SSmolak wrote:Despot wrote:SSmolak wrote:The new Hosek-Wilkie Sky Model is great but where is the most important option like sun intensity ? It was main problem with other Octane sky models too. We have power to set overall exposure like sky intensity but there is no individual setting for sun Intensity and it shadow. How to make ambient mood without going to HDR lighting ?
Increasing Turbidity doesn't decrease Sun intensity like it is in other Octane Sky Models - something wrong is here with Hosek&Wilkie model.
I agree, this is not correct behaviour...
+1 for a fix please.
Hosek&Wilkie Turbidity behavior is good while increasing it when we see more pronounced Sun aura and decreased Sky saturation - but Sun intensity stays the same, even at large Turbidity values. This is wrong. It works different not only because other softwares do it good but it works different than other implemented Octane Sky models where we have control over Sun by Turbidity. This is somewhat limited but better than nothing. Now we don't have any control over sun.
The Hosek-Wilkie daylight model is a direct implementation of the model as described in the paper by Hosek and Wilkie (if you're interested, the gory details are available at
https://cgg.mff.cuni.cz/projects/SkylightModelling/). As far as I know, the implementation in Octane is correct, but the model itself does have some limitations, such as not being able to simulate overcast conditions (the turbidity parameter controls sky haziness, but it doesn't simulate actual clouds and the range of the parameter doesn't go far enough to simulate thick fog).
The ideal solution would be for Octane to implement a model specifically designed to simulate overcast conditions. The only such models I'm aware of provide luminance only; no color or spectral data so they're not ideal for use in Octane, but they could possibly be adapted in some way.
We could add a sun intensity parameter to attempt to fake overcast conditions. This wouldn't produce very accurate results (for example it would be unrealistic to have a big bright sun aura being produced by a dim sun), which is potentially a problem as the goal of the Hosek-Wilkie model is simulation accuracy. It would come with other problems like tinting the scene slightly blue due to the sun no longer counteracting the blue color of the sky. It would however be easy to add (to all models) and sounds like a good idea as long as we made the potential issues clear to users. I can't promise when but I'll have a go at adding this parameter and seeing what the results look like.
For now, the best way to achieve the result you're looking for is to use either one of the other daylight models or an HDR texture.
Side note 1: Technically, the sun intensity does change with turbidity in the Hosek-Wilkie model. It's just very, very bright, so you can normally not tell the difference. When the sun is close to the horizon you can see the difference: increasing the turbidity makes the sunlight dimmer and more orange.
Side note 2: Something that might be useful to know is that the maximum turbidity supported by the Hosek-Wilkie model is 10, so any turbidity values above 10 are treated as 10 for the Hosek-Wilkie model in Octane. All the data comes from tabulated data provided with the paper, which only goes up to 10, so there's nothing we can really do about this.
