PDA

View Full Version : The weather procedural



prometheus
01-09-2020, 08:42 PM
Thankīs to denis pontoinner for his rman textures, in this case ..rock and weather procedural, the weather procedural has so many scale and other settings, it may be a bit difficult to get good results at once, you need to spend a lot of time with it, once you know it better..it has the potential to be better than turbulence for most clouds.

Iīts just unfinished half rendered screenshots, just wanted to showcase the weather procedural a bit..
the larger terrain, just a quick mockup from Gaea, only have the demoversion so that is only a 1024x1024 texture that is displaced, can not get any higher output than that with the demo, a bit over a millions in polys when subpatched some artifacts partly due to not having tripled it, and partly some settings in gaea, material is not pbr but standard material, some slight crackle bump as well..but not good enough since it should have the falloff fixed to only work on the higher rocky parts.

Anyway, the rest is the weather procedurals on 4 nulls for that larger gaea terrain.
for the other I think only one single large null and the weather procedural, and that single rock in the middle is just pure lw fractals.

the gaea landscape I didnīt wait for it to iterate refine enough before taking the screenshot, thus the clouds have too much grain in them, these images may go nowhere I just wanted to test the weather procedural with some rock and landscape approach, for gaea terrains
I would need to invest in the full version.(maybe) and the other pillar type with only procedurals is ongoing RND from time to time, that will be combined with blender sculpting later, and much more advanced rocks, the procedurals for that is two of a kind, namely the rman rock..main rock structure, and the native crackle procedural for finer cracks.

The newer sunlight type is used for more accuracy when the sun goes down a bit and color the clouds, and the new physical sky is used as well, radiosity and indirect sampling for it to work on clouds as well.
Never mind what seem to be a water plane, I just made it reflective, and didnīt focus on that to be ocean or sea, will adress that later perhaps.
some more rock types of different shape, and then instance them etc..together with better water, and it may be nice.
Clouds are roughly at large scale near real life I would say.

https://forums.newtek.com/attachment.php?attachmentid=146574&d=1578623618

https://forums.newtek.com/attachment.php?attachmentid=146575&d=1578623796

146574

146575

prometheus
01-09-2020, 09:02 PM
Accidently saved out from acdsee..which produced a horrible banding.

this one is better..

To further note, I also use a curve gradient in between the actual texture fractalto the texture input, as well as another curve gradient between the actual texture fractal and the texture density, this softens up the are of the edges of the cloud, but with a curve control to maintain the inner volume a bit more denser.

https://forums.newtek.com/attachment.php?attachmentid=146576&d=1578625340
146576

Tim Parsons
01-09-2020, 09:35 PM
Wow this is really cool!

erikals
01-09-2020, 10:20 PM
Wow this is really cool!
Yepzi!  https://i.imgur.com/bcwLfNX.gif
Prometheus knows clouds!  https://i.imgur.com/uVY2fVj.gif

gdkeast
01-10-2020, 02:41 AM
You wouldn't mind sharing your weather procedural settings, would you?

prometheus
01-10-2020, 03:05 AM
You wouldn't mind sharing your weather procedural settings, would you?

Maybe..but only part of it, I may need to keep some of it for tutorials later.
And to smoothly shy away from your request in a fine way, I need to go to bed now, got some more interesting procedural images to show tomorrow, and Iīll think about what to share etc.

prometheus
01-10-2020, 03:47 AM
Eh...Iīm supposed o sleep now, but sometimes these tweakings keep me up for hours.

This sample, not the weather, but the gardner clouds, in this case added with the procedural (multinode) which I actually isnīt that fond of, itīs deepest function parameters is two window modul popup required from that node, and these settings are not exposed and accessable as node inputs becuase of that, I think it is better to simply add the nodes you want and reroute them, enough about that now.

