Some research with implicit surfaces

[jacksc02]

stop it Dieter! This looks amazing, if you need a tester I will be more than happy to test it out.

[Geddart]

Wow, great work Dieter!
Looks i can finally stop bugging ca scanline with emails about that stuff.
These forms are really something that would be very hard to model. Sponges, rotten wood and coral reefs may not be the only things one can do with it. I'm sure we will find more use of it when we can start experimenting with it. So please keep it up!

For the name... how about VrayMarching, VrayStructure or VrayDensityMesh. I know they're all not really cool but it's hard to find a name i think.

[Da_elf]

well. hedges would be done well with this as well. rather than using displacement

[dimo]

Hi,

I was at FMX Expo and listened to the speech of Stephan Trojansky from Scanline CA about their R&D. Compared to what they do with implicit surfaces and their fluid simulation my stuff is kindergarden, really. Also the rendertimes they mentioned for their implicit surfaces are almost not believable.

I'm currently trying to get my stuff faster. I have a Voxelcache (regular Grid) now and my own Voxeltraversal Algorihm. it speeds up rendering in certain situations but not always. It introduces artifacts and needs a looot of ram. So no breakthrough yet.
Vlado proposed a BSP tree instead of the regular Grid but honestly I'm not sure how to implement this for Voxels. An Octree would be a good alternativ, too. I do have some papers for Octrees for Voxels.

I also have implemented a fast marching algorithm to fill the grid. This should in theory make it more independent from particle/vertex count. The results however look very different compared to the metaball algorithm.

But I'm still working on this.

best regards,

Dieter

[jacksc02]

Thanks for the update Dieter.
The scanline plugins look amazing, havent heard much about the render times.

keep us updated on the progress

[gafl]

Hi Dimo !
What about your work nowdays ?
Have you made any progress ?
Gaël

[dimo]

Hi,

too much pressure at the job right now. We don't have weekends until October from now on

Best Regards,

Dieter

[jacksc02]

Thats a shame Dieter.
Hopefully you can get some time to work on this!

[dimo]

Ok,
still crunching time here. I still have no time to do further testing and developing. ( I thought the 85h-week was history
Why not start alpha stage now? It doesn't help if the stuff lingers on my harddisk. So if some of you want to play with the actual state and are willing to contribute with some feedback, drop me an email. But beware: Feels still a bit like using Maxwell (ha ha could'nt resist this one).
And alphatesters are expected to put some cool images to this threat

For those interested. I put a small text together describing my new approach. You can find it here:

http://www.dimo3d.info/Implicitsurfaces.pdf

Best Regards,

Dieter

[stefan]

Hi Dimo, nice effort and good PDF too. I was looking for a Blob shader for a long time, Vray and AfterBurn don't work together very well.
I was very close to asking Vlado for the SDK myself, but I was affraid of the leaning curve (I'm not a die hard programmer unfortunately)

I'd love to take a test drive.

Cheers,

Stefan

[DaForce]

sweet, very informative PDF

email sent.

[gafl]

Hi ! yes, the explanation is very good.
Good Job
Gaël