BananaftJan 12
Christoph Peters
You cannot get physically-based rendering right without understanding radiometry. My new blog post explains it all, relying on familiar concepts from rendering algorithms as much as possible.
Part 2 will come next week.
https://momentsingraphics.de/Radiometry1Backwards.html
Radiometry, part 1: I got it backwards

Jan 12 at 7:43pmWeb
Show threadrfnixJan 12
@MomentsInGraphics I really wish optics books started with radiance, it's much closer to the basics of geometrical optics we learn in high school, you can say "it's like a single light ray" and you'd be mostly correct.
Show threadPekka VäänänenJan 12
@MomentsInGraphics This looks like a really useful framing, thank you. Will read this in more detail later.
Show threadmmbyJan 13

@MomentsInGraphics solid angle my beloved.

also it sometimes seems that more than half of tech-literate humanity doesn't know that optics and radiative transfer has conservation laws akin to the third law of thermodynamics

Show threadPaleosJan 13
@MomentsInGraphics I really had to dig deep into such concepts for my bachelor thesis. Thanks to Eric Veach's PhD thesis and Jim Arvo's work.
Show threadPekka VäänänenJan 28

@MomentsInGraphics Finished reading the first part now. Very neat explanation and I'll sure be referencing this in the future. Thanks!

The only thing that confused me was why the absolute value was used instead of max(0,x) in the irradiance definition.

Show threadChristoph PetersJan 28
@pekkavaa Glad you liked it.
Regarding irradiance, you may very well want to use max(0,n⋅ω) instead of |n⋅ω|. Especially when computing diffuse shading, that is the right thing to do. But then again, if you have a closed mesh, radiance below the surface should be zero anyway.
For the definition of radiant flux, it makes more sense to work with |n⋅ω| because I define the flux as a property of the surface A, without saying which side is outside.
Maybe I'll clarify that alongside some update.