Let’s see if this movie will attach…
- Chad
It didn’t 
Stupid Information Superhighway.
Script timed out. It’s 29MB, so maybe I should be putting this on your FTP or something.
I’ll make a more compressed version, too, and try again.
- Chad
10%ish
Wow. impressive stuff Chad!
It really went a long way from the previous test I saw!
How is the first turbulence done? It looks like it's creating some sort of net!
Cheers and keep up the nice work.
MAC
It’s a mix of 3 forces, 1 keep apart, and Box3, which sets initial velocities based on tissues, and does some prevalent velocity mixing to maintain clumps of flesh.
There’s a more refined version, but my licenses pooped out, so this is all I had to show, and I wanted to get something out there before siggraph. If I get new licenses, I’ll try to get the better version up soon.
- Chad
>How is the first turbulence
>done? It looks like it's
>creating some sort of net!
Sorry, didn't understand the question the first time.
The blood vessels under the skin are darker. So I told Box#3 to get the point color and apply the reciprocal of that as the length of the speed vector. End result is the blood vessels explode.
- Chad
No licenses doesn’t mean I can’t use the PRT Loader to see what partitions did manage to get rendered over the weekend. Pretty cool to see in viewport.
- Chad
No licenses doesn’t mean I
can’t use the PRT Loader to
see what partitions did manage
to get rendered over the
weekend. Pretty cool to see
in viewport.
Duuuuude! :o)
This looks fantastic!!!
Btw, we added color clamping for the viewport display today, because floating point colors displayed warping colors in all previous versions and it looked rather ugly.
I’m just happy it looked at all. I was waiting for the preview to crash, but it just kept working. The multipass depth of field in the preview render worked well too. Velocity vector display is really fast.
Dunno if I mentioned this before or not, but as a wishlist item, it would be cool to have integer color and density. With the volumetric density and the shadow maps, the high fidelity of float colors is largely lost. The large range isn’t always used either, depending on what the source it.
Position and velocity still make sense as float, it’s just that with all the mixing going on during rasterizing (in float) the accuracy of the float colors just isn’t coming across as much as the space savings (especially after compression) would be appreciated.
- Chad
Chad dude.
you're the most advanced Krakatoa user so far. Why don't you pull out a DVD with all the tricks and cool stuff you've found out! ;)
Allright, if you're not in on a DVD(cause it does take a hell of a long time) then perhaps you could put a couple of web pages up as notes or tutorials? Maybe Frantic would be keen to hosting them!(or use their wiki?)
Your questions and comments on the beta are very interesting and very deep into Krakatoa usage. I'm not there yet but it would certainly be welcome to have examples that I could try out and learn from.
Cheers,
Marc-André
Afraid Bobo will beat me to it. He works harder and smarter than I do.
Thanks for the compliment, though. Truth be told, I’m surprised there’s no Box#3 DVD’s out. Gotta talk to Chris Thomas about that. Limited market, I suppose, so it can’t be lucrative, but if you enjoy doing it anyway, why not? It’s a topic I’d be interested in putting time into producing, too. Ben’s been working in the Box#3 SDK, so maybe he and I could team up and make some tutorials along with some free sub-operators. With Krakatoa coming out, there’s going to be a market for people looking to push particles farther. It opens doors.
I’ll probably contribute to the Frantic Krakatoa wiki when it goes public. I contribute to the Vfxpedia, and that’s a lot of fun.
Nuther test.
- Chad
Here’s some more renders. Compression is not kind to particles.
- Chad
great tests!
especially the movement of the particles with forces and box#3. really currious about the dataflow setup. still a box#3 noob ;-(
i know u did tests on that for long now. but i never had the time to read how it was all done in detail ;( i hope for enlightment…
kind regards,
Anselm
Cool stuff…
How is the red particle cloud done? It looks like smoke
Is the initial hand rendered with particles too?
The segmentation is all done in Box#3. So we test for bone, ligament/tendon, muscle, fat, skin, etc. with Box#3.
We also use Box#3 to evaluate the neighborhood voxels to determine vectors, similar to how we do normals, but here we set the normal to be the explosion vector.
The first test showed it a little better, but there’s also some scaling of the velocity vector based on tissue segmentation and prevalent speed. The second set of tests was too violent for that effect to work without much lower integration steps.
There is no geometry in the scene. We did a test where we extracted isosurfaces with Blobmesh for the bones, but particles were faster and easier.
Next test involves taking my boss’s head CT and seeing what shotgun does to it.
- Chad
Hope I never become your boss 
Well, yeap there are many challenges you can’t see at first. I discovered that everything is particles. Wonder how many you took.
I had to render some stuff in 4k. I think your animation would blow your hard drive in that size.
Krakatoa doesn’t really care what your raster size is. It’s nearly the same speed no matter what. Of course, you might need to put more particles in there to make 4k look good, but it’s hard to tell. You might try rendering at 4k, and at 1k and resize it to 4k. See how they compare.
- Chad