CMY or RGB ... a mystery

[MichaelG.]

I have been investigating some low-power objectives, and have a rather surprising result

The attached image is a tiny crop, 100 x 80 pixels, from a 2592 x 1944 frame.
Objective = Zeiss 3.2x Achromat
Eyepiece = Zeiss Kpl 8x

Afocal imaging with 16mm lens on Toupcam 5MP
.

This looks like: CYAN.MAGENTA.YELLOW

22DD7FFF-8244-414D-9855-AC42A6A8AA2B.jpeg (5.86 KiB) Viewed 6613 times

.
The cropped detail is at the edge of the frame and [predictably] shows both Chromatic Difference of Magnification
and out-of-focus effects ... BUT the mystery is: Why does the CDM produce Cyan, Magenta, and Yellow spots ?

My understanding is that Cameras and monitors use Red, Green, and Blue
... CMY being for printers
http://daveoverbeck.com/webcolor/01.html

Why, in this jpeg file are we seeing CMY artefacts, not RGB ?

There is presumably a simple answer to which I am oblivious
... if anyone can explain, I really would be most grateful.

MichaelG.

[MicroBob]

Hi Michael,
the image on the sensor covers several pixels so any colour that is sensed at this spot can be recorded. The Zeiss 3,2 achromat is a simple seeking objective not intended for photography. probably it projects a colour range that is rainbow-coloured at the border of the immage. If the few colour corrections in this objective shift some wavelenghts to the side this might just lead to intensity peaks in the three colours you see.
What illumination was this?
If you would use the Plan 2,5 you would still have no answer for your question but might forget the question over time!

Bob

[Hobbyst46]

Hi Michael,
I am 99% sure that it is the selected format output of the specific image capturing software or pp software if you used one. I have seen this RGB/CMY output selection window in some software lately - Toupview perhaps - ca'nt find it now. But it must be a software artefact.

[MichaelG.]

Hi Michael,
the image on the sensor covers several pixels so any colour that is sensed at this spot can be recorded. The Zeiss 3,2 achromat is a simple seeking objective not intended for photography. probably it projects a colour range that is rainbow-coloured at the border of the immage. If the few colour corrections in this objective shift some wavelenghts to the side this might just lead to intensity peaks in the three colours you see.
What illumination was this?
If you would use the Plan 2,5 you would still have no answer for your question but might forget the question over time!

Bob

Thanks for your thoughts, Bob ... but I struggle to follow your explanation.

MichaelG.
.
P.S. ...My investigation is an ‘academic exercise’ for my own education: I am comparing the performance of this 3.2x Achromat against my Zeiss 4x PlanApochromat in the hope of understanding the aberrations of the achromat.
.
These first two shots have no great merit, but may be of interest:
https://www.dropbox.com/s/pr5o7u0qbsdc0 ... B.jpg?dl=0
https://www.dropbox.com/s/tqe6cq43y9tzr ... B.jpg?dl=0
The speck is on the left margin of the first image, but out-of-frame on the second

[MichaelG.]

Hi Michael,
I am 99% sure that it is the selected format output of the specific image capturing software or pp software if you used one. I have seen this RGB/CMY output selection window in some software lately - Toupview perhaps - ca'nt find it now. But it must be a software artefact.

Thanks, Doron ... I suspect that might indeed be the case

I am rather wary of jpeg processing, and will try to proceed with a different file-format.
These were just quick comparison shots, and I [perhaps foolishly] accepted the ToupLite defaults.

MichaelG.

[Hobbyst46]

Michael,

I spotted a dot in the north-east part of the images, along the "one o'clock" radial ray and near top border of the image. And extracted it with IrfanView within a 63X63 pixel square.
Here it shown, from the two objectives. I think that lateral CA (CDM) is evident in the 3.2X achromat, and none in the 4X apochromat image, not so ?

[MichaelG.]

Michael,

I spotted a dot in the north-east part of the images, along the "one o'clock" radial ray and near top border of the image. And extracted it with IrfanView within a 63X63 pixel square. ...

An excellent ‘find’ ... Thank You

I am very happy with the combined correction of PlanApo and Kpl

More photos in due course.
MichaelG.

[MicroBob]

Hi Michael,
in lhe lower left corner of your Planapo-image I found the sharp contrating detail I have enclosed in a enlarged form for better visibility.
There still is some colour error and there are these chessboard like artefacts. Of both I think this is not what the Planapo can offer.
Can you show a picture of an object micrometer? What condenser and camera setup do you use? How was the compression setting for the jpg-file?

Bob

[MichaelG.]

Hi Michael,
in lhe lower left corner of your Planapo-image I found the sharp contrating detail I have enclosed in a enlarged form for better visibility. [ ... ]

Thanks, Bob ...but fear of this level of scrutiny is why I only posted a specific question, and did not originally share the images. ... They are not good enough for such forensic analysis.

• They were taken on a lashed-up assembly, and the tube-length is about 176mm instead of 160

• The camera is not rigidly aligned with the eyepiece, and may be slightly off-axis

• The illumination is very basic … an LED ‘dome’ under a Wild M11 condenser

• The ToupLite software is new to me, and I don’t really trust jpeg image processing anyway

• With a frame rate of less than 3fps, it is also extremely difficult to focus the image on-screen

When I can get a more suitable test-bed configured, I will share some better images
... until then, I think your generous efforts are being wasted.

Thanks again
MichaelG.

[MichaelG.]

