Trouble grasping numbers and resolution with higher mag

Just bought that first macro lens? Post here to get helpful feedback and answers to any questions you might have.

Moderators: rjlittlefield, ChrisR, Chris S., Pau

Koorosh
Posts: 215
Joined: Fri Jun 20, 2014 3:39 pm
Location: London

Trouble grasping numbers and resolution with higher mag

Post by Koorosh »

Hi everyone,
so I was in the market for a 20x objective, and having spoken to Chris for a while he has mentioned that the difference in resolution between my 10x 0.25 and a 20x 0.4 isn't going to be all that significantly different, but I really don't think I have understood much of what he's said anyway- I find this whole subject confusing and it makes me feel like an idiot!

Chris steered me towards Charles Krebs relay optics excel spreadsheet, but I genuinely don't understand what is being said and what I should be looking at or interested with! Not least because pixel and Pixel appear to be something different to each other! What is going on? Why am I an idiot and why can't I understand this? I read Charles' explanation as well and I am still no better clued in on what this all means with respect to my next lens purchase.

The suggestion with Chris is to maybe look into the Nikon CF 50x elwd with 0.55 NA and shorten the tube length for higher resolution. In some sense I understand this, but I am obviously falling short because it really isn't clicking properly in my head.

I also know of calculations to do with diffraction, but have never performed them and I would imagine they are again relevant here too, and maybe this is what Chris is alluding to with a 10x 0.25 and 20x 0.40?

I'm going for a drink...

ChrisR
Site Admin
Posts: 8668
Joined: Sat Mar 14, 2009 3:58 am
Location: Near London, UK

3D

Post by ChrisR »

Some of that's as I remember it...
The 10x in question is a Mitutoyo planapo NA 0.28,
and the 20x under consideration is a Nikon 20x NA 0.4 ELWD from a rather suspiciously uncooperative ebay vendor.

Sensor is APS.

I referred to a forum search on Diffraction, Cambridge in Color on Diffraction, and Charlie's Relay Optics spreadsheet, where Koorosh correctly interpreted a 135mm "tube" lens on an infinite objective , being rather like a 0.675 relay lens.

I also referred to Rik's fly proboscis picture, showing 25x at NA 0.55 on a Nikon 50 ELWD. Those often come up priced comparitively reasonably.
Chris R

Koorosh
Posts: 215
Joined: Fri Jun 20, 2014 3:39 pm
Location: London

Post by Koorosh »

Maybe it would be better to state my purpose? I am wanting higher resolution for parasitic wasps that I mostly image. For the stuff I image it ranges between 3 and less than 1mm, and I find the 10x is not producing enough detail for body parts etc, for the smaller stuff.
I guess my wonder is if a 20x will be sufficient with a 0.4 NA or if the 50x shrunk to 25x would be a much better idea. And what is it going to look like? Is the point that you were making Chris, that iy it wouldn't look as sharp in spite of greater detail?

ChrisR
Site Admin
Posts: 8668
Joined: Sat Mar 14, 2009 3:58 am
Location: Near London, UK

Post by ChrisR »

If you run the numbers, as suggested in the links I gave, you'll find that your sensor is a reasonable match for a 10x 0.28.
You have about the right pixel density.
If you used a 20x 0.28, you'd spread very much the same detail over twice as many linear pixels.
Marginal improvements only. Empty magnification.
If you go to a 20x 0.42, you get 50% more resolution. You still have some empty magnification.

If you go to 20x 0.56, you get the same number of pixels per resolution limit on subject, as you have with 10x 0.28.

The numbers are somewhat simplistic but I'm sure you can get the idea.

I can't really judge which if any or none you'd find "enough", "much better", or " going to be all that significantly different".:(

If you use eg a 50x infinite objective you can change the magnification to suit your sensor.
If you use an Mplan, it'll cover a bit bigger than APS, but it's finite so you can't benefit from that. If you shorten the tube the quality drops because it's designed tor 210mm

A Mitutoyo 50x NA 0.55 is better for CA and has more working distance, but costs a lot more than the Nikon.
Chris R

rjlittlefield
Site Admin
Posts: 23562
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Post by rjlittlefield »

Koorosh, let me try a different explanation.

