Moving clouds ?

Started by nixx, January 23, 2007, 07:36:50 AM

Previous topic - Next topic

nixx

Hi,

Is there a way to move a layer of clouds ? I 've tried a few combos of vector/position nodes but nothing seems to work. Maybe it's just me (quite probable ;)) but I can't get it to work.
Say you have a cloud layer you 're satisfied with, but you need to slightly offset the clouds' position (or, in the animation-enabled version, maybe even animate the clouds drifting & billowing). How would you go about it ?

thanks,
nick
I 'm child, and man, then child again; the boy never gets older

dhavalmistry

I think wraping the clouds should work
"His blood-terragen level is 99.99%...he is definitely drunk on Terragen!"

nixx

I assume you mean "warping" ;) (if not, please correct me !)

That could be a possible workaround, yes, but what about straight-line motion ? Think time-lapse movies of low-altitude clouds being carried away by winds...

Any ideas ? I 'll keep looking into it myself, but if anyone has any suggestions I 'd love to hear them :)

thanks,
nick
I 'm child, and man, then child again; the boy never gets older

dhavalmistry

I am not sure what you are trying to say but you can animate the "Lead-in wrap effect" for you movement and probably animate buoyancy and clumping for you billowing effect. I am still trying to figure out what the whole "noise variation" part does but as far as I've tested those settings....they should work for you
"His blood-terragen level is 99.99%...he is definitely drunk on Terragen!"

buchvecny

yeah i think i know what u mean... and no i dont think its possible right now

nixx

Quote from: dhavalmistry on January 23, 2007, 10:58:56 AM
I am not sure what you are trying to say but you can animate the "Lead-in wrap effect" for you movement and probably animate buoyancy and clumping for you billowing effect. I am still trying to figure out what the whole "noise variation" part does but as far as I've tested those settings....they should work for you

Animating the lead-in warp does produce a certain kind of motion, but again, it's not exactly what I had in mind. I was talking about moving the cloud layer towards any direction you choose, which I still haven't figured out how to do (if at all possible). But your suggestion provides a limited workaround, at least until a better way to do it comes along.

Quote from: buchvecny on January 23, 2007, 11:16:40 AM
yeah i think i know what u mean... and no i dont think its possible right now

Well, like I said, I 'll keep looking into it, and if I find anything I 'll post it here. (unfortunately I still only have the free version, so no animation for me - so I use the "tweak-and-watch-the-patterns-change" method ;))

nick
I 'm child, and man, then child again; the boy never gets older

nixx

I can't believe it was that simple, and yet I missed it !

There's a "Transform Shader" there :) Just plug the cloud's density shader into a Transform shader, and then go ahead and adjust the XYZ translation and scale !

I think the reason I overlooked this was that I was looking for a solution in the function shaders, while this is located under the "other shaders" menu.

Now all I need is the commercial T2TP (and lots of patience for rendering) in order to do some "real" tests !

nick
I 'm child, and man, then child again; the boy never gets older

skyasay

Thanks for finding the transform node :)

I am having luck animating time lapse coulds using the transform node along with animating the noise octives.

Njen

While animating the transform would be a nice way of moving the clouds, what we need is some sort of 'phase' value, to make them change their pattern slowly without compromising on the valules that make up how it should look like.

Oshyan

Correct, the Transform shader (specifically the Translate function within it) is the best way to "move" a layer of clouds currently. Translate along the X or Z axis (or both) to get planar motion. If you translate along Y, rather than get real vertical motion, you will get something more like shape variation. The cloud plane essentially samples the noise function within an "altitude" range and translating along Y moves the noise shader relative to this sampled area. As the shader moves into and out of the sampled area parts of the cloud will appear to slowly grow and shrink, in a surprisingly realistic manner. Worth a try for the effect you're aiming for. A more specific "cloud morph" function for realistic cloud evolution would be nice as well, but it may not be something we can implement for the initial 2.0 release.

- Oshyan

nixx

Thanks for the detailed description Oshyan. So in other words, you 're not moving the clouds themselves, you 're just moving the "pattern" that makes them up. It's no different than the way procedural patterns behave in most 3D apps, really.
I 'm obviously just speaking for myself here, but I don't think a "cloud morph" function would be that important - the described technique works surprisingly well (judging from other apps).

nick
I 'm child, and man, then child again; the boy never gets older

Oshyan

Yes, that's correct. True, realistic cloud motion would take some more work and possibly a dedicated function or shader, but you're right that this simple approach yields surprisingly good results.

- Oshyan

stevehmeyer

I am a meteorologist and have a passion for 3D (volumetric) clouds and am always looking for a better way to animate clouds. For about 5 years I have been using Lightwave with a plugin called Ogo-Taiki. Like TG2 this plugin uses procedural textures and cannot specify individual clouds just a sky populated by clouds the number of, density of and form of which can be controlled. Like the Technology Preview of TG2, Ogo-Taiki requires long render times but that's the cost of realism.


