Imaging at 5X -- a tale of two objectives

[rjlittlefield]

Since the topic of CA has come up, I thought I'd run a test focused on that issue alone.

The target here is a piece of paper with laser-printed dots, shot with the CFI60 Plan 10X NA 0.25 objective mounted on a Sigma 105 mm f/2.8 macro lens focused at infinity, giving measured 5.15X magnification. Camera is Canon T1i as before. Images are from in-camera JPEGs, no CA removal at any stage.

The full frame:



Toner blobs near corners at actual pixels (cropped from 4752 x 3168):



The above are short stacks, Zerene DMap. I can't see any notable difference from the original images.

Here are some dots from image center, single frames focused behind, on, and in front of the dots.



I read this as negligible lateral CA; longitudinal CA visible but mild; vignetting barely visible.

This all surprised me again. I was happy with the moth scales, but I really didn't expect the dots to look this good.

--Rik

[Craig Gerard]

Am I imagining things or is there a definite 'bright spot' in the centre of the first 'Toner blobs' image? (and possibly some different hues left and right of frame).

Craig

[rjlittlefield]

You mean the full frame?

Yes, the center is brighter than the corners. You may be referring to what I called "barely visible" vignetting. If it's immediately obvious to you, then the wording is wrong. This level of vignetting is something I would not normally notice unless I were looking for it.

About the color, I did notice some variation in overall color from frame to frame. Don't know what caused that, since the camera was set on custom color balance. Flash variation? But I don't know how that could change color from frame to frame.

Usually I process in PMax, which averages out variations in both brightness and color. But these were done with DMap because I wanted the output to locally reflect the input as close as possible while still not having to pick and choose individual frames. So each area of the composite inherited the color of its respective source frame(s). If you do histograms of the crops, you'll see that the RGB peaks for the white paper line up a little differently

--Rik

[Craig Gerard]

Rik,

Yes, I was referring to the full-frame image.

I recognised the pattern. I first began to notice it when taking single images of a micrometre slide, so it was immediately familiar when I saw it in the 'Toner blobs' image.

Thanks for the explanation.

Craig

[ChrisR]

I fear I may have muddied the waters.

Yes, I find the same, that the 10x 0.25 CFI60 objective is as others do, considering my, er, technique.
It's the previous family of CFs which bothers me.
I'm thinking "can they be that bad, or am I doing something wrong?"
As I said, maybe under normal circumstances they're pretty good and we're just looking harder than most.

This is nearly actual pixels from one of the older series of CF, on 200mm. Centre of the pic is about where the red spot is. The yellow printed circle would be 10mm diameter on the sensor, so it's all well within the expected good zone. At the arrowed black chips of pencil lead though, I think I'm seeing blueish inner margins. That's the undesirable lateral CA, isn't it?
Have I now decimated the prices I was hoping to get for the things when I chuck them back on ebay ?

[Craig Gerard]

Chris,

That's an abstract masterpiece

Reminds me of Blue Poles: Number 11 by Jackson Pollock.

In 1973, the work was purchased by the Australian Whitlam Government for the National Gallery of Australia for US$2 million

Craig

[ChrisR]

He needed alcohol and broken glass to produce his, too.

[Blame]

He needed alcohol and broken glass to produce his, too.

Well maybe only the alcohol..... drink enough and the broken glass comes automaticaly.

[rjlittlefield]

A question was asked by email:

So, what do you figure the NA is when used with the 105mm? The same?

Yes, the subject-side NA at 5X will still be 0.25, same as the objective's nominal NA at 10X. That's because neither the working distance nor the entrance pupil changes, so each point on the subject has the same entrance cone as before. On the camera side, the effective f-number does change, from f/20 at 10X to f/10 at 5X.

NA 0.25 at 5X is pretty impressive -- wider than any of Nikon's current 4X or 5X objectives as far as I can see. It's a interesting question how the CFI 10X pushed down would compare to one of the older wide-aperture apochromats pushed up. (I can't answer that question, since I don't have such a lens to test and I'm not curious enough to buy one just to check.)

--Rik

[ChrisLilley]

NA 0.25 at 5X is pretty impressive -- wider than any of Nikon's current 4X or 5X objectives as far as I can see.

