn't hope to achieve with my limited live budget, so it will be refracting the live action in the rain drops. I figure if I can get a small short made maybe I can get further funding later on. Sorry sometimes my mind doesn't work in straight lines, get focused on one aspect and forget about the rest.

[blesson2k]

Couldnt u try reactor or realflow, the latter better equipped for this kind of stuff?

[FutureMan]

Yeah you might be right about RealFlow, I don't have it now but I'll look in to it. Have some other ideas that need a good liquid simulation.