The gardner clouds is here inverted, and in the second image, pretty much the same main procedural gardner clouds, but I copied it but did not invert it..and then fed the output to the first gardner clouds rotation node, and we know by that it will transform the procedural (distort it)..so that way we can get more whisp and undulations, you can choose many other procedurals to do this as well, like the weather procedural, or turbulent noise etc...this additional procedural that goes in to the rotation input, can be controlled in strenght by just lower or increase opacity, you can also go in there and change scale and a lot of other parameters.

In this case, no emission value..(only use it if you absolutely have to fake illumination) and in this case, no radiosity in the scene..quicker that way when I just need to showcase, and in this kind of cloud type, multiple scattering is less appearant anyway.

A smooth step function with some adjustment helps getting control of how much of the volume should be dense or filled with the texture, so that can clip texture density off, or more fuller cloud over the sky, you can of course use curve gradient also, but they can be harder to control.

Only one larger primitive item in volume mode for these two latest images.

Settings for the weather procedural have to look for it tomorrow.

And as usual, thankīs Denis P for your Rman Collection.

https://forums.newtek.com/attachment.php?attachmentid=146581&d=1578649404
https://forums.newtek.com/attachment.php?attachmentid=146582&d=1578649434


146581

146582

fishhead
01-10-2020, 04:16 AM
These look really, really nice, Prom! :-) Thanks for sharing!
And now: Sleep well, you earned it! ;-)
Cheers

JohnMarchant
01-10-2020, 05:39 AM
Very nice Prometheus, hope you will share your scene file settings with us.

ary3d
01-10-2020, 09:11 AM
Beautiful

prometheus
01-10-2020, 06:03 PM
Beautiful

Thankīs David.

Any plans on upgrading from 2015 to 2019..so you get access to the new more realistic volumetrics?
sometimes backdrop imagery is of course to be prefered if you can match lighting with the scene etc, but othertimes you may want the exact match of
the light acting the same on clouds as on the landscape with full raytraced casted shadows, it of course comes at a rendertime cost.

prometheus
01-10-2020, 06:30 PM
I wonīt share the scene, Only the node for the weather procedural from the first image, but the clouds was 3 primitive items..pretty much the same, so youīll get only the procedural node setup, and you have to feed the multiply node in to the scattering channel, and the alpha output to the texture input (weird that the full tree can not be exported..missing the main output node)

But it takes a lot more to get it to look like this, first off, you need to set the radius of the primitive to 100 m, also scale the cloud ref to 225.340, this will not scale up the cloud size..only the texture size of the procedural, you would have to connect it yourself also to the setup in the primitive volume.
The actual texture is also a little tweaked in scaling x and y 1,5 and z 1,7
Texture amplitude for the voxel is 2,3

So donīt expect to get exactly what I got there, position of the null is also not described, the rest is a learning process ..so for whatīs it worth, if you manage to get something decent you can always post here and I could probably tell where it can be improved.
The cloud lives at the height of 119 meters, the actual cloud volume is also scaled to..
x 2,826
y 0,350
z 6,026

ary3d
01-10-2020, 08:45 PM
Thankīs David.

Any plans on upgrading from 2015 to 2019..so you get access to the new more realistic volumetrics?
sometimes backdrop imagery is of course to be prefered if you can match lighting with the scene etc, but othertimes you may want the exact match of
the light acting the same on clouds as on the landscape with full raytraced casted shadows, it of course comes at a rendertime cost.

I have upgraded from 11.6 to 2018 some time ago, soon I will upgrade to 2019, I have used the new volumetric system in my last render:
https://www.artstation.com/artwork/qAm8Xn
I have doing some experiment with the volumetrics also, some low clouds or morning myst:

https://forums.newtek.com/attachment.php?attachmentid=146587&d=1578710645

146587

prometheus
01-10-2020, 09:15 PM
I have upgraded from 11.6 to 2018 some time ago, soon I will upgrade to 2019, I have used the new volumetric system in my last render:
https://www.artstation.com/artwork/qAm8Xn
I have doing some experiment with the volumetrics also, some low clouds or morning myst:

https://forums.newtek.com/attachment.php?attachmentid=146587&d=1578710645

146587

Great, and I recommend you upgrade ASAP, some bugfixing that was crucial I think, some minor physical sky improvements..though I would say octanes sky look much better...at least in blender which I can see.

