Interesting, does it work? I was also thinking that one could use the normal equation based on XYZ to get the seed matrix in the ballpark... The seed matrix could be calculated before calling the matrix finding routine and passed to it as a parameter.
Yes, but it doesn't seem to make things faster. That is counterintuitive. I need to check my work.
That's gonna be faster, but I'm looking for something that will allow CFA optimization in a way that's more general than the SMI way.
It's the physics.As noted there are infinite other ways to define the functional mapping. However, thephysicselectronicsof the device constrains what form one
The physics has to do with the way the universe works, not an inner product.
There are a few good reasons why data sheets contain what they contain. One is consistent and repeatable design and testing protocol, the other is compatibility. There are others, too.
Yet dimensional analysis is a strong indicator.It's the physics.As noted there are infinite other ways to define the functional mapping. However, thephysicselectronics
The physics has to do with the way the universe works, not an inner product.
Yup, it's where physicists start for a back of the envelope conjecture. Helps that these days we have our units sorted.Yet dimensional analysis is a strong indicator.It's the physics.As noted there are infinite other ways to define the functional mapping. However, thephysicselectronics
The physics has to do with the way the universe works, not an inner product.
I think the other reason is what they are designed for. If you're trying do design a device to micmic a human visual response, photometric units is a good start - you wouldn't want to be saddled with unit conversions right at the start.
So, if we multiply by something, and divide by the same afterwords it is not an indication that we may have our formula wrong/redundant; and consequently, our understanding doesn't need some adjustment.Yup, it's where physicists start for a back of the envelope conjecture. Helps that these days we have our units sorted.Yet dimensional analysis is a strong indicator.It's the physics.As noted there are infinite other ways to define the functional mapping. However, thephysicselectronics
The physics has to do with the way the universe works, not an inner product.
One is consistent and repeatable design
I am not quite sure what your point is here.
Relative QE. It needs to be superimposed over responsivity data.
µV/e- is output sensitivity.
No, it is not provided.
Relative QE.
Right. I guess one can add additional filters like IR-cut etc. on top of that. What I find interesting in the context here is that the green filters typically peak earlier than one would expect and blue filters are stretched out a lot more than the others. That makes sense because it's the one that has the most to gain, starting from such low ideal values.
What's missing, other than the UV/IR/AA filters?
"combined CFA + sensor spectral response" is missing. QE is no replacement for that.
With the code above, you need to drop the transposition of M a couple of lines further down. I could change it so that you don't have to do that, and I will if you want, but that will make it slightly slower.
What you really want to say is that physics determines the inner product they use.
Have you followed the links I provided?
Absolute QE is specified for KAF8300.
Sure. It is the one place I would expect photometric units to be used.
and
I have not seen spectral responsivity specified in V/lux-sec, and provided three counter examples.
Again, as far as I can tell, the examples I provided are all absolute QE for the combined sensor response + CFA transmission spectrum.
I think you overrate the word absolute here, data is in percent if you look at the Y axis.
Apologies -- I was looking at the 2012 Truesense version of the data sheet, rather than the 2015 ON Semiconductor 2015 version, where the same graph appears on page 11.
Quantum efficiency is defined as electrons released per incident photon. The theoretical maximum QE is 1.0, or 100%, at least in the regime of interest where multiple electrons per photon is highly unlikely.
