Thom Hogan 2011 Predictions

[pico]

Advertisement (gone after registration)

Update: Thom Hogan's email response from Jan 13, 2011, 06:09 CST:

 

I guess the next question would be:

 

What are the commonalities in the production of all Bayer arrays which would lead to variances in green filtration?

 

Have at it, technical conspiratorialists!

 

Two different hues of green in the same sensor. Different patterns, such as Kodak has made, and finally, an extra pixel that has no filter - used to detect luminance w/o hue.

[mjh]

Two different hues of green in the same sensor. Different patterns, such as Kodak has made, and finally, an extra pixel that has no filter - used to detect luminance w/o hue.

But these sensors haven’t found widespread acceptance. None of the major camera vendors are using sensors with Kodak’s panchromatic pixels and Sony’s 4C sensor with “emerald” pixels found use in exactly one model, more than 7 years ago. So what is that “dirty secret” the vendors should come clean about?

 

Maybe I should start spreading a rumour: Hogan just throws us a few crumbs (i.e. cryptic remarks not explaining anything) because he is actually in on it – he’s one of Them!

[stunsworth]

My personal experience is that if there's a colour that can sometimes be problematic with digital sensors it's red rather than green.

[mjh]

Of all the colours along the visible spectrum, violet is the only one beyond the red-green-blue range, so strictly speaking it should be impossible to tell violet from blue, just going by the output of an RGB sensor. Surprisingly most cameras manage to cope anyway, even when violet is notorious for being a tricky colour to get right.

 

The main issue with red is that saturated reds can easily overexpose the red channel, resulting in artificial looking colours. The same can happen with blue, though, cyan-coloured skies being the prime example for this issue.

[ho_co]

Website statement speaks of "mismatching greens in Bayer filtration." Compare that with the much weaker formulation in Thom's first response:

... it’s difficult to make the greens match the way that the Bayer layer is currently created. The RG row tends to have a slightly different green than the BG row.

[emphasis added]

 

Is the Bayer pattern laid down in the same process with CMOS and CCD sensors? (There's only one "Bayer pattern," right?) If the application of the filter layer is done differently by Kodak, say, and by Sony, or on CMOS and CCD, then there's an inconsistency in Thom's logic.

 

If the G's in the red rows are different from those in the blue rows, is that of any real consequence? Aren't the G values in fact averaged in producing an image? No one believes that the colors of a print or a transparency or a digital sensor reproduce reality.

 

On that point, how do we know that all B's or all R's are the same?

 

If we build another pattern like the Sony RGBC, are we sure that all filters of a given color pass identical wavelengths?

 

Are we certain that all pixel cells (i.e., the cells of the photoresponsive substrate beneath the filter) have the same sensitivity to begin with?

 

Aren't all these concerns taken into consideration by the camera and in post-processing?

 

 

In other words, I think we're into esoterica. Or less.

 

I'd like to hear from someone familiar with Bayer chip construction (or familiar with the deposition of any filters on silicon) on the matter.

 

So far, none of the companies whose alternative arrays have been mentioned claimed that their arrays "eliminate the mismatched green problem of the Bayer production."

 

 

Definitely an interesting idea; I'd like to know more. But at the moment, I think calling it a "dirty little secret" inherent in today's techniques of "creating" Bayer arrays may be a bit overextended.

 

Indeed,

... he’s one of Them!

[ho_co]

... luminance w/o hue.

pico, I say hueless is clueless!

 

That's why even Kodak isn't using that design...

[mjh]

Advertisement (gone after registration)

If the G's in the red rows are different from those in the blue rows, is that of any real consequence? Aren't the G values in fact averaged in producing an image?

The green channel is generally used for creating a luminance image and since (a) there are twice as many green-sensitive pixels than red- or blue-sensitive pixels and ( our impression of sharpness mostly depends on the luminance channel, this serves to maximize sharpness.

 

Now suppose the transmission curves of the green-sensitive pixels in the red and blue rows or columns did differ, i.e. there were green1 and green2 pixels being a slightly different green. These two kinds of green pixels would then yield a different response if confronted with most solid colours. This colour would usually be closer to the maximum of one of the greens than to the maximum of the other, resulting in a checkerboard pattern in the luminance channel. This pattern would be difficult to get rid of since depending on the exact colour either green1 might yield a stronger response than green2, or it might be the other way round.

 

Now of course one could use two different greens for improving colour renditition within the green part of the spectrum, provided one was prepared to let go of the resolution currently possible because 50 percent of all pixels are the same colour. You cannot have both the resolution expected from a Bayer sensor and at the same time the improved colour rendition only a four-colour sensor would be capable of.

[pico]

As this discussion goes on, Fovean technology is more appealing now than ever.

[Nicoleica]

As this discussion goes on, Fovean technology is more appealing now than ever.

 

+ a great big fat 1. :D

[Riley]

Only there are no secrets, dirty or otherwise. Sony at the time made a big fuss about their non-Bayer pattern (only to dump it immediately after …), as did Kodak with their panchromatic pixels. In any case I suppose it’s just a conspiracy theory, but at least I would like to know which conspiracy theory – this seems to be one I’m not familiar with.

 

there was this

An Open Letter To The Major Camera Manufacturers

[mjh]

there was this

An Open Letter To The Major Camera Manufacturers

That was about a completely different issue. If only someone believing that camera vendors were conspiring against us (and I don’t mean by designing products were are lusting after, eager to part with our money just to get one – that much is taken for granted) would compile a table of who is supposed to be guilty of what. With easy to understand examples, if possible.