Resolution on subject is directly proportional to the NA of the objective. In going from NA 0.28 to NA 0.40, you will gain resolution on subject in the ratio of approximately 1:1.43 (comparing 0.28 to 0.40).

There is a related but different measure for sharpness on sensor. The sharpness on sensor is directly proportional to NA/magnification. In going from 10X NA 0.28 to 20X NA 0.40 at rated magnification, you will lose sharpness on sensor in the ratio of 1:1.40 (now comparing 0.28/10 versus 0.40/20).

On the other hand, if you use relay optics or a short tube lens with that 20X NA 0.40 objective so that it actually gives 10X magnification, then compared to NA 0.28 at 10X you would gain sharpness on sensor by (again) that ratio of 1:1.43 (since we are then comparing 0.28/10 versus 0.40/10 .)

I can't tell, either, what you would find sufficient. But perhaps this will help to explain how the numbers work.

--Rik

Koorosh
Posts: 215
Joined: Fri Jun 20, 2014 3:39 pm
Location: London

Post by Koorosh »

OK I think I am with you both now :)
I am still a little confused by the numbers that came out when I was doing it though. Namely, with the Nyquist statement being two pixels are needed at least in ideal conditions to resolve the smallest detail of the lens. It isn't that which I find confusing, but the numbers that follow when looking at resolving with 3.5 pixels. The number that comes out in the sheet is smaller than if allowing two pixels. Surely if you want a greater number of pixels to allow better detail resolution in non ideal conditions, that number would be a larger value? This is one of the things I'm struggling with.
The second is whether I was reading the right values. I read from the Nikon crop which is roughly the same dimensions as my Sony a6000. I then looked at the number of megapixels required for a goven magnification and NA. If I have read it right, I would have more than enough with 24.3mp. Or have I misunderstood that part?
Lastly, I only have a 135mm so would end up with about 30 odd times magnification. My wonder is if I should consider buying a 100mm or 90mm to add to my rig in order for a lower magnification and higher resolution? Would a 90mm be stretching the 50x too far though?

ChrisR
Site Admin
Posts: 8668
Joined: Sat Mar 14, 2009 3:58 am
Location: Near London, UK

Post by ChrisR »

Which numbers?
It's hard for us to know which ones you're looking at which appear strange. Resolution is in microns, or in lines per millimeter ( times a constant) - one is the inverse of the other
If you could write them here, with words to say how you're interpreting...

Yes you have more than enough pixels, if that's 6000 x 4000 on APS.
Haviong more wouldn't make a difference to the resolution you'll get on your sensor.

Yes going from 10x to 30x is a step in field of view which you may find too much - it comed down to what you need.
If you're going to display them at 1024 pixels wide, then you can crop the 10x objectives images without missing any/much resolution. I mean, at 20x, your image is still 3000 sensor pixels wide. At 30x it's still 2000 so the higher magnification lens takes over.
'Scope objectives traditionally went 10x, 40x, 100x - big gaps!

I doubt you could use a much shorter "tube" lens than 135mm without vignetting. I'll try it if I find time, unless Rik remembers.
Chris R

Koorosh
Posts: 215
Joined: Fri Jun 20, 2014 3:39 pm
Location: London

Post by Koorosh »

I haven't got my laptop to hand, but at the top near where you put the values of the magnification and NA, there are a set of values below and to the right that say 2pixels, 2.5 pixels, 3 pixels etc and a value next to those which I don't understand.
With Riks image of the fly gob at 25x, I presume that was with a 100mm.
I do crop my images for the little stuff that doesn't fill much of the sensor, but I don't find the resolution all that exciting, and in many instances the details are too soft to capture many of the smaller details necessary for identification purposes. Actually, I often have to resort to using my rig to image because the detail is far better than my stereo microscope can pick up, which probably isn't surprising gicen the difference in NA. For the really small stuff though, I do need more magnification. 30x will probably be too much (f/3.5 minimum for my lens- will that make matters even worse for higher mag?), but 25x seems a little closer to what I need. If so I need to source a 100mm m42 lens ideally, which will push expenditure up, but I need one anyway :)

Koorosh
Posts: 215
Joined: Fri Jun 20, 2014 3:39 pm
Location: London

Post by Koorosh »

Ok I understood the first set of numbers now! I was getting confused with number of pixels and pixel SIZE. If the minimum resolvable detail of an objective is 10um, then if following the Nyquist rule, you would need 5um pixels for the 2 pixel rule. So if you have 3, you need 3.33um pixels and so on.

