[fulkrum]

Now we already have grass in Maxwell; why not a realistic phisical sky with a cloud editor and spectral lighting effects? It would be great for Architectural and outdoors renderings.
What do you think? Maybe for Maxwell V3?

[segnoprogetto]

+1

[Half Life]

I'm for any and all improvements to the Physical Sky, so +1

Best,
Jason.

[JTB]

+1 because I need it!

[Fernando Tella]

+1 Yep, yep.

[dariolanza]

Hello fulkrum,

Although it sounds nice, it is much more complicated than what it seems.

Rendering clouds will be very time consuming, as the volumetric light interaction with water vapor is complex.
This will make that, even if Maxwell could simulate them, the render calculations will be really slow.

In addition to that, not only the light interaction, but the clouds shape fractal is very complex itself. Just think about other softwares specialized in producing cloudscapes for years now, and those cloud fractals don't look very natural either.

In addition to that, the user interface will be very complicated, to host all the variables involved in the clouds formation (wind, vorticity, turbulence, air temperature, humidity, proximity to earth or sea, etc).

And finally, we can end with clouds that are slower to render, difficult to understand and set up, and that look CG or pretty similar among renders.

All of this can be skipped by placing a real sky picture (with your desired weather conditions) as background and let Maxwell to render just the models (faster, easier and better looking).

Cheers!

Dario Lanza

[Half Life]

All practical points, but there are counterpoints to that line of thinking as well.

Using HDR for sky is very limited because you only have data for a particular instant of time, in a particular location, with little to no control over any of the variables -- essentially putting the user at the mercy of the photographer.

Besides that, when was render-time ever a priority here -- I mean it is still impossible to render strong (eg. Physical Sky) dielectric caustics without massive rendering time/resources... but that doesn't stop dielectric caustics being presented as a valid feature of Maxwell.

As long as we have to use AGS or other poor gimmick trickery to render common things like water and glass in sunlight I can't see why people would not be equally accepting of the idea of massively resource consuming clouds as a valid feature.

I thought the name of this company was Next Limit -- this choice of name implies that pushing boundaries is what this company is meant to do... "playing it safe" and doing the practical thing is for other software companies. Here is the home of bleeding-edge and render-time be damned!

Best,
Jason.

[bograt]

KABOOM!!
I kind of know what Dario is getting at though, it seems like more of a thing for hdr light studio

[Fernando Tella]

Probably an integrated HDRI generator which would generate different kind of animatable skies straight in the scene would be more useful and easy (and will attract a lot of clients) than trying to mimic 3 dimensional clouds. I see no real advantage on the latter.

[AlexP]

I both agree and disagree with Dario. Right point is CG clouds always look CG, but today there is no efficient software able to generate such clouds in good way - both vue and terragen have slow rendering engines (vue one doesn't have gi on clouds at all, only "ambient light"). But people do make nice pictures in both, much better in terragen I think and some are hard to distinguish from photo.
Slowness is another story, simple clay scene lit by vue was about 10x slower then lit in maxwell keeping quality at nice level. Haven't tested that scene in tg, but I made some calculations and nice scene with really good quality settings, volumetrics etc. would take about 2 months to generate on my old 2600k CPU. I am talking about 20k x 10k resolution sky only here of course.
Advantages are obvious and one of them is possibility to animate clouds and animate lights on clouds, something hard to do in reality.
Moreover taking picture of sunlit clouds even with most flare resistive lenses always lead to bloom near sun and brightly lid clouds not talking about taking photo of sun itself.
Maybe NL team can tell how much faster in volumetrics is path tracing compared to classic rendering engines.

[fulkrum]

The negative ttitud from Next Limit staff surprised to me.
If I'm not wrong, Next Limit was the first in bring the non-unbiased rendering to us. And the first version of Maxwell Render was very slow.
I have not used Real Flow but I guess to simulate fluids should be also slow right? So, why not a realistic sky?
I think all people that use to rendering architectural exteriors or any kind of exterior scene would enjoy to have a functionality like that no mather if it takes a time.
As it had been said in the replies above, there are another programs like Vue that makes the duty, I have tried it and it is too slow even for a simple small scene and there are a lot of people that use it so, why not in Maxwell?
Anyway, I think most of the users that need to render a complex scene chose to use a service like The Ranch so the time for rendering is a relative therm.
So, don't say just no, let's analyze the possibility.

[gadzooks]

I have to agree with dariolanza on this. I would much rather see their efforts go towards getting caustics to clear up at a round sl 2 that at sl 32. Also be able to encase emitters into a glass object to simulate light realistically. That to me is way more important that clouds.

[joaomourao]

I must say that I tottally agree with Jason...
Physical Sky is a so powerfull but lacking clouds! We are talking about a software that reproduces reality! The unbiased way!
So, you guys that defend this way (like me!), how do you get your background clouds done!? Photoshop? Come on!!! Why not caustics in Photoshop then!?
You want some new "next limit"? This is one of them for sure...
I know Lumion is the opposite of Maxwell... they prefer "The fake way", but check out their cloud system and how it interacts with the scene...
Why not a channel for this or a kind of multilight thing
Controlling time in a render is a thing of the future, I know... but aren't you NL guys, people that came from the future?!?
Cheers!

[jojojoj]

really want to see which amount of time you'd have to wait, just to get a clear render with full volumetric lighting inside unbiased clouds.... if I remember well, things like caustics in cinema 4d never were rendered quickly, but creating volumetric effects for example with pyrocluster was a pain in the a.... compared to that and we're talking about quick renders for the caustics compared to what we'd have to wait for in Maxwell. I think we'll have to wait for some much much newer cpu gen to get things like clouds implemented in a usable way. Just my two cents.