Bidir - Metropolis.

Use this forum for general user support and related questions.
Forum rules
Please upload a testscene that allows developers to reproduce the problem, and attach some images.
Post Reply
JuandeLara
Posts: 9
Joined: Wed Oct 28, 2020 2:39 pm

Bidir - Metropolis.

Post by JuandeLara » Thu Dec 17, 2020 3:00 pm

Hello all,

SO basically I am trying to obtain the most reliable and PBR behavior of alight in a complex 3D space, reason why I chose LuxCore. I have set up a scene and I am calculating the sun behavior in it as obtained with external data for sun positioning. However when setting the renderer I have some questions.

I am using Bidir - Metropolis.

For the best possible result in terms of behavior of light, is it fair to say that 'Light PAths>Max Bounces> Eye Depth and Light Depth should be set at the highest values (for example 800). I attach a test in between that value at 10 vs. that value at 800. The result is that the interior is better lit (the walls are a glossy type of material to replciate marble). Is this rationale correct? Is the value 800 a more accurate representation of real live light behaviour?
Thanks!

doubts
Capture.JPG

CodeHD
Posts: 408
Joined: Tue Dec 11, 2018 12:38 pm
Location: Germany

Re: Bidir - Metropolis.

Post by CodeHD » Thu Dec 17, 2020 3:41 pm

Hi,

yes, in general a higher path depth is better, just your example of 800 is very excessive for almost all realistic scenarios. Let me give you some example numbers:

First off, you can find an example of a situation requiring >100 describes in the follwing thread: https://forums.luxcorerender.org/viewto ... f=4&t=2667. This scenario is based on multiple total internal reflection in a glass or acrylic material.

For other surfaces, it depends on the reflectivity. A really good white paint could reach up to, lets say, 0.9 total reflectance. If you bounce that between 10 surfaces it becomes 0.9^10 = 0.35. So after 10 bounces it still carries 35% the energy, which can be significant. after 50 bounces it is down to 0.9^50 = 0.005. Without cosnidering HDR images or gamma correction, this is approximately one step in an 8-bit png or jpg image (1/255 = 0.004).

If your scene contains also darker materials, your average relfectance will be maybe 0.7. with 10 bounces this becomes 0.7^10 = 0.028. So still a possible contribution by higher path depths, but not too much.

You have only provided an image, and not a .blend file that shows all material setups, so hard to say what should be expected for your setup.

User avatar
Dade
Developer
Posts: 4974
Joined: Mon Dec 04, 2017 8:36 pm
Location: Italy

Re: Bidir - Metropolis.

Post by Dade » Thu Dec 17, 2020 4:02 pm

800 is an over the sky value, you are still using floating point 32bit precision, it is like an order of magnitude more than a very high value.
Support LuxCoreRender project with salts and bounties

JuandeLara
Posts: 9
Joined: Wed Oct 28, 2020 2:39 pm

Re: Bidir - Metropolis.

Post by JuandeLara » Thu Dec 17, 2020 4:37 pm

CodeHD wrote:
Thu Dec 17, 2020 3:41 pm
Hi,

yes, in general a higher path depth is better, just your example of 800 is very excessive for almost all realistic scenarios. Let me give you some example numbers:

First off, you can find an example of a situation requiring >100 describes in the follwing thread: https://forums.luxcorerender.org/viewto ... f=4&t=2667. This scenario is based on multiple total internal reflection in a glass or acrylic material.

For other surfaces, it depends on the reflectivity. A really good white paint could reach up to, lets say, 0.9 total reflectance. If you bounce that between 10 surfaces it becomes 0.9^10 = 0.35. So after 10 bounces it still carries 35% the energy, which can be significant. after 50 bounces it is down to 0.9^50 = 0.005. Without considering HDR images or gamma correction, this is approximately one step in an 8-bit png or jpg image (1/255 = 0.004).

If your scene contains also darker materials, your average relfectance will be maybe 0.7. with 10 bounces this becomes 0.7^10 = 0.028. So still a possible contribution by higher path depths, but not too much.

You have only provided an image, and not a .blend file that shows all material setups, so hard to say what should be expected for your setup.
Oh I see. Thanks for this info. So I think I am just being very lame with this and not understanding some material processes at all.

Basically, lets assume the whole scenario is built in marble. I have estimated the marble as the attached material. How did I do this estimation? Well by eye, which is not very scientific. Is my starting point (setting a glossy material for it) correct? I estimated the roughness to be 0.1, which I guess is the reflectance? I would say the current setup looks "like marble"....Is there anywhere we the reflectance value for a material like gold or marble can be sourced? Am I missing an important point here?

I think I need to go back to some material basics. For example I don't know what Absorption depth stands in this context.

Would then the Max bounces at 30 be proper for such scenario?

CodeHD
Posts: 408
Joined: Tue Dec 11, 2018 12:38 pm
Location: Germany

Re: Bidir - Metropolis.

Post by CodeHD » Thu Dec 17, 2020 6:11 pm

JuandeLara wrote:
Thu Dec 17, 2020 4:37 pm
Basically, lets assume the whole scenario is built in marble. I have estimated the marble as the attached material.
There is no attachment, did you forget it?
JuandeLara wrote:
Thu Dec 17, 2020 4:37 pm
Is my starting point (setting a glossy material for it) correct?
Yes and no. Marble usually has subsurface scattering so you would rather need glossy translucent and a homogeneous volume.
JuandeLara wrote:
Thu Dec 17, 2020 4:37 pm
I estimated the roughness to be 0.1, which I guess is the reflectance?
No, reflectance would be the color setting.

Roughness just tells you if the surface scatters more like a mirror (ray bounces off at the same angle as it is incoming), or like a diffuse surface (ray bounces in a random direction regardless of where it comes from).
Now, roughness can reduce the reflectance depending on the rougness model, but that's an issue of the algorithms used, not the real-world physics behind it.
JuandeLara wrote:
Thu Dec 17, 2020 4:37 pm
I would say the current setup looks "like marble"....Is there anywhere we the reflectance value for a material like gold or marble can be sourced? Am I missing an important point here?
I'm not sure of a database right now. Perhaps someone else can help. Marble is a natural material, so there can be variations of course. It all depends on what your goal is. Artistic; replicating an existing room; predicting a building to be constructed; ...
JuandeLara wrote:
Thu Dec 17, 2020 4:37 pm
I think I need to go back to some material basics. For example I don't know what Absorption depth stands in this context.

Would then the Max bounces at 30 be proper for such scenario?
Yeah, most of the time that will be fine.
One more detail to add since you talk about BiDir: The total path depth can be as high as the sum of eye depth and light depth! So even 15+15 would be a lot. It can be very scene dependent what you really need, depending on the involvement of caustics.

Post Reply