My apologies for the melodrama- as always I really do appreciate your help, and thank you for being patient with me. Chris particularly on this occasion. :)

Dare I ask where next to go to work out a little more in depth as to what detail is being lost due to diffraction, and perhaps some other factors to consider?

Particularly with my f/3.5 135mm lens as an example, can I expect a considerable difference if I were to use f/2.8?

I stumbled across this site, although maybe you mentioned it Chris and I forgot that you did? In any case, I have really enjoyed learning from it, although it is somewhat basic (good for beginners like me though).

http://micro.magnet.fsu.edu/primer/digi ... asics.html

ChrisR
Site Admin
Posts: 8668
Joined: Sat Mar 14, 2009 3:58 am
Location: Near London, UK

Post by ChrisR »

Dare I ask where next to go to work out a little more in depth as to what detail is being lost due to diffraction, and perhaps some other factors to consider?
To be honest, it's my belief that if you look at the basic numbers as you have, then assume the world is likely to be a slightly worse place, and things are more or less borne out by other people, then I don't think there's a lot of point - unless you enjoy it.

Your pixel "dead space" will lose you a bit, the AA low pass filter will lose you a bit, and so on. I have in mind a thread here where someone analysed such losses (for an MP-E65) which was interesting but wasn't something I thought worth putting into my estimates.
Lighting makes a huge difference.
Particularly with my f/3.5 135mm lens as an example, can I expect a considerable difference if I were to use f/2.8?
I wouldn't. If it's not vignetting, then fine. I have Nikkors 135 f/2.8 and 3.5, & no difference that I've noticed in quality.
Rik may have used a 100mm tube lens, and it may have vignetted, but it wouldn't matter, in the sense that if you have more than enough pixels at that magnification then you've captured everything the objective can give. You just enlarge the good part of the image to whatever you need it for.

From 10x it looks like a big step to 50x, but I don't know of a good intermediate option for NA and WD. Sure, if an apo 20x pops up at an affordable price, go for it!
Chris R

Koorosh
Posts: 215
Joined: Fri Jun 20, 2014 3:39 pm
Location: London

Post by Koorosh »

OK so I guess my next question is about lighting. Will shining janjsos through a Styrofoam cup suitably diffuse light enough for the 50x?
And I also apologise for misquoting you from our messages- I completely misunderstood something and it was just a mistake with my understanding, not yours. Using actual quotes would be far handier but I usually post from my phone where it is a bit of nightmare.
I will leave the other calculations for now but I would like to understand the theory a lot more because I do enjoy it and find it very interesting. Perhaps for now it would be better to focus my energy on understanding lighting techniques, backgrounds and diffusion.

ChrisR
Site Admin
Posts: 8668
Joined: Sat Mar 14, 2009 3:58 am
Location: Near London, UK

Post by ChrisR »

I can't use continuous lighting at 50x - vibration.
Flash yes, and you can try letting let a little of the flash creep over the edge of the cup so it bounces around inside, but it mustn't hit the subject directly of course.
Chris R

naturepics43
Posts: 88
Joined: Mon May 11, 2015 9:21 pm
Location: Hocking County, Ohio , USA

Post by naturepics43 »

Koorosh wrote:Maybe it would be better to state my purpose? I am wanting higher resolution for parasitic wasps that I mostly image. For the stuff I image it ranges between 3 and less than 1mm, and I find the 10x is not producing enough detail for body parts etc, for the smaller stuff.
I guess my wonder is if a 20x will be sufficient with a 0.4 NA or if the 50x shrunk to 25x would be a much better idea. And what is it going to look like? Is the point that you were making Chris, that iy it wouldn't look as sharp in spite of greater detail?
If I'm out of line posting on this thread, please advise me & delete it.

