Experiments..

Caronte · Post by **Caronte** » Wed Jun 13, 2007 9:50 am

:shock::shock::shock:
Awesome

Ricky · Post by **Ricky** » Wed Jun 13, 2007 11:04 am

tinman999 · Post by **tinman999** » Wed Jun 13, 2007 11:06 am

OnoSendai wrote: I'm still thinking about how to handle the volume data formats etc...
Any more suggestions are welcome!

Off the top of my head

Assuming you're using grids to represent the 3d data,
the simpliest choice would be a spar list of 2d image, when storing the datset on file each 2d image can be compress by zlib or some other compression library. For example assuming you got a grid xyz of size 30x30x30, only x at 0, 5, 10, 12, 14, 25, 30 contain non-zero data you could stored it like this

xyz <- index order
7 <- number of entries
0, image 1
5, image 2
10, image 3
12, image 4
14, image 5
25, image 6
30, image 7

trilinear interpolation is simple to do and more advance sampling and reconstruction method should be ok too

Another option would be to replace the 2d image slice with n 1d polynomial depending on how empty it is and how smooth you want the data to be.

eman7613 · Post by **eman7613** » Wed Jun 13, 2007 11:48 am

manitwo wrote: you can be really proud of yourself - so go and get yourself a beer now - you deserve it

and a woman or two!

I wont pretend that i can contribute to the theories being represented, but i have a question. One of the new toys (well, for me anyway) in Cinem4D is pyrocluster (clouds and smoke and that stuff) but it can also make them illuminate nearby objects * stuff ie: fire. Could you code this into indigo with the new toy?

Anthony · Post by **Anthony** » Wed Jun 13, 2007 4:07 pm

Lord of the Rings Junkie wrote:Hold your horses guys, an Earth-sized sphere is a far cry from a planet with full terrain! Awesome feature though, absolutely cannot wait!:D What's next, a scale Sun model 93 billion miles from Earth?:) (Hopefully it wouldn't take light 8 minutes to reach your render...)lol

[Edit] Both linear and rotational motion blur would be cool things to see in 0.9

lol he was making a not so subtle reference to the book "The Hitchhiker's Guid to the Galaxy" where mice are actually the most intelligent beings on Earth and if I remember right they build planets, or at least someone builds planets, and the most fun parts to make are the coastlines and fjords.

Zom-B · Post by **Zom-B** » Thu Jun 14, 2007 12:35 am

OnoSendai wrote:It would be relatively simple to read in a list of particles, and 'splat' them into a volume map, using some kind of 3d filter for each particle (e.g. gaussian, step function, etc..)
Then the volume map could be used for rendering.
Alternatively the particles could be stored in their own data structure, which might be better for sparser sets, but that would be a bit trickier to code.

Does this means that a new generic Indigo particle Format is needed, that the 3D aplications have to export to (via new developed export Plugin)???

This would perhaps give you (Ono) the most flexibility for your interaction with this particle files.... ergo better and faster visual results.

The Drawback will be the developement time for each 3D application ...

StompinTom · Post by **StompinTom** » Thu Jun 14, 2007 1:05 am

ZomB wrote:
OnoSendai wrote:It would be relatively simple to read in a list of particles, and 'splat' them into a volume map, using some kind of 3d filter for each particle (e.g. gaussian, step function, etc..)
Then the volume map could be used for rendering.
Alternatively the particles could be stored in their own data structure, which might be better for sparser sets, but that would be a bit trickier to code.
Does this means that a new generic Indigo particle Format is needed, that the 3D aplications have to export to (via new developed export Plugin)???

This would perhaps give you (Ono) the most flexibility for your interaction with this particle files.... ergo better and faster visual results.

The Drawback will be the developement time for each 3D application ...

well i think most, if not all, 3d programs used here have particles so i think its a matter of translating particle position/density and maybe even speed and direction into an Indigo particle format. or if particle density could create a 3d grid (maybe what tinman999 was talking about?). this is really cool.

Kram1032 · Post by **Kram1032** » Thu Jun 14, 2007 2:00 am

SketchUp don't have paritcles...

Post by **OnoSendai** » Thu Jun 14, 2007 2:01 am

Sketchup doesn't count

Deus · Post by **Deus** » Thu Jun 14, 2007 5:10 am

thats great. Ill need that

Kram1032 · Post by **Kram1032** » Thu Jun 14, 2007 5:11 am

Deus wrote:thats great. Ill need that

arneoog · Post by **arneoog** » Thu Jun 14, 2007 6:03 am

Kram1032 wrote:
Deus wrote:thats great. Ill need that

Agree

Post by **OnoSendai** » Thu Jun 21, 2007 6:07 am

Just a little update...
My scattering coefficient *was* off by an order of magnitude

I now have (approximate) Mie scattering for the sky as well as Rayleigh scattering.

Phase function for the clouds is still completely wrong

The camera viewpoint is raised 100m off the ground here, so you can see the shadows of the clouds.

Zom-B · Post by **Zom-B** » Thu Jun 21, 2007 6:19 am

I'm really interested about the final possibile parameters for the sky...
Could you drop some lines about the controllable parameters?!

hope to see some more pics these days ^^
(a clean cloud rendering [over night] would be sweeeeeeeet!)

Caronte · Post by **Caronte** » Thu Jun 21, 2007 6:22 am

Looks Better and better

The shader is very nice

when the clouds have a better shape, the results will be awesome

Thanks!

Experiments..

Who is online