Yes, nice image..the new volumetrics is great for that kind of either low clouds and mist i think, may be even easier to manage than in vue for instance.
Still hope they could manage to implement yet another volumetric global scattering handler, and this time a full spectral one that physicly is much better for a sky than the current volumetric global scattering, after that we only mostly need infinite volumetric cloud planes as well, then they could start to focus on displacements and terrain fractals.

By the way, the Rman terrain node may be of interest, Denis P did implement that not long ago, it has some interesting patterns if using jordan and suiss mode, though it has to be together with some other main typical terrain fractals such as FBM and Ridged multifractal.

I will have to showcase some samples of that later as well.

gdkeast
01-11-2020, 03:14 PM
This is very helpful and thank you for sharing. I always try to figure things out myself, but sometimes I just need some starting points. Thanks again!!!

Bernie2Strokes
01-12-2020, 07:43 AM
146600

Hi. Months ago, I mentioned how I wanted to do type of dark cloud cover coming from the horizon while it was still bright in the opposite direction. This photo I took today is somewhat an example of what I'd like to achieve some day.

BTW: there's ashfall outside my house. But since it's already night time I can't take any decent photos of that.

prometheus
01-12-2020, 01:12 PM
146600

Hi. Months ago, I mentioned how I wanted to do type of dark cloud cover coming from the horizon while it was still bright in the opposite direction. This photo I took today is somewhat an example of what I'd like to achieve some day.

BTW: there's ashfall outside my house. But since it's already night time I can't take any decent photos of that.

You mean you need it to be animated and move in? perfectly doable, depends on how fast, and what kind of general motion it should have.
That kind of dark cloud you have in the image, has very little texture to it, just some small density variation, and soft edges...that isnīt hard to do at all, the bright side is just a matte of setting asymmetry the right way, and of course having the right sun angle for it, then itīs a matter of adjusting scattering color and perhaps absorbtion color..I may look at how this could be setup later today.

prometheus
01-12-2020, 05:14 PM
A little godray test I did mid december.

Canīt recall what procedural I used here, have to check the scene first which I havenīt right now, just a sample of some slight sunbeam, godray shining through, could be enhanced more by post, or tweaking the cloud density, sharpness and some more in the actual global scattering.
Cloud movement can be made with either null reference and moving it, or use dpontīs speed time node and feed it in to a vector node and choose directions, both wind direction and a slight up y direction for cloud change if necessary.

Sorry for the clip being to short, and I only did smaller resolution of the render 1123x478, so it was scaled up by youtube and thus loosing a bit of detail and sharpness.

Youtube description..
"
This is the lightwave volumetric primitive in volume mode, and in conjuction with the global scattering volumetrics, it is tricky to set it up depending on where the clouds are, how large they are, and position of clouds, tricky in the sense that it doesnīt behave by defualt such as vue or terragen for this effect, you have to adjust volumetric distance per scene case, and itīs harder to get good quality when rays are close to the clouds.

Cloud movement can be done with a null reference, or the speed time node from dpont fed in to position of wind direction and also a slight up-y direction.
Physical sky environment, and the newer hosek wilkie sunlight model.
Very short and subtle effect of the sunbeams, I will make longer and better ones later on."

And ..now youtube is enforcing harder child protection privacy, so you have to set vids to be non appropiate for children, if you want your vids to have comments active.


https://youtu.be/7x4oAzv7ZxA

prometheus
01-12-2020, 08:24 PM
146600

Hi. Months ago, I mentioned how I wanted to do type of dark cloud cover coming from the horizon while it was still bright in the opposite direction. This photo I took today is somewhat an example of what I'd like to achieve some day.

