9
$\begingroup$

There is a new volumetric shader node called "Principled Volume" in Blender 2.79 as of this commit. It appears to support a new temperature attribute, using blackbody temperature.

Using the more or less familiar setup as pictured below, I can get a good enough material, but I would like to know how to achieve the same result with this new shader node.

"Old method" and result:

enter image description here enter image description here

Setup with Principled Volume node:

enter image description here enter image description here

Here is the current file to play with. It does require a current (as of 3/5/18) buildbot build to use the shader.

The question:

How do I set up a material with the Principled Volume node that renders visually the same as the old setup?

$\endgroup$
  • 1
    $\begingroup$ I will see if I can play around with this soon, but in the meantime there are a couple of things that look off to me. The flame-heat-colorramp group of nodes is connected to the emission color in the first setup, and the volume color in the second. The first setup also has a much higher emission strength than the second (may have to be this way, not sure yet). Finally from the link you posted I think the blackbody input should be connected to the flame attribute. I will test this when I can and update accordingly. $\endgroup$ – Brenticus Mar 5 '18 at 6:20
  • $\begingroup$ Would you be able to add your blend file? That way it will be easier for me to try and match the initial result rather than trying to recreate what you have and then match it. $\endgroup$ – Brenticus Mar 5 '18 at 17:15
  • $\begingroup$ @Brenticus see my edit, have added the file as requested. Haven't had time to test flame to blackbody input, but will look at that. $\endgroup$ – Timaroberts Mar 5 '18 at 22:57
5
+50
$\begingroup$

Here is my final result.

The result using the first node setup:

enter image description here

And the result using final principled setup:

enter image description here

Let me start of by saying that smoke simulation and rendering is not something I am very familiar with, so despite being able to replicate the material fairly well, I don't necessarily understand everything behind why it works the way it does. But I'll explain what I can.

A quick summary of what I did:

  1. Left the default color as white, and used the colorramp as the input for the Emission Color.

  2. Increased Emission Strength to the value you had, but then dialed it back some as it was too strong. I noticed that the two images aren't exaclty the same, and further reducing the Emission Strength could probably fix that.

  3. Increased the Density to half of the value you used in the first setup, as this looked closest.

  4. Left everything else as the default. I tried changing the color and temperature attributes, as well as the Blackbody settings, but they seemed to have no effect with this setup.

So, from what I understand at the end of this, the Principled Volume Shader is basically setup to replicate the original node setup you have, and it does it well. There does appear to be some differences in how the shader works compared to the old method, namely in how the density works. I noticed that (before emission was added, which made the details hard to see) the Principled smoke looked more dense than the manual smoke, but if I decreased the density it would get to the point of areas being nearly transparent before the main density looked similar. More details were visible in the principled smoke, and since in this setup the smoke was colliding with the domain, the principled setup had more defined boundaries, whereas the manual setup was much less defined. Perhaps with some experimentation, or a better understanding of how the shader works, as my knowledge is limited, a more similar result could be achieved.

The good news in all of this is that smoke is now much easier to setup, as the final setup I arrived at was just this:

enter image description here

Hope that helps!

| improve this answer | |
$\endgroup$
  • $\begingroup$ Having emission be linked to fire makes more sense to me. As I understand it, color would be for the volume scattering part of the volume node. I'm still not clear if the color ramp is necessary or the heat information should be somehow connected to the blackbody. $\endgroup$ – user1853 Mar 6 '18 at 2:08
  • $\begingroup$ That's a good way to think about it. In this case then the color input would be the smoke color, which makes sense. The color ramp is what defines the color of the fire. Without it it's essentially black and white. From what I have read, the heat/flame attribute should be connected to the blackbody, just in this case with the intense emission there was no visible change, so for the sake of simplicity I left it out, and also in case I got it wrong. $\endgroup$ – Brenticus Mar 6 '18 at 2:41
  • 1
    $\begingroup$ I just realized that I never tried connecting the flame to the Blackbody tint, only to the intensity. That makes sense the more I think about it, and increasing the temperature would make the effect more noticeable. I'm liking the results so far with this node. $\endgroup$ – Brenticus Mar 7 '18 at 2:17
2
$\begingroup$

Here's my take on fire (I'm not trying to replicate the example, but trying to make one using blackbody and color temperature).

enter image description hereIt seems that decent looking fire is now quite easy with a ramp on the Blackbody tint, and controlling the temperature with the heat attribute.

| improve this answer | |
$\endgroup$
  • 1
    $\begingroup$ I like this answer, but now it is sparking another thought, and I'm wondering if it is possible to now use brush and canvas, to animate the temperature falloff when the canvas trail dries. Any thoughts? $\endgroup$ – Rick Riggs Mar 8 '18 at 11:09

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.