I divide cloud animation into three parts. 1. large scale motion of clouds in the atmosphere i.e. clouds follow the winds, 2. growth/decay - evolution of individual cloud elements i.e. is condensation or evaporation dominant. This could be expressed as vertical growth in cumulonimbus (thunderstorm) clouds or increasing density in stratiform clouds or a combination of these or other factors you may come up with. 3. Small scale variations, these I treat as random variations that have smooth transitions from frame to frame. In other words I do not want things just popping into and out of existence for a single frame. They are fine detail which follow from the physical processes that cause the clouds. Examples are the cauliflower-like bulges on a cumulus cloud or thin patches in a deck of stratus clouds, or edge detail (wispy vs. hard).  At the landscape scale we do not worry about physical reality of these small small variations, they are aesthetic detail and add realism.

In TG2 a transform shader is great for #1 (in Lightwave an object is moved and the clouds follow the motion of the object to which the texture is assigned).

I am new to TG2 and because I could not do what I wanted with clouds in TG I never really used it much.  So those of you who know TG2 much better than I do (almost everyone!) try to divide your cloud animations into three parts and deal with each part separately. It really does simplify the process. Move the clouds with the translate function of a transform shader, and come up with techniques on increasing vertical extent and/or density for item #2 and come up with ways to manipulate procedural textures for fine detail in item #3.

Here is hoping that TG2 will some day allow a volume, defined by a 3D object to be filled with a 3D procedural texture to create an individual 3D cloud. In Lightwave this seems to be an easier way to get individual clouds than a particle system.

While I learn TG2, anyone have any ideas?  By the way the node network of TG2 is great!

-Steve Horstmeyer


Cyber-Angel

You know as well as I do that cloud movement, formation and dissipation is a highly complex process involving thermodynamics, Gas and Hydrodynamics and the Navier-Stokes Equations among other things. In its current form the Atmosphere model in TG2 is no where near sophisticated enough to do cloud motion in a physically correct, photorealistic manor and you would have to rewrite it from scratch again and that would take vast amounts of time.

In its current form Terragen dose not have an atmospheric circulation model that I can see for air currents and air circulation, also while it has a sun or multiple suns these play no part in the formation of clouds within Terragen and there is no evaporation model between the sun/s and water bodies at this time.

Real clouds have complex movements of air inside them which are a major source of cloud movement as well as temperature gradients within the cloud and relative wind speed and direction.

Other things' missing for the model and this is by no means complete:   

•   Advection (Follows Isobaric Surfaces, Predominantly Horizontal)
•   Air Density
•   Air Mass
•   Cloud Albedo (Variable from less than 10% to more than 90%: Dependant on droplet sizes, liquid water or ice content and sun zenith angle).
•    Cloud Dispersion
•   Cloud formations expand and react to other clouds
•   Cloud Optical Thickness
•   Cloud Particle Sedimentation
•   Cloud Top Pressure
•   Cloud Top Temperate
•   Cloud Translucency
•   Coagulation
•   Coalescence
•   Cold and Warm Fronts
•   Collision
•   Condensation Level
•   Condensation Nuclei
•   Condensation Rate
•   Convection
•   Convergence
•   Degree of Saturation (Rising Air)
•   Dew-Point
•   Differential Light Scattering
•   Droplet Size Distribution  (Maximum Drop Size 5.5mm)
•   Dry Adiabatic Lapse Rate (Consistent Value at 1°C/100 Meters)
•   Dry Air
•   Dynamical Cooling/ Heating
•   Eddy Updraft
•   Equilibrium Point
•   Equivalent Potential Temperature (Also called: Theta-e)
•   Evaporation Rate
•   Geometrical Optics
•   Humidity
•   Latent Heat
•   Lifting Along  Frontal Zones
•   Liquid Water Path
•   Orographic Lifting
•   Potential Temperature
•   Relative Humidity
•   Saturation Vapour Pressure
•   Solute Effect
•   Supper Saturation
•   Vapour Partial Pressure

There would be alot of work requiered to do this so that clouds behaved correctly durring an animation in a cnvincing mannor.

Regards to you.

Cyber-Angel



old_blaggard

That's a pretty comprehensive-looking list you've got there.  However, I have to disagree with you on the point that all of that is necessary for realistic cloud animation.  If your goal is to simulate cloud patterns exactly, then it's definitely good to have all of those parameters, but then we would have to get into chaos theory and the movements of the individual molecules - and as far as I understand chaos theory, you can't calculate an exact outcome, which will make any kind of simulation slightly inexact.  I think that for hobbyist and even studio-grade work, a competent implementation of the methods described by Steve would be adequate for results that look realistic (even if they actually aren't).
http://www.terragen.org - A great Terragen resource with models, contests, galleries, and forums.