BTW: there's ashfall outside my house. But since it's already night time I can't take any decent photos of that.

I took a look at this in Lighwave, though due to the texture of the cloud, it is a bit boring to mess around with, so my isnpiration for that type isnīt the best...the tricky part is to get all lighting temperatures, and scattering colors right, itīs in so many parameters, from the hosek wilkie light temperature, to itīs scattering color, and turbidity, then you got the actual cloud scattering, and then you got the physical sky and itīs scattering, there is also softer background cloud whispīs in white that blocks part of the sun, and also affects this cloud so it is a bit brighter at the side top than below..so a part of that effect I think is due to the secondary dispersion clouds as I call it, then you need to set the right assymetry, and the right scattering amount and light angle..and then you need to balance it with the proper amount of shadow intensity.

So trying to match a photo exactly in color or mood is the hardest part, the clouds texture, movement and density is of lesser difficulty.

prometheus
01-13-2020, 03:20 PM
And fog item


just added a primitive item in volume mode, with very small density etc, you can see the settings, this allows for making haze and scatter ligthcolor in a more spectral way sort of, while not physical correct in any means, it yields a more natural look to the sky and obscuring clouds with some hase/fog.

I have yet to find out how to work with global scattering this way, if possible..some more advanced nodes and math for extracting lenght and depth etc I guess, but that is a path I am reluctant to go since I want to work more artisticly than scientificly..when I can do so.

Still hoping they either rewrite the global scattering system, or implement another drop down handler of a scattering system more spectral physics, and more proper controls for height and depth.

When adding these extra volumetrics, on top of other volume items, rendertimes can go sky high, in this case 52 minutes ( with a lot of downloads going on in the background) and I should have increased step size, cause it wasnīt needed to have so low step size I did here, so I could crunch the time down to mayb 35 minutes...but 45-50 minutes is probably what you can expect for terragen and vue as well with this type of atmospherics, depending on your hardware also of course, but in general the 3D software equals up in time. Octane is a different matter though.

Take note of color change in scattering channel and those values, a slight bluish scattering results in a more redish pink and green tone when entering the fog volume which is located near ground, I like how it scatters from deepblue to cyanblue to bluegreen yellow, then orange almost pink, this is also what I often see when sun is almost going down on clear days without too much turbidity etc, there are so many different conditions though depending on temperatur in the air, amount of water droplets, time of year etc.

Never mind the little noise here and there, I could adress that..just testing and didnīt want to spend time too much on full quality.


https://forums.newtek.com/attachment.php?attachmentid=146612&d=1578950065
https://forums.newtek.com/attachment.php?attachmentid=146613&d=1578950080


146612

146613

erikals
01-13-2020, 11:03 PM
And ..now youtube is enforcing harder child protection privacy, so you have to set vids to be non appropriate for children, if you want your vids to have comments active.
not as far as i know.
but Youtube sometimes has problems showing comments, a bug they need to fix.
they might also have started to remove comments with links.

it is a bit of a hassle, but better child protection is good in the long run.


i just don't like the upload page 1-2-3 thing they got going. very inefficient.

prometheus
01-13-2020, 11:37 PM
not as far as i know.
but Youtube sometimes has problems showing comments, a bug they need to fix.
they might also have started to remove comments with links.

it is a bit of a hassle, but better child protection is good in the long run.


i just don't like the upload page 1-2-3 thing they got going. very inefficient.

