"Never use 100% black or white for diffuse and reflection" myth

[ThomasMiller]

I've heard multiple times that you should never use 100% black or white in the diffuse or reflection slots as it might cause issues, but it's never been explained what kind of issues. It seems to me that this advice might have been useful in the past, but doesn't apply anymore. I frequently use both 100% black and white in diffuse and reflection and it's never caused me problems.

Where did this come from? I see it repeated so many times without any explanation that it sounds more like superstition.

Am I wrong in thinking that this is a myth?

[oglu]

You can do what ever you want but there is no white or black in real live. If you make something pure white there is no room left for a highlight caused by reflections.
If you like to fake the real world dont use white or black. If you want a good looking image just do whats needed to get the desired look.

https://google.github.io/filament/im...rial_chart.jpg

[vlado]

There is no problem using pure white for reflection so long as the Fresnel option is also turned on.

Pure white for diffuse tends to look unnatural though. Pure black should be used for glass.

Best regards,
Vlado

[Xntric]

Check this great article and explanation of why it makes your image looks flat to use high albedo.
https://www.racoon-artworks.de/cgbas...otorealism.php

In Corona render, it even makes rendering slower, and i assume it's the same in V-ray. Actually i would like to test if it's making renders slower in V-ray and post my results.
https://coronarenderer.freshdesk.com...hat-is-albedo-

I never go above 220 white, you should rather compensate with lighting and exposure.

[Xntric]

I did a quick test on an old scene.

Just applied an override material. One render with diffuse of 200/200/200 and one with 255/255/255.
I tried to match the exposure with camerasettings.

I'm surprised that it didn't affect rendertimes like it does in Corona, it somehow seemed a little faster. I guess the lack of contrast makes the adaptive sampler measure less noise?
This was just a single test done on Vray GPU though, so other scenarios might show something different.

What this test does show, is how flat and unrealistic your image becomes when using unrealisticly high values like 255 diffuse.

[^Lele^]

The second image is clearly different to the first: exposure outside of the window, and the brightest sun-lit parts are way lower in intensity.
That's rather why it's quicker: shaders are brighter, reflecting more of the incoming light, but the light flux to be bounced is much lower, so it extinguishes sooner.

[Xntric]

Makes sense.

Only thing i have changed is the camera exposure between the two, and ofcourse the override material.
I obviously had to lower camera exposure a lot on 255 one. I don't think it would make sense comparing if the exposure wasn't more or less the same.

My test was more about showing that Unrealistic high albedo values breaks the realism and makes your image appear very flat.

[^Lele^]

My test was more about showing that Unrealistic high albedo values breaks the realism and makes your image appear very flat.

I agree on it not making much sense as a comparison, particularly when it seems to deduce that higher albedos render quicker.
One of the key issues of high albedo *is* longer rendertimes, and this is valid across all engines on the market: a light ray will take longer to extinguish, and with a 1.0 white is likely to hit the max bounces limit (which for us is by default at 100 for GI).
If you want to read about this, there's still a monster thread lurking around the forums, here, from way back when.
Little changed, conceptually, in regard to albedo.

[Karol.Osinski]

You can do what ever you want but there is no white or black in real live. If you make something pure white there is no room left for a highlight caused by reflections.
If you like to fake the real world dont use white or black. If you want a good looking image just do whats needed to get the desired look.

https://google.github.io/filament/im...rial_chart.jpg

Thanks for sharing the chart, though the glass reflectance values seems very odd to me...I mean, good luck getting nice looking glass with reflectance value set to 119 grey... I often struggle to get the glass looking "glassy" enough with almost white values.

Could anybody explain the reason why above chart indicates such values? Also I don't really get how 4% percentage of reflectance converts to grey being half way on the slider.

[Xntric]

Thanks Lele, yeah i also couldn't understand why it wouldn't be slower.

[oglu]

This writeup does a good job explaining the PBR workflow.
https://google.github.io/filament/Filament.md.html

[Karol.Osinski]

Great lecture !! Thanks oglu !!!