A candle test! Using the gradient texture + blackbody to get some color variations across the flame. (there is a blue section at the base, but it’s a bit hard to see in this shot).
Shader setup: (as a group so I can use it with both the volume emission and the surface emitter that’s the real light source. Volume object is hidden from diffuse/shadow rays)
Also, is there a plan to do some kind of “volume sample” option for branched path? The flame needed a lot of samples to completely get rid of the noise, and I couldn’t find a way to actually add more samples just on the flame.
Ok, Cloud Generator guy here trying to make clouds. Am I missing something? What I mean is Volume absorption is cool, but don’t we need volume reflection/scattering? I don’t see any light reflected into the camera from the volume, all I see is light being eaten by it. Am I correct in saying clouds aren’t possible due to this?
That is correct. Scattering is not in the regular release but you can try it out in the GSOC release it works quite nicely but is a bit slow as you might expect. I have followed your adventures with clouds for a while and I admire your tenacity!
J_the_Ninja: Nice candle test. Absorption and Emission Volume (the volume itself) should clear up after 1-2 samples only, and as as you noticed for heterogeneous volumes a lower step size can help.
Scattering needs to be added still in master, Brecht is working on it. I am not sure if we will have Dedicated Samples for this (for Branched Path), but might be a good idea.
Thanks so much for the info DingTo. Matt Ebb and Brecht are heroes for their volume efforts. I look forward to scattering. If scattering is controlled separately from absorption this implementation is shaping up to very intuitive indeed. As much as I loved volumes in BI, it was VERY difficult to get natural looking clouds, I look forward to what Brecht comes up with:)
Right now, a scattering medium would obviously require a lot more samples than a normal situation without volumes, but the difference should be narrowed a little in the future as it doesn’t currently have various optimizations (including one that’s already in the volume branch).
Brecht has written in the wiki that there’s currently no support for having the camera inside of a volume, you can however put the camera in a sort of ‘bubble’ with the normals facing inward which I found works just fine (as long as there’s an outer surface with the normals pointing outward).
@Tekneek, I like to share my build but I cant build static and think it run only on my system.
If you use also Opensuse Linux 13.1 64 bit it should work but even Ubuntu did not, I fear.
If you use Windows we could poke one of the builder to activate GPU for volumes.
Until you realize of all the things that could be done to improve Blender yet
Absolute perfection is more or less just a myth in the software world, for example, 3D Studio Max has been developed by a paid team for how many years now and there’s still plenty of complaints and suggestions on improvement.
There have been a few fixes, it is better now. Use a homogeneous Scatter volume, that’s the fastest.
The World surface light is very far away, so it’s to be expected that you don’t see it. For this case you should use a volume object (like a large cube or so).
As for the opposite color, use an add shader and combine scattering and absorption to get the effect you want. If you just use Scatter node, the color change is correct. This has been discussed in the Easter Egg thread already.