Well, let me describe it like this, when I upload a video and go to settings, the ability to use comments isnīt active, checked...and I get an error when trying to choose it because of some setting, the vid was marked as appropiate for children, so I changed that..and voila..the comment field became choosable, and to further put that in perspective..I think there was a statement in the new rules for youtube, that comment sections are not allowed for a video marked as appropiate for children, so if you choose that option, you seem not able to have it commentable as described.

erikals
01-13-2020, 11:47 PM
ok, hm, seems alright with these settings at the westcoast
https://i.imgur.com/yTi6rAV.png

prometheus
01-14-2020, 12:23 AM
ok, hm, seems alright with these settings at the westcoast
https://i.imgur.com/yTi6rAV.png

that screenshot doesnīt say anything about the commenting section, all you got there is the option to set it as inappropiate for children, and you also have the description field which is a different matter.
somwhere there you must have the setting where you choose to allow comments, maybe in the next step? check more options.. and if you have choosen not appropiate for children, you should not be able to choose allow comments either.

erikals
01-14-2020, 12:56 AM
you have a global setting, i don't touch that one, other than that i don't know.
anyways, works ok, if you use the right settings. tho not sure what those "right" settings are. Youtube doesn't like to play easy.


and if you have choosen not appropiate for children, you should not be able to choose allow comments either.
works here.  https://i.imgur.com/B8QSFA7.gif

note tho that kids >are< allowed to watch it, since i activated this last setting >
https://i.imgur.com/SJyNDSd.png

Bernie2Strokes
01-14-2020, 08:19 PM
And fog item


just added a primitive item in volume mode, with very small density etc, you can see the settings, this allows for making haze and scatter ligthcolor in a more spectral way sort of, while not physical correct in any means, it yields a more natural look to the sky and obscuring clouds with some hase/fog.

I have yet to find out how to work with global scattering this way, if possible..some more advanced nodes and math for extracting lenght and depth etc I guess, but that is a path I am reluctant to go since I want to work more artisticly than scientificly..when I can do so.

Still hoping they either rewrite the global scattering system, or implement another drop down handler of a scattering system more spectral physics, and more proper controls for height and depth.

When adding these extra volumetrics, on top of other volume items, rendertimes can go sky high, in this case 52 minutes ( with a lot of downloads going on in the background) and I should have increased step size, cause it wasnīt needed to have so low step size I did here, so I could crunch the time down to mayb 35 minutes...but 45-50 minutes is probably what you can expect for terragen and vue as well with this type of atmospherics, depending on your hardware also of course, but in general the 3D software equals up in time. Octane is a different matter though.

Take note of color change in scattering channel and those values, a slight bluish scattering results in a more redish pink and green tone when entering the fog volume which is located near ground, I like how it scatters from deepblue to cyanblue to bluegreen yellow, then orange almost pink, this is also what I often see when sun is almost going down on clear days without too much turbidity etc, there are so many different conditions though depending on temperatur in the air, amount of water droplets, time of year etc.

Never mind the little noise here and there, I could adress that..just testing and didnīt want to spend time too much on full quality.



That's really amazing. So far I have no plans to animate that kind of atmosphere, it was just a look that I dreamed of simulating.

Also I suspect what I photographed was actually ashfall from the recent eruption of Mt. Taal. I hope that didn't confound your tests.

prometheus
01-14-2020, 11:36 PM
That's really amazing. So far I have no plans to animate that kind of atmosphere, it was just a look that I dreamed of simulating.

Also I suspect what I photographed was actually ashfall from the recent eruption of Mt. Taal. I hope that didn't confound your tests.

Yeah...sorry to hear about that vulcanic ashfall, hope you are in a safe place?

I suspected that it is some kind of dust cloud with particles from earth rather than water droplet particles.

Regarding setting up an atmosphere this way with fog/haze layers to scatter the light in various colors etc, either you go the full global scattering or by scaled volume item layers....adding that effect is changing the atmosphere immensly with
all that advanced scattering, much much more realistic and color when you can get that light and color scattering, it spreads the light also depending on assymetry settings to me much more realistic in the sky when the weather conditions is a particular way.

Comes at a cost though, longer rendertimes if you want decent quality, and which method to use, the global scattering has no quality control per say, and you would need to do the trick with nodal setups to control height and depth I think, which I still do not know how to do.
And the volume item node with scalable layers, will pose issues of scaling properly and position it properly...a single still may work quite easy, but for a scene you would need to pan or move camera through, there will be problems.