I noticed that Corona dev has no intention of improving the multi-scatter of volume rendering in Corona - very important feature, which V-ray 6 has done very well, the latest version of V-ray allows realistic cloud rendering ( with deep scatter and dark edge ) with fast render times. I hope the developers can focus on this
i can’t talk about the speed, i havent yet compared with v-ray,
but the scatter phase feature you linked is also in Corona Render (forard/backard scatter value).
I’ve tested the Moana cloud some times ago in Corona (for Max), and while it was slow - I agree - it looked quite good.
There are a couple things to keep in mind for a fair comparison. Do you have similar phase function settings across all renderers? This is one of the most deciding factors for how realistic rendered clouds will look. The 0.5 value in Vray’s examples will be typically too low to generate dark edges, a more realistic value is in the range of 0.7 to 0.9.
Corona’s performance tanks once you use higher values for the phase function (positive values mean forward scattering) and it’ll be quite slow for values close to 0.9 - generally increasing this value will result in slower rendering since way more samples are needed to resolve the scattering within the cloud.
Another factor is how many bounces you use. Corona’s default will be way more than most renderers. However for a realistic result you’ll need to go even higher.
Ofcourse I set all of them is 0.7 at directionality. Depth ray = 100. Butafter many tests i have to admit that the light passing through corona cloud is not deep enough to form cloud shading
->> well go to scatter tab (in volume grid), the setting “directionality” is exact the phase function you point to in the vray manual. the similar you have in volume material by the way.
edit: vray has also gi depth 100(see lightcache advanced) settings
Judging from the render you posted above your density is too high, your cloud looks too solid. For me, density for the Moana cloud had to be around 0.4/0.5.
For darker shadowing you have to play with the scattering multiplier.
nothing special of course, but i think scattering is not so bad i feel, and even renderspeed also is ok (i raised step value a bit to make it faster).
the scale value fro absorption and scattering is most important in my experience, so it fits the size of scene and cloud.
If you use the density channel for scattering you are basically diminishing it where the density increases - while it gives you some additional artistic control, it’s technically wrong. Scattering should be a constant color, and then play around with the value - it has a ceiling value where it won’t change with increased values. It should produce a better result.
You can see that there are still dark inner of your cloud. Due to weak scattering. And that will be even worse when you increase thickness of the cloud
@pokoy: ok good to know, i will try constant later
@thacmac: if you adapt the scale this goes away, but i found it quite realistic as also in reality clouds are darker where they are thicker if light comes from behind, or not?
maybe share a setting, so we can play and try with it ?
I always set scatter to 100% white, no channel at all. So, please take it seriously and accept that Corona volume is not deep scattering enough for cloud.
If you mean my response, I was referring to Stefan-L, not your tests.
I’m really taking this seriously as this is important for my work, and I have done many tests and am a bit torn on this topic myself. But to make it a serious conversation, not just a random “Here’s a better looking result” thread, my question still is - what is the other renderer in the pic you posted?
Notes to whomever
Take into account retro-reflective properties of water droplets/mist - so set directionality accordingly
also, set sample intensity & ray depth as high as you can afford, otherwise clamping occurs
EDIT
There also seems to be a kind of bug w/ use of VDB grid - roughness on sphere is gone inside the grid… (?)