How would they know that they had exhausted all possible test light colors?

jgforsberg

@RichardChen9 I don't think its possible to exhaust all possible test light colors. If we think of the real plane, we cannot test all possible points (because there are infinitely many) but we know every point on the real plane can reach using [0,1] and [1,0] as basis vectors. I think they are using a similar idea and using 3 primary colors to generate every color in a large set of test lights. As long as our set of test lights is a representative sample of the entire color spectrum (or close enough), we can assume we only need 3 primary lights to span the color spectrum.

psmanohar

One thing you can do is test all possible monochromatic lights: just vary the wavelength of the light source from ~380 nm to ~740 nm in 1 nm increments. Mathematically you haven't tested all possible colors, but scientifically this is enough.

john-b-yang

For a succinct explanation, there's a Udacity video describing the color matching function based on RGB values (part of their computer vision course!) - https://bit.ly/2WSOLuZ. I like how they included a real time example of "stacking" the primary colors and tuning them until they achieved the test light color.

julialuo

It was mentioned in lecture that any three (different) colors can be used as the three "basis" for the color spectrum. This seems surprising to me as it seems like mixing blue, light blue, and dark blue would not be able to yield a red overall color. Also, in the context of linear algebra three vectors can only form a basis if they are linearly independent. Is there an equivalent while discussing colors?

julialuo

Oh welp he talks about it in the next lecture and it's pretty obvious --- linearly independent means any one of the colors can't be a mixture of the other 2.

kavimehta

Thinking about the same properties in another way, the combinations being linear means that we can combine the equalities the same way we combine mathematical equalities

jeshlee121

Ren talks about this during the 3rd Annual Berkeley AR/VR Symposium re. Oz Vision!

How would they know that they had exhausted all possible test light colors?

@RichardChen9 I don't think its possible to exhaust all possible test light colors. If we think of the real plane, we cannot test all possible points (because there are infinitely many) but we know every point on the real plane can reach using [0,1] and [1,0] as basis vectors. I think they are using a similar idea and using 3 primary colors to generate every color in a large set of test lights. As long as our set of test lights is a representative sample of the entire color spectrum (or close enough), we can assume we only need 3 primary lights to span the color spectrum.

One thing you can do is test all possible monochromatic lights: just vary the wavelength of the light source from ~380 nm to ~740 nm in 1 nm increments. Mathematically you haven't tested all possible colors, but scientifically this is enough.

For a succinct explanation, there's a Udacity video describing the color matching function based on RGB values (part of their computer vision course!) - https://bit.ly/2WSOLuZ. I like how they included a real time example of "stacking" the primary colors and tuning them until they achieved the test light color.

It was mentioned in lecture that any three (different) colors can be used as the three "basis" for the color spectrum. This seems surprising to me as it seems like mixing blue, light blue, and dark blue would not be able to yield a red overall color. Also, in the context of linear algebra three vectors can only form a basis if they are linearly independent. Is there an equivalent while discussing colors?

Oh welp he talks about it in the next lecture and it's pretty obvious --- linearly independent means any one of the colors can't be a mixture of the other 2.

Thinking about the same properties in another way, the combinations being linear means that we can combine the equalities the same way we combine mathematical equalities

Ren talks about this during the 3rd Annual Berkeley AR/VR Symposium re. Oz Vision!