Procedural clouds

[Jiri.Matys]

Photos generally have some kind of tone mapping applied.

Best regards,
Vlado

Sure, but even without tone mapping look different. My eyes see them different. Without color correction, LUT or other types of post vray sky just looks different.

[^Lele^]

Sure, but even without tone mapping look different. My eyes see them different. Without color correction, LUT or other types of post vray sky just looks different.

Well the sky model is semi-empirical (i.e. measured).
It still won't be exactly as the one you see outside of the window (i believe that was measured in L.A., so west coast US facing the pacific. Very specific colours.).

Do you have a reference from some other sky model, and pictures showing non-graded skies (say, a camera raw, instead of a post-processed jpg)?

p.s.:it's being suggested that you grade the v-ray sky as well. It's hardly possible to get *all* color grading off a camera, even with Raw images.

[Paul Oblomov]

I'd assume MIE comes into play for water droplets.
Rayleigh is the one that gives sky its blue color, but the interplay happens between light and particles about the size of the wavelenght.
Water vapour's bigger -> MIE -> greyish horizons.

Can you show samples of the look you don't like, Paul?

Oh, I assume, that phase function volumetric shader is called that way. Like famous hyperion cloud example...
Or maybe this database of measured functions
https://gitlab.com/hjhornbeck/vsf-database
Or just make OSL compatible with volumetrics. Sadly no progress in this direction for years makes my wanna cry.

[^Lele^]

Sorry, i couldn't find any mention of Rayleigh in your link or the Hyperion cloud sample (that's HG, a wholly different thing. And their work is essentially a-chromatic, with colour added in without proper spectral calculations.).

I doubt our clouds use a full-blown multi-bounce scattering solution, given how quick they are to converge.

We do have volumetric phase now, and for hero clouds, you should be able to get that working with phoenix (and/or the volumegrid) at some point in the near future.

[Paul Oblomov]

So sad, that Ive been waiting almost forever for this to come

[sirio76]

Sure, but even without tone mapping look different. My eyes see them different. Without color correction, LUT or other types of post vray sky just looks different.

Human eyes are one thing, photo cameras are a different thing, render engine linear output is yet another thing. Vray output is in linear because that's how light works (before devices, from camera to human eye perform adjustment).

[kosso_olli]

A thing that is related to V-Ray Sky: Right now there is no way to shift the horizon downwards. You can only shift it upwards with the offset parameter. Can the offset be made to accept negative values as well?

[JonasNöll]

A thing that is related to V-Ray Sky: Right now there is no way to shift the horizon downwards. You can only shift it upwards with the offset parameter. Can the offset be made to accept negative values as well?

Isn't it the oposite way? There is no way to shift it upwards, only downwards at the moment.

[kosso_olli]

Isn't it the oposite way? There is no way to shift it upwards, only downwards at the moment.

Yes, you're right. Sorry.
Anyway, it would be great if we could shift to both directions.

[kosso_olli]

Tone mapping? Here is my results using default sun sky, highlight compression, filmic tone mapping, LUT:

I am really curious how exactly you achieved this. Getting the sky to look like that is something I never got right. Does the LUT play a big role?

[^Lele^]

See attached.
All it is is a contrast operation, however one achieves it.

[^Lele^]

The new cloud system is really awesome! Really makes using static HDRIs in many cases obsolete now!
2 ideas that came in my mind though:
- Is it possible to get a mask that matches the current render view? Like a VRayCloudMask renderelement or so?
- Is it possible to have a spherical cloud mask (generated during rendertime) so that I could pipe the whole VRaySky into a composite and then use that mask to only modify the skycolor without affecting the clouds or composing other elements into the sky behind the clouds directly in 3dsMax?

Counter-EDIT: It can indeed be masked out just fine using a merge expression node in nuke, in this case removing pixels which are not *exactly* identical.
Feathering can ofc be added in.
However (and that may also be the case if the mask was provided in advance), there is a lot of faint cloud work on the sky behind, so the mask may not be as one expects it, and making changes to one may still reflect on the rest in non-pleasant ways.
I'll be curious to see how the proper mask will look.

[kosso_olli]

Oh. That is quite simple indeed. Thanks for leading me into the right direction.

[sirio76]

I am really curious how exactly you achieved this. Getting the sky to look like that is something I never got right. Does the LUT play a big role?

Here is an example on another scene, since the tone mapping affects also every other material you need to build them using the same corrections.
In this case ther's also a slight camera offset so the vignetting on top is a bit too much, physical camera ISO100/F16/SS100, default sun/sky.

[kosso_olli]

Thank you very much!