… Sigma and Wabi Sabi

Peter Slovakia wrote:

Hi DMillier.
It's here, the first objector and it is clear to me that you take photos with a camera with a Bayer sensor.
Well, I didn't write which color is better, because Bayer can't read color, he's as blind as a patron, he can't see any colors. Everything I wrote about Bayer and Foveon sensors is principled and cannot be questioned in any way, because they are facts. So this is not about some "human methods" of comparison. The Bayer sensor is color blind and done! It records only brightness. It has only one layer, in which one point is divided into three, so it does not reach the image density and therefore not the point sharpness as Foveon has, for the same reason Bayer is not able to produce a 3D image, because it needs more layers of sensing stored under each other - as film had and as Foveon has.

I only stated the facts, not my opinion. For me closed. I will not continue the discussion on this topic. My previous reply was to Don, not you, my friend.
Peter, Slovakia

Edited:

Peter

Are you here to state your opinions to be absolute facts? Or to discuss?

I have a SD9 and a SD14 and used to have a DP1 and a DP2M. That's that cleared up. You are so keen to launch to the attack that you didn't even bother to understand my point (which had nothing to about taking sides as to which sensor was better, as you seem to think).

Also, some of your so-called facts can be challenged. For the record, there is nothing that can detect colours directly because colours don't exist as a physical property, they are purely constructs of the brain. Light has a number of properties: wavelength, frequency, intensity, polarisation etc, but colour isn't one of them.

Sensors, Bayer and Foveon based (and retinas), detect intensity filtered into bands of wavelengths. Bayer sensors use organic dyes to split the wavelengths into 3 bands, Foveon uses silicon depth as the filter. The filtering methods are different kinds of filters, but the concept is the same: measuring the intensity of 3 bands of wavelengths.

The practical similarity is that both approaches end up with one red, one green and one blue measure for each spatial pixel in your file. Once you have colour triplets, the brain can compare the relative strengths of each of the components and assign the result to invent the perceptual experience of "colour" but it isn't a physical thing.

The advantage of Foveon is that each of the three detectors are located at the same spatial location on the 2D grid, while the Bayer detectors are offset and require some clever image processing to recover the full colour triplet. The absence of this colour processing gives Foveon higher acutance and better colour resolution for a given number of 2D grid pixels. The disadvantage is that the silicon filters are not really aligned with the colour perception of the human eye and the processing required to convert to a reasonable colour space adds noise and the occasional colour misfire.

So, advantages and disadvanges to both, you pick your poison.

But nothing in the above has anything to do with my previous post. That was all about well researched facts that go way beyond image editing. If you want to understand subjective perceptions across a broad spread of people (rather than just yourself), the only game in town is to do research using double blind methods to remove observer bias.

For example, if I were investigating whether 50 people taken at random could tell the difference between Foveon and Bayer photographs, I might start by providing 50 prints, half taken with a Foveon, half with Bayer and asking my test subjects to say whether there was anything about the pictures that would cause them to divide the prints into two groups. And to please sort the pictures into two groups.

If the differences were purely imaginary, you would expect people to split the pictures into two random groups. But if (say) there was a big and obvious difference between Foveon and Bayer, you would expect to find all the foveon images selected together and all the Bayer images selected together.

It the latter result occurred and the exercise was well designed and executed under purely blind conditions, it would be suggestive that randomly selected people can distinguish an obvious difference between the sensor types. You'd then be able to dream up ever more subtle tests to explore exactly what it was people were detecting.

I'd love to participate in such an exercise, wouldn't you?

ps

In the ongoing leadership of the UK Tory party/Prime Minister selection competition, a pollster ran a survey to find out how well known the candidates were. They included a made up fake candidate in the survey. 12% of respondents considered the fake candidate to be well known. Makes you think, doesn't it?

https://www.indy100.com/politics/stewart-lewis-tory-leadership-poll

We can't trust our brains.