[Bf-committers]

Jean-Luc Peuriere jlp at nerim.net
Tue Apr 5 19:24:39 CEST 2005


Le 5 avr. 05, à 07:30, Matt Ebb a écrit :

>> 3) Patches/Tracker
>> -Minnaert shader should be changed so that user can choose either the
>> standard Minnaert shader or a variant, to be called 'minnaert-B
>
> Uh, this would be *really* overcomplicating the issue. Minnaert is  
> such a specific usage shader already, having two variants that are  
> only marginally different would be overkill. I second Jorge's  
> suggestion in the dialogue he's started on the tuhopuu-devel mailing  
> list (  
> http://projects.blender.org/pipermail/tuhopuu-devel/2005-April/ 
> 001277.html ) - just keep the current nice, simple and usable  
> implementation as is, and rename it to something like "velvet/satin"  
> or whatever, so pedants don't get offended that it's not the 'real'  
> Minnaert formula. To most users, the end visual appearance is what's  
> important.
>

The issue is the following :

  On the net and in other apps, Minnaert refer to a defined shader (less  
used than lambert agreed but still well known), and the parameter for  
it is k also known as darkness. For k = 1, gives same result as  
standard Lambert.

NVidia diffused a different version of this shader, where results are  
similar but  inversed. it is known on several internet referral pages  
as Minnaert-B or NV-Minnaert.

Jorge implementation mix the 2 in one implementation, with a darkness  
parameter where darkness = 0 gives Lambert.

The problem is only there, same name, same parameter name but gives  
different result ! Not good in my book.

And it is only a matter of UI, as if the latter provides a darkness  
parameter where k = 1 give Lambert, and a toggle to choose the version,  
we can use the existing code and be conform to the docs you can find on  
internet.

Besides, if you look at all the shaders existing in blender they are  
named by their standard name, so changing the name to 'velvet' or  
things like that would not be good either.

-- 
Jean-Luc



More information about the Bf-committers mailing list