PDA

View Full Version : Cloud Time Lapse


ruchitinfushion
05-22-2010, 09:14 PM
Tell me how to get proper look of "Cloud Time Lapse".
I made one test with ready preset from visor & added time multiplier expression to texture origin but didn't get smooth rhythm.Is there any more good idea for this??
Here is my reference:http://www.vimeo.com/1156338
______________________________________________________________________________
Ruchit Bhatt.

arkangel-fx
05-23-2010, 06:18 AM
try a fluid, with lots of swirl, emit from a large volume and set the simulation rate scale to .2 or .3, that should do the trick. Maybe you need to increase the substeps.

ruchitinfushion
05-23-2010, 05:09 PM
Hey Bro ,thnx for reply but still hard to get this...@ mid rez i m not getting texture sharpness & motion is not good @ all.
Test Scene:CloudTL1 (http://sites.google.com/site/researchfluid/r-d/CloudTimeLapse1.ma?attredirects=0&d=1)
http://i410.photobucket.com/albums/pp186/ruchitinfushion/Cloud_test1.png?t=1274630699

Duncan
05-26-2010, 12:51 AM
You are going to have a difficult time trying to simulate that, especially at a good resolution. (not impossible, I think, just a lot of work)
You might try a purely textural solution. Start with one of the cloud example scenes and animate the texture time and offset. In particular an offset moving upward in Y can look like cloud formation. In some ways this can look better than the dynamic approach in that with real clouds the appearance is somewhat as if the density of the cloud remains in a fixed position while a texture moves up through it. I think this is because while the air is moving upwards the altitude of condensation stays roughly the same, so the cloud stays in roughly the same position even though the air seems to flow through it. (I'm not sure, though... perhaps travelling pressure waves also have an effect)

Duncan

HowardM
05-26-2010, 11:45 AM
im confused...as to why this would be hard?

his ref video looks like slow moving thick fluids, like many people have done here... like either volcanic smoke or a large oil fire smoke.. just moving slowly...

ruchitinfushion
05-26-2010, 03:18 PM
Here is my second test just added time*integer to texture Time & Yaxis origin.
CloudTimeLapse (http://sites.google.com/site/researchfluid/r-d/CloudTimeLapse.ma?attredirects=0&d=1)
But not satisfy with motion of moving cloud.
@Duncan: can u plz chck my attached file & tell me how to fix this.
@HowardM:Just try it ,and feel it ..ha ha great huh.

Duncan
05-26-2010, 05:46 PM
im confused...as to why this would be hard?
his ref video looks like slow moving thick fluids, like many people have done here... like either volcanic smoke or a large oil fire smoke.. just moving slowly...

One could probably get the rough look pretty easily. But one of the harder things to simulate is a really convincing cumulus cloud formation with updrafts. I've never yet seen a simulation that would fully convince me(perhaps it would make a good FX challenge). There is a quality of the internal fluid billowing outward from inside the volume as is rises that is subtle and complex. It is not simply pushing around of cloudy density but rather formation of condensation due to pressure drop with altitude and several other effects. Add to this that the directional light scattering and coloration effects are very important to the beauty of a cumulus cloud.

I suppose I'm only familiar with the cloud formation behavior thanks to time lapse photography(it's not something one sees in common experience), so perhaps people who have watched less of this would be more easily convinced in terms of the motion( the lighting is another story )

Duncan

Duncan
05-26-2010, 06:33 PM
But not satisfy with motion of moving cloud.
@Duncan: can u plz chck my attached file & tell me how to fix this.

I think to match the reference more closely you will need to both have a subtle texture animation AND have a dynamic density. Fortunately there is an aspect of the cloud behavior that is a bit like a density moving through a texture that is slowly animating downwards.

Make the density and velocity grids dynamic on your scene and make the opacity input= density instead of constant. You can either paint the initial state of the clouds or emit into them and set initial state from current. Given that you are painting the cloud density you may wish the opacity gradient ramp to shift to the left a bit. In addition to the initial state you may also wish to have several emitters for the clouds, perhaps with density dissipation.

Initially I would try and just do a simple fluid animation... a few blobs of density moving slowly and shearing a bit with most of the detail and motion in the texture. (you might not want any density buoyancy, perhaps just push the fluid with a weak force or have some speed on the emitter)

One could try a more complex simulation, for example one could have a large flat volume cube emitter at the bottom of the fluid with high turbulence on the emission so that it has a chunky emission, perhaps also emitting heat into the simulation. A problem I've found, however, is that simple rising heat makes the fluid become streaked upwards like flames. I think one wants more little localized upward puffs within the fluid. There is a lot going on so I'm not sure what is the most critical effect for the visual look of cloud formation. In addition to an emitter along the bottom of the fluid one might also want an emitter towards the top that sucks density out (use the replace method on the emitter with density = 0 and emission rate relative low with dropoff). One might also have a replace of temperature in a similar fashion on the same emiiter. This could simulate the effects of change in pressure and temperature with altitude.

If you go for a more complex simulation you may want more resolution on the fluid grid and less texturing. One can increase the texture frequency and lower the depth max, essentially replacing the low frequency iterations of the noise with the grid based fluid density. At a certain level of complexity one could do away with the texture altogether, although it will be slow to simulate and you might run out of memory.

The rendering should probably also use shadowing( currently the shadows on that fluid are faked with a y gradient incandescence ramp that is also textured ), but you probable don't need to turn on real lights. Ambient diffusion would also help. Mixing the clouds with a sky fog fluid can help to make them blend into the sky, instead of being in front of a sky texture, although it depends on your camera viewpoint.

Duncan

ruchitinfushion
05-27-2010, 09:17 AM
Last night i did two test
1.With Texture Time-in this i found unrealistic movement ......where density die .........they start to pop up again back in reverse direction @ same place.
2.With Texture Origin :-i think this is very very hard to understand for real application.....when i add time multiplier exp to this field...it start to shift up & spread out bit.
I think it will be better ,if i spend some time to clear real science fundamental behind this.
I have good link to study cloud
http://neo.dyn-o-saur.com/ref/clouds.html

ruchitinfushion
05-30-2010, 09:24 PM
Last Test
http://www.vimeo.com/12086462
Don't know why density is moving in zig zag.
@ Dunacn can u plz make good test ..will appreciate for that.
______________________________________________________
Born To Make History
Ruchit Bhatt

CGTalk Moderation
05-30-2010, 09:24 PM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.