New Sky/Atmosphere model in 4.24

Darn, forgot about that, I guess I figured it just be available out of the box oh well, TY o_0

Hi guys, for me is impossible to go in the space . I set the ground radius 600 but this not works. What’ wrong? I’m lost something?

Hi folks, another noob here suffering from my lack of a PhD in atmospheric physics! :wink:

Quick question: when i open up the Games blank template (), i see lens flares.

Yet i don’t see any camera in this level.

Where / how are these lens effects created and can i edit them?

Thanks!

Not sure if this is the correct place to ask, but is there a way to test the volumetric clouds meant to be released with 4.26 yet?

I’m curious to find out how they will look with an ocean water surface simulation plugin we’re using.

@Dazzer1234567 I do not understand what you mean. That level only has bloom I think unless it has changed. In any case, lens flare are unrelated to sky so I would suggest you start another thread. Thanks a lot.
@roberteker Another dedicated thread would have also been good for volumetric cloud :slight_smile: The new volumetric cloud component that will be available in 4.26 is available from our dev-rendering github.

Thanks for your work on the clouds. Is it supposed to be useable yet? I tried a build of that branch yesterday but the tooltip for the material slot says ‘It must be a CloudVolume domain material’ but I dont see that sort of domain listed in the material editor.

Hello @SteveElbows ,
On the next question please create a new thread for the cloud support separated form the this one which is about SkyAtmosphere component.
“CloudVolume domain” should be a “Volume domain”, I have fixed this now and it will be soon copied up, thanks!
VolumetrcCloud component It works in dev-render branch yes, documentation and example data will be coming at some point.

Hello,

Is it possible to link the fake color of Exponential Height Fog to the atmosphere, to derive it’s color from the actual atmosphere color?
Enabling **"**Support Atmosphere Affecting Height Fog" seems to do it only for the Volumetric part of it.

Hi @Maximum-Dev ,
Thanks for hte suggestion. That is not possible today: instead of that, what we have discussed about so far, is to have some controllable albedo for the height fog when it is linked to the SkyAtmosphere (to keep the height fog separated and simple). This is however not implemented today.

@SebHillaire

Thanks for this work. I love Sky Atmosphere!

Just trying to remove a few artefacts - the main one seems to be related to raymarch step size and is visible when inside the atmosphere volume at height, as shown in the image. I tried the most obvious console parameters from the docs page but didn’t find one that helped this. Any idea?

Thanks very much and sorry for using the wrong thread. I started a new one for the clouds, although I didnt have much to say on it right now: https://forums.unrealengine.com/unreal-engine/feedback-for-epic/1779103-volumetric-clouds-for-sky-coming-in-4-26

@mrboni Hello!
Yes this is because it looks like you have some small radius planet (by default the settings are optimsied for ground views).
So for tiny planets, you likely want to increase both r.SkyAtmosphere.FastSkyLUT.SampleCountMax (used by thje fast sky lut when inside the atmosphere) and r.SkyAtmosphere.SampleCountMax (used by the raymarcher when outside the atmosphere).
Have look at https://docs.unrealengine.com/en-US/…nce/index.html. You might also want to tweak cvars with DistanceToSampleCountMax for both of this cases too for your specific case.

In 4.26, there will be a *ray marching quality scale *(clamp to config quality settings) on the SkyAtmosphere component itself to try and avoid having to deal with CVARs as much as possible.

Also, if the aerial perspective is not showing up on your mesh, you need to increase r.SkyAtmosphere.AerialPerspectiveLUT.Depth. It should be all in the reference document.

@SebHillaire Thanks! I’ve got good results now, but only after disabling r.SkyAtmosphere.FastSkyLUT and r.SkyAtmosphere.AerialPerspectiveLUT.FastApplyOnOpaque whcih I guess means the same rendering approach is used inside the atmosphere as well as outside?

FYI increasing *r.SkyAtmosphere.FastSkyLUT.SampleCountMax *did reduce most of the artefacts, but not at the threshold between the sun shadowed area

Disabling the FastSkyLUT and FastApplyOnOpaque will indeed go the higher quality, but more expensive, per pixel tracing path. If that is in budget for you then all good :slight_smile:

Otherwise the recommended path is to keep those LUTs as much as possible for performance to remain independent of the screen resolution (and play with their resolution and sample count to hide artifact)

@SebHillaire Yes I have the budget thankfully :slight_smile:

Btw do you know anything about parameters affecting (non directional) light sources at these large distances? - https://udn.unrealengine.com/questio…distances.html

SkyAtmosphereLightDirection always returns 0,0,0 when used as a Light Function. Is this a bug or expected behavior?

lfdbg.jpg
Left: Light function applied to a light
Right: same material applied to a mesh

Hi,

so I am now trying 4.25 and whenever I drag Sky Atmosphere actor in the scene, the area below horizon remains pitch black. Few pages back, @SebHillaire mentioned there will be an added option to define atmospheric scattering inside the planet volume or something, but it appears it’s still not present. :confused:

To me, the new atmosphere is still not usable due to this.

And one more question. How do I actually get physically accurate ratio of sun light to sky light? If I enable directional light to be sky atmosphere sun, the intensity value still has an effect, so the proportion of the sun light strength to the sky light intensity seems pretty much random. All the other physically based sky models I know have this taken care of.

“define atmospheric scattering inside the planet volume” I do not remember what I said but I do not promise anything here, just discuss things and possibilities.
The SkyAtmosphere component simulates a terrestrial planet, not a gaz planet. There is a virtual ground. Just put your terrain or your planet mesh on top of it. Looking at Fortnite, UE5 tech demo, Hourence SkyAtmosphere demo, Ryan planets and other projects I saw here an there: it looks very usable to me. It can even be translated and rebased now. But yes it is true I do not know what you are doing exactly…

That being said, to render a gaz planet yourself, you should be able to:

  • reduce the planet radius, increase the atmosphere radius
  • tweak the atmosphere density to make it look like a gaz planet
  • then put you level at the position that suites you. The virtual planet can now be moved around in 4.25 if you want your level to still be at the origin. Just put it where you want.

Oh and you can also use a sky dome and a shader to do anything you want in it with any custom colorization and whatnot.

Very good question about the sun illuminance! I think this is explained in tool tips but not in the documentation? (I will check and we will update the documentation if needed). The sun illuminance is specified as illuminance reaching the ground when the sun is at zenith. So if you put 1000Lux, the ground will always receive 1000Lux, even if you make the atmosphere thicker. It is done the same way as this, section 4.1. The outer space sun illuminance is simply deduced from the ground illuminance and atmosphere properties. This is more practical for artist to specify this way because usually this is where things happens and they do not know/want to guess the outer space luminance to setup the sun light relatively to other lights. Then when you move the sun around, everything is coherent and ratio between sun and sky are still respected because we recover the original outer space sun illuminance (and this is the input to the sky system). Example of sun/sky measurement have been done in this document, section 4.6. I did some measurement in unreal and it matches (you have to be very careful and disable anything that can influence measurements such as AO). More faithful result are obtained by using r.SkyAtmosphere.MultiScatteringLUT.HighQuality 1.

Thanks for the info about the sun illuminance.

As for the color below horizon, I was referring to this quote of yours few pages back:

@Rawalanche Thanks for asking about the sun illuminance: it does seem to lack proper description and even tool tip! We will fix that with documentation in 4.26.

For a gaseous atmosphere: great then, the process I described above should be the one to follow then (I have only played with tiny planets and thick/large atmosphere but I think it should work).