New Liquid Plugin(not fluid plugin) coming eventually

New releases and general discussions.

Post » Sun Aug 15, 2010 5:17 am

ok
I need to figure out what can be done with boundaries, because for now it's not as flexible as I'd like, and I'm going to finish the fluids plugin first

and lastly, in this demo, the black squares are adjusting to fit around the liquid, not the other way around, so again, the liquid is not reacting to the black squares:

but look: liquid physics in construct:
http://dl.dropbox.com/u/1013446/liquid.rar
Spriter Dev
B
87
S
21
G
12
Posts: 3,240
Reputation: 16,461

Post » Sun Aug 15, 2010 5:30 am

no sure what's happening but itlooks pretty cool
B
16
S
10
G
5
Posts: 255
Reputation: 3,934

Post » Sun Aug 15, 2010 9:31 am

Looking great so far. Excellent work.

Just how efficient is it though? Looks like there's a lot going on.
B
2
S
2
G
3
Posts: 143
Reputation: 1,536

Post » Sun Aug 15, 2010 10:14 am

Looks promising :)

Is this going to be a pure liquid physics simulation or will the plugin care about the graphics as well? And, will it be fast enough to be integrated to a game?
Image
B
23
S
8
G
10
Posts: 1,820
Reputation: 8,242

Post » Sun Aug 15, 2010 1:45 pm

[quote="Shady":2g8tblbz]Looking great so far. Excellent work.

Just how efficient is it though? Looks like there's a lot going on.[/quote:2g8tblbz]
once again, not as efficient as a gpu solution, but I'm going to have a look at making a directcompute, or opencl implementation. Depending on how similar they are to CUDA, as there is a open source for a CUDA version as well. If not, I'm also going to have a look at making this one work for multiple cores.

I can get about 1800 particles before it starts slowing down

[quote="tulamide":2g8tblbz]Looks promising :)

Is this going to be a pure liquid physics simulation or will the plugin care about the graphics as well? And, will it be fast enough to be integrated to a game?[/quote:2g8tblbz]
it's going to work how the final version of the other plugin will
actually, that's what it's doing now, those are sprites, and after I get some better container manipulation working, I'll see how fast it goes while other stuff is going on


also, check it out...
about the happy faces...don't ask:

http://dl.dropbox.com/u/1013446/fluid2.exe

and fluid attracted to a central point of gravity:

http://dl.dropbox.com/u/1013446/fluid3.rar
Spriter Dev
B
87
S
21
G
12
Posts: 3,240
Reputation: 16,461

Post » Sun Aug 15, 2010 2:28 pm

Freaking awesome!

Performance is a little slow on my machine. Here's hoping you can get that opencl version working!
Moderator
B
87
S
32
G
33
Posts: 3,005
Reputation: 27,397

Post » Sun Aug 15, 2010 4:18 pm

lucid, I sent you a pm.
Image
B
23
S
8
G
10
Posts: 1,820
Reputation: 8,242

Post » Sun Aug 15, 2010 6:07 pm

just pure awesomness
B
11
S
3
G
4
Posts: 622
Reputation: 3,186

Post » Sun Aug 15, 2010 7:08 pm

It is of course looking good so far. I really hope you find a good solution in the end, because the performance does not seem to be there.

Dual core systems are pretty common, just don't want to have to require a quad core system for a 2D game to perform at a solid frame-rate.
B
2
S
2
G
4
Posts: 156
Reputation: 1,762

Post » Sun Aug 15, 2010 7:18 pm

small update, unfortunately, it seems I don't possess the skills yet to add gpu or multicore acceleration to this simulation, but all hope is not lost yet
the simulations I've uploaded here are about 1500 particles each if I remember correctly
the speed problems come from displaying 1500 sprites.

I ran the simulation with 150,000 particles and no display and it was about 150fps
obviously we need to be able to display the particles in some way, but a more efficient form of display undoubtedly exists to find an outline of particles, although greater minds than mine have had problems finding a perfect one:
http://www.rchoetzlein.com/theory/?p=129

however, when limiting the simulation to 2 dimensions the problem might be simpler
I won't tackle this until after I make an alpha release of the fluid (smoke) plugin, but I have one idea where I set up a grid like the smoke simulation, and count the particles in a given grid space, and use a single sprite which changes size depending on the number of particles, when the number is too small, it will split and be represented by multiple sprites

also, though it doesn't seem possible without rewriting the whole simulation, which I'm not up to, if I find a way to actually limit the simulation to calculating in 2d that could greatly speed things up. even the 2d looking versions youve seen are of 3d water in a 2d container, so I believe there is wasted cpu time on the 3rd dimension even though it isn't doing anything

I'll keep you posted
Spriter Dev
B
87
S
21
G
12
Posts: 3,240
Reputation: 16,461

Next

Return to Construct Classic Discussion

Who is online

Users browsing this forum: No registered users and 2 guests