It is interesting to compare the reflection models we learned in class to the options that Blender provides for creating reflective materials. Some of the options are highlighted here: https://docs.blender.org/manual/en/2.79/render/blender_render/materials/properties/mirror.html.
misha-wu
Why is there a max(0, m*l) term in the L_d breakdown?
elaineqian02
The Blinn-Phong reflection model is a modification, developed by Blinn, of the Phong model. It is simpler and less computationally expensive to implement, making it a more popular choice for real-time rendering. Instead of computing the reflection vector, the specular reflection term is calculated using the dot product of the halfway vector (the normalized sum of the light source and the view vectors) and the surface normal. This image compares the specular highlights in the two models: https://en.wikipedia.org/wiki/File:Blinn_phong_comparison.png
wilrothman
I have seen these models before in the context of animation but I never knew what they meant. I wonder, then, where the formula for each one of these models comes from. Particularly the Specular L_s component.
jbf11
@misha-wu, My understanding is that the max(0, n.l) term ensures that we do not compute negative lighting. If the light source is behind the surface, then we expect none of its light to reach the surface and reflect off of it, so we must get 0 reflective lighting. Mathematically, in this case, we would compute n.l < 0, so without the maximum we would compute negative light (which doesn't make sense), and is why 0 is the correct value to use instead.
colinsteidtmann
@misha-wu,
I think you mean max(0, n*l), not "max(0, m*l)," it's needed because the normal vector for the surface pixel may be at an obtuse angle to the light vector (e.g. the back of a cup). "*" is the dot product. Taking the dot product of two vectors gives the cosine of the angle between them and cosine for angles that are between 90 degrees and 270 degrees is negative. Negative illumination doesn't make sense
Edge7481
At a fundamental level what is the motivation for separating lighting into these 3 types? Can materials in real life be modeled with a "diffuseness" and a "specularness"? It also seems like this model is limited by assuming that the object itself doesn't emit light and only works if the light source is out of view
sueyoungshim
It's quite fascinating to learn that something as simple as the way an object is shaded in a digital image is actually a complex combination of different lighting components. Before seeing this, I didn't realize that the realistic appearance of an object in a computer-generated image is due to a deliberate mix of ambient, diffuse, and specular lighting. The mathematical formula blending these components together to achieve the Phong Reflection is especially intriguing, showing the depth of thought put into what might seem like a straightforward visual effect.
j-nn
This is such a cool visual, to show how something 2D can look so 3D. It no longer relies on just the reflection vector, but also uses light direction vectors.
MillerHollinger
Does this need a normalizing constant? I notice that the lighest parts of the specular image are white. Doesn't that mean that adding the values could cause large amounts of the image to become white?
stephanie-fu
I found an example at the bottom of the page here: https://garykeen27.wixsite.com/portfolio/blinn-phong-shading
I don't quite understand the specular highlights here - why are they so blocky when we have an r^2 falloff term?
stang085
It's very interesting that you can calculate every part of the object individually and then add everything together to get a really three dimentional effect.
stang085
I wonder if you didn't add them but did other things, how would that effect the way the surface looks?
Alescontrela
it's interesting that as the scene gets darker, things look more realistic. The brain fills in the missing details to make things look more realistic.
It is interesting to compare the reflection models we learned in class to the options that Blender provides for creating reflective materials. Some of the options are highlighted here: https://docs.blender.org/manual/en/2.79/render/blender_render/materials/properties/mirror.html.
Why is there a max(0, m*l) term in the L_d breakdown?
The Blinn-Phong reflection model is a modification, developed by Blinn, of the Phong model. It is simpler and less computationally expensive to implement, making it a more popular choice for real-time rendering. Instead of computing the reflection vector, the specular reflection term is calculated using the dot product of the halfway vector (the normalized sum of the light source and the view vectors) and the surface normal. This image compares the specular highlights in the two models: https://en.wikipedia.org/wiki/File:Blinn_phong_comparison.png
I have seen these models before in the context of animation but I never knew what they meant. I wonder, then, where the formula for each one of these models comes from. Particularly the Specular L_s component.
@misha-wu, My understanding is that the max(0, n.l) term ensures that we do not compute negative lighting. If the light source is behind the surface, then we expect none of its light to reach the surface and reflect off of it, so we must get 0 reflective lighting. Mathematically, in this case, we would compute n.l < 0, so without the maximum we would compute negative light (which doesn't make sense), and is why 0 is the correct value to use instead.
@misha-wu,
I think you mean max(0, n*l), not "max(0, m*l)," it's needed because the normal vector for the surface pixel may be at an obtuse angle to the light vector (e.g. the back of a cup). "*" is the dot product. Taking the dot product of two vectors gives the cosine of the angle between them and cosine for angles that are between 90 degrees and 270 degrees is negative. Negative illumination doesn't make sense
At a fundamental level what is the motivation for separating lighting into these 3 types? Can materials in real life be modeled with a "diffuseness" and a "specularness"? It also seems like this model is limited by assuming that the object itself doesn't emit light and only works if the light source is out of view
It's quite fascinating to learn that something as simple as the way an object is shaded in a digital image is actually a complex combination of different lighting components. Before seeing this, I didn't realize that the realistic appearance of an object in a computer-generated image is due to a deliberate mix of ambient, diffuse, and specular lighting. The mathematical formula blending these components together to achieve the Phong Reflection is especially intriguing, showing the depth of thought put into what might seem like a straightforward visual effect.
This is such a cool visual, to show how something 2D can look so 3D. It no longer relies on just the reflection vector, but also uses light direction vectors.
Does this need a normalizing constant? I notice that the lighest parts of the specular image are white. Doesn't that mean that adding the values could cause large amounts of the image to become white?
I found an example at the bottom of the page here: https://garykeen27.wixsite.com/portfolio/blinn-phong-shading
I don't quite understand the specular highlights here - why are they so blocky when we have an r^2 falloff term?
It's very interesting that you can calculate every part of the object individually and then add everything together to get a really three dimentional effect.
I wonder if you didn't add them but did other things, how would that effect the way the surface looks?
it's interesting that as the scene gets darker, things look more realistic. The brain fills in the missing details to make things look more realistic.