new Hosek sky behaviour.

[super gnu]

is it normal for the new sky to go almost black at the horizon when the sun is low in the sky? looks pretty wierd to me.

[vlado]

That's probably due to the blending with the ground color, but from what I've seen on photos and in the real world it looks more or less like that. It is on the "to do" list to be able to control this. In the meantime, you can change the ground color to brighten it up.

Best regards,
Vlado

[super gnu]

to be honest i realised it does it with the old preetham model too.. you might be right about it blending with the ground colour..

i still feel a sunset or sunrise is beyond the vray sun/sky system.. ive treid with the new aerialperspective thing, but i cant get a very nice result.

if i have enough to haze out the distant parts of my model a bit ( circa 7km distant, would be a bit desaturated and grey by that distance) ) then it completely obliterates the sky, and is generally a bit of a pig to control.. i know i mentioned this before but some decent non-physically-correct min/max ranges, maybe a falloff curve, and some more control on blending with the sky would make it much more useful. sunsets and sunrises are quite "artistic" events after all

as it is ive only managed to use it to hide the horizon line of an infinite plane, with a view distance of 100 km..

added to that the glare/bloom doesnt work properly with the sun/sky or the intensity mask, so you cant put a glow on the sun. of course i can fix that in post, but still..

one day i hope we will have a full sun/sky system, with clouds, and even night-time, stars and the moon. i can wish!

- and Ozone is not nice to use, and feature-limited so you buy vue..

[^Lele^]

Sunset in Vancouver, looking West, circa 2012.

The dark silhouettes at the left and right of the picture are land masses, the middle bit is clouds.
Still, the darkening is quite apparent.
It's just we hardly ever get to see a pure horizon, normally...

[^Lele^]

The first links to a contact sheet with frames of a little sequence (linked afterwards as uncompressed QT) of the scene used for the introductory video to SP3 (scene courtesy of the amazing 3d department at chaos).

https://drive.google.com/file/d/0B1P...ew?usp=sharing

https://drive.google.com/open?id=0B1...0hfRGlNY0swM2c

Is there anything specific you can't achieve, and would like us to look at?
Perhaps it's a limitation of the technology in itself, perhaps something not clear in the UI, or docs...

[Neilg]

Your first shot there does have a black gradient creeping up into the sky haze that shouldnt be there, it becomes the darkest thing in the shot. it's only on that first couple shots though.

Actual horizon darkening is more like horizon desaturating, like this:



It goes grey, not black.

[^Lele^]

That's actually with the sun between 0.5 and 2 degrees above the horizon.
At the same height as your shots (although i ought to render a sequence with the sun well visible...) it behaves much like you see there.
I posted the picture of the sunset in Vancouver as the sun literally dove into the ocean, and was by the time i shot, BELOW the horizon (something we can't do quite yet).
In other words, an extremised version of the lower angles of hosek/preetham (which are both principled from empirical measurements, by the way.).

The fact it should go grey, instead of black, is entirely down to the color mapping between CCD and LWF render, exposure, moisture in the air (ie. it belongs to the Aerial Perspective, not the sky model, to desaturate. The sky model changes the hue at the atmosphere Edge, the Aerial Perspective changes the rays from thereon into the cg world), and so on (again, the Van shot was taken in a particularly clear day, with contrast at its best.).
In fact, even in your shots, behind the smog and humidity, the dark band (clouds always help it bits, but it's there on a clear day too) is quite visible.

[^Lele^]

This is a properly calibrated, custom-made, 6-stop HDRI reference taken in early September (again, very clear, dry day, with a northerly breeze.).
Unlucky i can't quite see the unoccluded horizon line from here, so most of the stuff happening after the sun goes below the horizon is lost (i do have the sequence going well into early evening, i just left it out as not really relevant.).
Notice the last image has the sun still a good few degrees above the horizon line (around 5: i am around 30 meters BELOW sea level.), and yet the darkening is quite visible already (compare the last image to the one right above it.).
I believe it's the planet's self-shadowing soft edge as the sun goes down, so unless there's something catching the shadow (a cloud, or heavy mist/fog), the darkening goes unnoticed.



The attached preview is small, grab the drive image below for the full size.
https://drive.google.com/file/d/0B1P...ew?usp=sharing

[super gnu]

The first links to a contact sheet with frames of a little sequence (linked afterwards as uncompressed QT) of the scene used for the introductory video to SP3 (scene courtesy of the amazing 3d department at chaos).

https://drive.google.com/file/d/0B1P...ew?usp=sharing

https://drive.google.com/open?id=0B1...0hfRGlNY0swM2c

Is there anything specific you can't achieve, and would like us to look at?
Perhaps it's a limitation of the technology in itself, perhaps something not clear in the UI, or docs...

hm, well if i could achieve the look present in that file id be very happy.. id love to know how they got the vrayaerialperspective to get the further reaches of that scene fogged out so nicely, without either a mismatch with the sky at the horizon (if not affecting env rays) or a completely fogged out sky (with affect env rays turned on)

unless of course the scene is massive and those domes are 60-70 km from the camera.


only other parameters are fog height, which needs to extend well into sky to avoid an obvious horizon line, and that scatter parameter. , which just looks odd if you turn it down.


i guess it also depends on sun position.. in my tests sun is in front of the camera.