Metaballs for VRay

[jonnybefree]

that looks realy nice!
what are cal./render times for it?

[dimo]

Thanks for your intrest.

...
This stuff is still so slow thats its probably not practical to use it for animation, for high quality stills requiring correct motion blur its perfect.

That's my experience, too. But on the otherhand with very huge particle counts this approach might again be interesting, because meshing takes so long.
Perhaps we should go the route to mesh it dynamically like VRayFur does.


Dieter

[MasterBercon]

I just can't get the Zhu-Bridson field working.

Here is the code:

for (int i=0;i<count;i++) {			
	r = (pointCloud[points[i]]-pos).lengthSqr();
	if (r > blobSize2) continue;
	w = function(blobSize2, r);
	totalP += pointCloud[points[i]] * w;
	totalW += w;		
}
res = 1.0 - (pos - totalP / totalW).length() / blobSize;

And here is the result:
http://www.niksula.cs.hut.fi/~jylila...taballs_22.jpg

Some ugly/buggy disks floating between particles.

[joconnell]

Looks really good! It's headed directions to make realflow renders a lot nicer it seems - will also sort the motion blur issue! I always had issues where I couldn't use small enough metaballs to get a nice surface in realflow - the lumpy nature of the meshes really cheapened what looked nice when viewed as particles only. Something like this is really interesting in that regard

[Da_elf]

updates? cant have this post all the way back on page 3

[MasterBercon]

Unfortunately I've been busy with other things such as my bachelor's thesis which I finished yesterday. Hopefully I'll get something ready for you guys to test next month.

[dellis]

Great stuff! What is the topic of your thesis by the way?

[MasterBercon]

A bit out of topic, my thesis' title is "Future payment systems - busines point of view" and is mainly about Single Euro Payments Area, so its not exactly computer graphics related.

[jujubee]

Uh-oh - we losing MasterBercon in the 3d field to business and banking?

[duke2]

MasterBercon has now achieved refugee status.

[dimo]

Hey Jerry,

I looked at my code. Here is what I did to get rid of those discs:
if ((sum_w)<(Isoparams.radius))
{
sum_r*=smoothstep(Isoparams.cutoff,1.0,sum_w/(Isoparams.trim));
};

I have two additional parameters cutoff and trim

Hope this helps.
Best regards,
Dieter

[MasterBercon]

Thanks dimo! That does indeed help with the discs.

http://www.niksula.cs.hut.fi/~jylila...taballs_23.jpg

Ireal r;
Ireal w;
for (int i=0;i<count;i++) {			
	r = (pointCloud[points[i]]-location).lengthSqr();
	if (r > size2) continue;
	w = function(size2, r);
	totalP += pointCloud[points[i]] * w;
	totalW += w;		
}
... ... ...
float radius = blobSize;
if (field.totalW < blobSize)
	radius *= smoothstep(field.totalW / trim, cutoff, 1.0); // TODO: Optimize 1.0 part
field.res = radius - (pos - field.totalP / field.totalW).length();

I should do some major refactoring with the code and then start working on UI. I wont add any more features before I'll release some beta version. It'll probably take me a month or two to get it done.

[dimo]

Hey,
yes, that looks pretty much the same as what I have.
I am looking forward to your version.
What I did here, was I wrote a new marching cube implementation to view the mesh in the viewport.

Best regards,
Dieter

[stilgarna]

Hello,

I just want to thanks you guy...

Your work is very impressive. As dimo, you're doing a great job for the whole community and it's not that easy to keep such a hight quality every time !!!

As soon as I reactivate my paypal account I'll make a donation. And my company to.

Best regards... (I'm out of this thread !)

[raymarcher]

why don't you two guys work together to make something even better