Hi Michael,
in lhe lower left corner of your Planapo-image I found the sharp contrating detail I have enclosed in a enlarged form for better visibility.
There still is some colour error and there are these chessboard like artefacts. Of both I think this is not what the Planapo can offer.
Can you show a picture of an object micrometer? What condenser and camera setup do you use? How was the compression setting for the jpg-file?

Bob

Bob,

I was intrigued by your 'chessboard like artefacts' ... so I compared your enlarged image with a simple enlargement [using Preview on the Mac] of that area in the file that I posted.

The image here https://www.dropbox.com/s/mp8r0mks986j3 ... .tiff?dl=0 is a screen grab of the pair side-by-side ... Please forgive the slight size discrepancy; it would have been tedious to match them exactly.

It seems reasonably apparent that the localised 'chessboard like artefacts' in your image are the result of whatever re-processing you performed ... whereas the original image shows very clear pixellation throughout.

Note: I have left the screen-grab image as a tiff file, to avoid any further re-processing.

It may be a few weeks before I can post any 'reliable' images that are worthy of analysis ... but meanwhile, thanks again for your interest.

MichaelG.

[MicroBob]

Hi Michael,
you came upon an interesting aspect: These image details are enlarged beyond useful measure. On my Windows 8.1 laptop there seems to be a system function that "enhances" the result instantaneously, while your Mac just shows the big pixels as they are. This enhancement has worked well nearly everywhere, but in some places it didn't step in or even increased the pixelation. I tried this with Irfan view (quite good programm for light image editing) and the basic Windows image editor too - same result.
Then I took your un-enhanced version and tried to give it a smoother look with GIMP. I employed Gaussian softener and got rid of the pixels but retained the information that was in the image. Then I tried to remove the "dreamy" look by employing unsharp masking. As soon as the settings led to more visible local contrast the pixels in just the same places came back!

While normally one doesn't enlarge images that far it is good to keep this in mind for the day it is necessary.

This is a very present problem: Photographs are increasingly the result of simple image capture and digital "improvement". Increasing or reducing depth of field, sharp free hand pictures in very dim light, resoultion and colour that is based on artificial intelligence, lens abberation correction... This all works very well - up to a point. Now try to find this point!

Bob

[MichaelG.]

This is a very present problem: Photographs are increasingly the result of simple image capture and digital "improvement". Increasing or reducing depth of field, sharp free hand pictures in very dim light, resoultion and colour that is based on artificial intelligence, lens abberation correction... This all works very well - up to a point. Now try to find this point!

Thought for the day, folks ^^^

MichaelG.
.
.
Footnote: As the great Lou Reed says, in ‘The last great American Whale’

They say things are done for the majority
Don't believe half of what you see
and none of what you hear

[ the New York grammar is suspect, but the message is there ]

[MicroBob]

Don't believe half of what you see
and none of what you hear

In the distant past it was always the problem to gather information. The internet brought easy access to information but the selecting became a problem. Today we more and more have the problem what to trust and to what degree. The same methods that are used to enhance images are used to enhance sound. This always brings the risk of artefacts. With information it is the same: We have to live with the fact that it comes to us in "enhanced" form and sometimes there is a tiny little artefact here or there.

Lets see what the NEXT problem will be!

Bob

[MichaelG.]

I may have deduced the answer to my opening question !
...Could someone please review my thinking ?

Let’s assume that there is actually a black dot on a white background
... a reasonable approximation of what should be seen in my ‘tiny crop’

We know that the Achromat will suffer Chromatic Difference of Magnification
We know that the imaging channels are Red, Green, and Blue

If the dot is sufficiently small, its Red, Green, and Blue versions will be easily identified
... this is the case in my ‘tiny crop’

NOW TO THE TRICKY BIT:

If the dot is black, then those channel versions will be:

• Red = 0

• Green = 0

• Blue = 0

But these are overlaid on a white background !

So, at those locations we will see:

• Red = 0, Green = 255, Blue = 255

• Red = 255, Green = 0, Blue = 255

• Red = 255, Green = 255, Blue = 0

Which equate to:

• Cyan

• Magenta

• Yellow

and ... I hope it is obvious that the presence of a White background will reduce the saturation of those colours.

Q.E.D.

Please correct me if I am wrong ...but it works for me.

MichaelG.

[Hobbyst46]

Q.E.D. IMHO too.
You certainly extracted more information from that humble 3.2X than what Zeiss had expected in the 1930's...

[MichaelG.]

Lets see what the NEXT problem will be!

Unless you were thinking of world politics, Bob [in which case, all bets are off] ... I think this might be it:

Slo-mo is the real stand-out here, at least on the page – 960fps has become the high-end mobile standard for slo-mo video, but the Mate 30 Pro goes up to 7,680fps, for 256x slo-mo.
This, unsurprisingly, uses a whole lot of interpolation. That is where you take two consecutive frames and then add one or more interstitial frames, using clever algorithms to guess what the content of them might be were they were not fabricated.
W

.
https://www.wired.co.uk/article/huawei- ... pro-review

Having worked with genuine high-speed cameras [both film and video] ... I have to wonder whether there is ANY technical use for a video comprised mostly of interpolated images.

MichaelG.

[MicroBob]

technical use

Hi Michael,
they probably have more common uses in mind, e.g.youtube videos with titles like "What happens when I drop my tomato"

A non-"enhanced" slow motion function is useful though. You can buy LED lamps today that flicker heavily, but with a frequency a human can just not quite see. This is easily detectable with a slow-motion video.

Bob