So I was thinking about the problem of shadows in images. From an image processing point of view, would you classify cast shadows in a scene as an additive phenomenon or as a multiplicative phenomenon on top of the shading?

From what I can understand, for the case of non-lambertian image formation equation, the specularity is formulated as additive because it is dependent on the view direction. But for the case of shadows, is there any similarity to that? Or should shadows be formulated as just as a scaling term over the shading?

Any views or pointing towards proper papers or materials would be highly appreciated. Thanks!

  • $\begingroup$ not an image processing guy, but wouldn't a "shadow" be "lack of light", actually, and then your question becomes "is illumination an additive phenomenon?". $\endgroup$ – Marcus Müller Jul 28 '18 at 15:02
  • $\begingroup$ What are you adding or multiplying in this case? Shadow is the lack of light. $\endgroup$ – Cris Luengo Jul 28 '18 at 15:02
  • $\begingroup$ I am thinking of adding cast shadows on top of a normal based rendering of a shading. So given an object, you can get a shading with the normal and the light direction. But this does not calculate the shadows cast on it by another object. This is only done by a ray tracer. So I was wondering if this shadow is an additive to the shading or more a multiplicative to the shaded pixel value. $\endgroup$ – Blade3000 Jul 29 '18 at 16:30
  • $\begingroup$ You ask an interesting question. The term additive and multiplicative can be used to characterize noise. A shadow seems more like a response to an input, an illumination source. In system ID, a probe signal can only reveal partial characteristics of the system. A shadow would seem to be a minimal information response to a probe $\endgroup$ – user28715 Jul 30 '18 at 23:08

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.