Transparency doesn't work with cloudy volume scatter?
rokketman last edited by
@ghostship I just ran a sim without any of those issues too. I used the primitive box as the atmosphere. No problems at all, until I started messing with the math modes with cycles. I don't know enough about cycles to know how the math modes respond, so I just tried to use them as I would on the atmosphere tab. No bueno.
ghostship last edited by
Thank you for your answers, but all your proposals have no effect. It doesn't matter wich base node I use or if I use math nodes or not, or if path brachned is on or off.
But the effect is not strong in every case. The effect is stronger if you use a big scaled clouds. scale = 100 for example.
A year a go we discuss this case and we found no solution.
it depends on some factors that get this effect. It depends on lights also how strong the effect is.
I can not uplaod any files here. so I can't share it to show you my example
Maby you can download this file here and test it
anomalaus last edited by
@joker a potential problem with props containing a volumetric shader, is that scaling the prop requires inversely scaling the density, since the density is in geometry units, not worldspace units.
Here's my comment from a Volumetric Shader thread on this forum:
Absence of scaling is an important requirement for the density settings of volumetric absorption and scattering in SuperFly. My experiments with standard props scaled up to large volumes to enclose the camera and scene objects has required equivalent (and logically appropriate) scaling down of density settings (to maintain their mesh-relative units). I suspect there may be some fixed precision operations involving density, which are intended to speed up the calculation of volumetric effects during renders, beyond what can be achieved by using floating point operations. The side-effect of using fixed precision is that tiny density scales can underflow to zero, thus reducing the possible effects to either on, but too much (totally opaque) or off, as I have observed in such experiments.
Without object scaling, changes to volumetric effects are easily seen with small value changes to the density inputs from their default of 1. If one needs to scale a volumetric prop to match a scene, the simplest way to revert to unitary scales is to open the grouping tool, select all the facets of the scaled prop (or actor) and spawn a new prop, then delete the original, scaled prop.
I think you may also need to bump up your Transmission Bounces, perhaps.
joker last edited by joker
@anomalaus Thank you for your answer, but I don't understand what and how it's to change.
But maby anybody can change this example file in this way, that the transparent plane is not visible anymore in this cloudy fog cube?
I can reproduce the issue using the provided test scene. No matter what i use: PoserSurface or CyclesRoot node, CPU, GPU, BPT on or off. As soon as the volume is modulated i get the trans issue.
@nagra_00_ Thank you for reproducing this issue. I tried so many things but nothing worked to solve this problem with trans in modulted volume scatter.
Maybe anybody will find a solution? Or maby this is only solvable by the software developers?
@joker Does this happen with cloud modulation only or does any modulation like e.g. a simple density gradient has the same problem?
@nagra_00_ The issue is with any kind of modulation the densitiy of the fog
@joker Ganz schlecht :( Looks to me as if this could be fixed by the devs only.
@joker Looks like i found a solution:
The setup looks strange but seems to work.
@nagra_00_ Ok, forget it does not work...
@nagra_00_ Yes, that's what I am thinking also. The devs can fix this issue only. But the question is, will this happen anytime?
Atmosphere effects are very important in my eyes. It makes a scene much more idyllic. I try to communicate this bug a long time here but nobody finds a solution.