View topic - How to measure if lighting is correct??

How to measure if lighting is correct??

General discussion on lighting, backgrounds, Global Illumination, rendering parameters and tonnemapping.

How to measure if lighting is correct??

Post Wed Nov 01, 2017 5:09 pm

Hello,

A user in BlenderArtists has posted a very interesting observation, please see this:
https://blenderartists.org/forum/showthread.php?436598-YafaRay-news-v3-3-0-released&p=3253155&viewfull=1#post3253155

The problem is that he is comparing YafaRay with Corona and I have no way to know which one is correct or whether there are other differences in the scene and material setup that could justify this.

On the other side this could be pointing to a bug (not sure yet if that's the case). To be 100% sure I need your help please.

I need to know how to check if the lighting in a scene is physically correct. For example I would need a simple "universal" scene similar to the CornellBox but mostly empty and with certain dimensions for the walls and very specific settings for materials and the area light.

I can create that scene, but I don't know how to check if the lighting I get is physically correct. So please will anybody here help to clarify this?

Thanks and best regards!
User avatar
David Bluecame
 
Posts: 404
Joined: Mon Jan 21, 2013 12:42 pm
Location: Spain

Re: How to measure if lighting is correct??

Post Thu Nov 02, 2017 9:08 am

hi David

Phisically correct lighting is only an objective parameter IMO when we measure it in surface materials, not in the camera plane or light source properties. The basic premise is that objects reflects lights according to phisically based laws and no object can reflect more light than it gets, adding all components (diffuse+specular+SSS) and substracting absorpted light. I think the best source of information for that is the PBRT book.

What the camera is seeing or what correct light power settings should we use, has nothing to do with surface properties IMHO. A camera in YafaRay only simulates tone mapping (linear, s-type, etc) so it is a subjective parameter to the user needs, and light source power is only in fact and "exposure" setting, so it is also subjective as well. More light power meas more "scene exposure". That a light type sheds more light in YafaRay than in other programs basically means nothing but that both programs don't have compatible "exposure" values. For instance, that for the same exposure in Yafaray we would need light power = 3 instead of 1 like in another engine. I have tried HDRI in YafaRay and in Marmoset and they produce the same results BTW.

However, it is important that there exist a correlation between background and area light source exposure. For instance, certain value in a light source power setting should be seen the same in a mesh light than in a hdri background, and should produce the very same scene exposure. For instance, an objective enviroment to try this is an HDRI background with only an area as a light source with a known intensity (RGB parameters) and then compare it with a HDRI render of a 'physical' area light in the same scene position with the same light power, both linear mapped in into 8-bits.

In real life we should always differenciate between "ligth shed" and "light seen", while in YafaRay both parameters could be mixed in the same setting which is light power. It is like white balance, it is a camera setting we in fact could simulate with light colors settings. It is possible to modify exposure and white balance in tonemapping but for the sake of comparison tests it is better keeping the camera factor as a neutral one.

For instance, another objective enviroment to test a correct result is to compare an HDRI enviroment image with the HDRI render of the same objects in a 3D scene, and correlating power settings and light sources exposure of both results, given that materias are correct. For that we would need a good deal of information about how the HDRI image was taken, as HDRI lighting depens on camera exposure despite the fact mapping is linear. The result is that we would correlate YafaRay lighting power to a certain exposure settings in a particular camera, nothing else.

IMO there is a good deal of "engine philosophy" about how to correlate lighting power to a given exposure results, and that's a completely subjective area, since not only tone mapping or exposure settings are involved, but also sensor and lens behaviour, etc. IMO many commercial engines try to produce "good" results in a given range of exposure settings for usability reasons. That's why is better concentrating on what every independent factor, light source, material and camera, is separatelly doing.

In fact the only way to solve this for once is implementing some sort of physical camera imo and discarding light source power as an exposure paramenter, then we could draw meaningful comparisons.
User avatar
Samo
 
Posts: 3072
Joined: Tue Dec 20, 2005 10:39 am
Location: Spain

Re: How to measure if lighting is correct??

Post Thu Nov 02, 2017 11:54 am

I can't test Corona either, but I did have a quick look at the shared scene. It uses a Shiny Diffuse material. Switching to a Glossy material gives a result more similar to the one it is compared with.

This may not be relevant as I don't know the materials used in Cycles and Corona . . .
User avatar
stoneage
 
Posts: 1233
Joined: Sun Dec 23, 2007 1:06 pm

Re: How to measure if lighting is correct??

Post Thu Nov 02, 2017 12:12 pm

I will try to prepare some scenes about it.
User avatar
Samo
 
Posts: 3072
Joined: Tue Dec 20, 2005 10:39 am
Location: Spain

Re: How to measure if lighting is correct??

Post Fri Nov 03, 2017 9:38 am

preview.JPG


Modo version with 1.0 W/srm2 of emission level.

Regards
surena
 
Posts: 97
Joined: Sat Apr 26, 2014 6:37 am
Location: Spain


Return to Lighting and Rendering



Who is online

Users browsing this forum: No registered users and 3 guests