As a beginner myself to high magnification focus stacking, I spend a lot of time on this great site looking for information that may help me achieve my end goal which I think is similar to yours, Sharp, detailed images of small (less than 3 mm) insects. If I may, I have a couple of questions:

1. What body parts of the wasp are you wanting to look at? Antenna segments, wing venation?
2. Could you post a couple of your images?
3. Are your specimens mounted or are you using a vertical rig & laying the specimen on the stage?

I have had limited success stacking a mounted 0.75 mm parasitic wasp but am easily able to zoom in & view the shape, size & count the antenna segments. I use a Nikon D7000 (APS-C, 16.2 MP sensor) with a 10X objective. Lighting has been my biggest hurdle.

Koorosh
Posts: 215
Joined: Fri Jun 20, 2014 3:39 pm
Location: London

Post by Koorosh »

Hi there, sorry I hadn't received a notification for responses on this- must have accidentally clicked the unsubscribe link instead of following the post!
I am taking bits and bobs of everything really- for smaller stuff, I wanted higher magnifications to see some actual body parts, because I think a lot of the time the keys used are asking for details on wasps so tiny that at 75x I can barely make out, let alone judge! Antennae will usually be OK, but mandibles, unless they're conspicuous in form, no chance.
What exposure times are you using? And continuous? What diffusion are you using? And step size?
I posted a few of my stacks in the study macro, I think titled 'some stacks of mine (about time).' They are my better ones though :) the Lamprtotatus face is at 10x with 7um steps. The mandibles on it are obvious, but that is quite large for Parasitica. A lot of the stuff I image has heads that fill about a quarter of the frame, and the detail isn't impressive or useful a lot of the time when cropped.
I've just got a 50x which I need to do some runs with. Using it on a 100mm to get a more suitable mag, and preliminary shots suggest it could do the job nicely, but I reserve my judgement for the time being :)

naturepics43
Posts: 88
Joined: Mon May 11, 2015 9:21 pm
Location: Hocking County, Ohio , USA

Post by naturepics43 »

Koorosh wrote: I am taking bits and bobs of everything really- for smaller stuff, I wanted higher magnifications to see some actual body parts, because I think a lot of the time the keys used are asking for details on wasps so tiny that at 75x I can barely make out, let alone judge! Antennae will usually be OK, but mandibles, unless they're conspicuous in form, no chance.
I see your point. At 10X, the mandibles on a 1mm wasp would difficult to see clearly.
Koorosh wrote:What exposure times are you using? And continuous? What diffusion are you using? And step size?
I'm using ISO 100 @ 1/200 using 2 Nikon SB-800 flash units, usually around 1/16th or 1/32nd power. My diffusion at the moment is 1 layer of a paper towel surrounding the specimen. As to step sizes, for my 10X NA 0.28 I use .005mm steps and for my 10X NA 0.20 I use .011mm steps.
Koorosh wrote:I posted a few of my stacks in the study macro, I think titled 'some stacks of mine (about time).' They are my better ones though :) the Lamprtotatus face is at 10x with 7um steps. The mandibles on it are obvious, but that is quite large for Parasitica. A lot of the stuff I image has heads that fill about a quarter of the frame, and the detail isn't impressive or useful a lot of the time when cropped.
I've just got a 50x which I need to do some runs with. Using it on a 100mm to get a more suitable mag, and preliminary shots suggest it could do the job nicely, but I reserve my judgement for the time being :)
I've been following your other posts with great interest. Your posted images, as stated by Pizzazz & rjlittlefield, are very good. Without scale bars, it's difficult to determine the size of the posted insects.

Are your smaller wasps ( 1mm - 3mm) point mounted or do you use a different method? I would really like to see your attempts at a 1 or 2mm wasp. Mine have not been very successful due to the clear adhesive holding the wasp on the point.

Post Reply Previous topicNext topic