This does raise the question of why Nikon provide microscopes with 4x and 10x objectives and a fixed 200mm tube lens, rather than microscopes with a 10x objective and a rotatable upper turret of 200, 150, 100mm tube lenses (or indeed a 200-100mm zoom tube lens).

Answering my own question, innate conservatism in purchasing and working practices in laboratories may well play a large part. Also, the quality requirements for things like routine clinical screening are actually rather modest. If the customer isn't complaining, sell them what they know and like and are used to.

But wouldn't a variable-tube-lens give a better result, for photomacrography at least?

[ChrisLilley]

Since the topic of CA has come up, I thought I'd run a test focused on that issue alone.

The target here is a piece of paper with laser-printed dots, shot with the CFI60 Plan 10X NA 0.25 objective mounted on a Sigma 105 mm f/2.8 macro lens focused at infinity, giving measured 5.15X magnification. Camera is Canon T1i as before. Images are from in-camera JPEGs, no CA removal at any stage.

Thanks for doing these tests; CA (well, false colour of any sort) is something I care about a lot.

Just checking, since I am more familiar with Nikon than Canon - the in-camera JPEG engine provides no suppression of lateral CA at all?

Here are some dots from image center, single frames focused behind, on, and in front of the dots.

Thanks in particular for the axial CA test. I much prefer to use lenses that have minimal secondary colour. The purple (red and blue/violet) is much more visible than the green.

I understand that this 10x objective covers a fairly wide range of wavelengths: Near UV, Visible, IR (380–1100nm). Depending on sensor sensitivity, strength of IR-cut filter in the camera, and the UV absorption of the tube lens there may be some effect from near UV or near IR being detected by the sensor.

On the plus side, people interested in IR photomacrography might find this lens to be of interest.

[ChrisR]

But wouldn't a variable-tube-lens give a better result, for photomacrography at least?

Mitutoyo have zooms in their literature for video use. Would the smaller video camera sensor make things easier?
They also do a 400mm tube lens (MT4 from memory?)
Not sure if you could get a step-down ring for one of these..

[ChrisLilley]

Lots of people make 400mm lenses; they are called +2.5 diopter close-up lenses. No need for a 7 thousand dollar bazooka.

Diopter = 1/focal-length-in-metres = 1000 /f

[ChrisR]

It was a suggestion for amusement, ChrisL.

[rjlittlefield]

This does raise the question of why Nikon provide microscopes with 4x and 10x objectives and a fixed 200mm tube lens, rather than microscopes with a 10x objective and a rotatable upper turret of 200, 150, 100mm tube lenses (or indeed a 200-100mm zoom tube lens)

It seems to me that integrating an adjustable tube lens into a microscope would be far from easy. One of the constraints is that the tube lens has to play nicely with the fixed optics of binocular heads etc.

Remember also that we're getting a little bit silly with our pixel-peeping here. Even at 10X NA 0.25, when I compare what I can see in the captured image with what I can see in an eyepiece view, there is obviously more in the captured image. Push that same NA down to 5X, and the optical image would be far sharper than justified for direct viewing. That would be OK if there were not corresponding costs, but there are: loss of DOF. It actually makes a lot more sense for a 4-5X objective to be limited to around NA 0.10-0.15, since that matches the resolution of an eye looking through the eyepiece, while giving roughly twice the DOF of NA 0.25 .

For amusement, I just now printed the 5X NA 0.25 dots at 300 ppi on a Canon Pixma iP4300 printer, rated at 9600x4800 dpi. The full frame is 15.84" x 10.56 inches. The print does look more crisp in the center than in the corners, but even in the corners, I have to haul out a magnifying glass to see in the print everything that I can see in the pixels. Amazing.

Just checking, since I am more familiar with Nikon than Canon - the in-camera JPEG engine provides no suppression of lateral CA at all?

I don't know if it provides CA suppression for lenses that it knows about, where the CA parameters are known beforehand. But in this case the macro lens is third-party and the camera has no way to know about the added objective. So the only way to remove CA would be matching up the RGB bands within the images as shot. I've never seen that approach work with anything other than carefully constructed calibration targets, so I'd be dumbfounded if the camera is doing it as a matter of course.

--Rik