Reduce magnification

Have questions about the equipment used for macro- or micro- photography? Post those questions in this forum.

Moderators: Chris S., Pau, Beatsy, rjlittlefield, ChrisR

Lou Jost
Posts: 6322
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Re: Reduce magnification

Post by Lou Jost »

Let me explain what I meant by "matching lens and sensor resolutions", since mjkzz may be misunderstanding this, and we don't want to be arguing about words, we want to argue about optical phenomena.

When people (on this forum at least) talk about matching sensor resolution to lens resolution, they usually seem to mean choosing a sensor whose resolution equals the resolution of the lens.

My point is that this always causes a significant loss of resolution in the final image. To recover most of the resolution of the lens, you must use a sensor that significantly outresolves the lens.

If this were in film days, the formula I gave would be correct and would show you exactly the trade-offs involved. Economics, not optical laws, would decide how you make those tradeoffs.

I do not know how to modify the formula for digital sensors. But in general, a chain of noisy information channels will always produce a noisier result than any element of the chain. A lens + detector channel in which both components are equally noisy will produce a final image much noisier than either channel alone.

Oversampling is always good, but of course leads to diminishing returns eventually.
Last edited by Lou Jost on Mon Jan 20, 2025 7:28 pm, edited 1 time in total.

lothman
Posts: 1046
Joined: Sat Feb 14, 2009 7:00 am
Location: Stuttgart/Germany

Re: Reduce magnification

Post by lothman »

Lou Jost wrote:
Mon Jan 20, 2025 6:43 am

When people (on this forum at least) talk about matching sensor resolution to lens resolution, they usually seem to mean choosing a sensor whose resolution equals the resolution of the lens.

My point is that this always causes a significant loss of resolution in the final image. To recover most of the resolution of the lens, you must use a sensor that significantly outresolves the lens.
hmh but which sensor do we have to measure the best lenses and which lenses do we have to measure the best sensors? This discussion makes me thinking?

As shown in my thread about the Sigma 105mm macro the lens seems to outresolve the 60MP sensor of my Sony A7R5 since pixelshift images show considerably more detail. So what is the resolution of this lens and how could we measure it?

Lou Jost
Posts: 6322
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Re: Reduce magnification

Post by Lou Jost »

lothman wrote:
Mon Jan 20, 2025 10:06 am
Lou Jost wrote:
Mon Jan 20, 2025 6:43 am

When people (on this forum at least) talk about matching sensor resolution to lens resolution, they usually seem to mean choosing a sensor whose resolution equals the resolution of the lens.

My point is that this always causes a significant loss of resolution in the final image. To recover most of the resolution of the lens, you must use a sensor that significantly outresolves the lens.
hmh but which sensor do we have to measure the best lenses and which lenses do we have to measure the best sensors? This discussion makes me thinking?

As shown in my thread about the Sigma 105mm macro the lens seems to outresolve the 60MP sensor of my Sony A7R5 since pixelshift images show considerably more detail. So what is the resolution of this lens and how could we measure it?
If you want to measure just the lens resolution, you can't put any sensor into the system unless the pixels were infinitely small. But you can easily measure the resolution of the aerial image by photographing it with a high-NA low-m microscope objective (essentially giving you nearly infinitely small pixels). I used that technique to show that my f/1.0 85mm Repro-nikkor outresolves my sensor at m=1

IMAGOμM
Posts: 18
Joined: Sat Aug 24, 2024 12:33 am

Re: Reduce magnification

Post by IMAGOμM »

lothman wrote:
Mon Jan 20, 2025 10:06 am

hmh but which sensor do we have to measure the best lenses and which lenses do we have to measure the best sensors? This discussion makes me thinking?

As shown in my thread about the Sigma 105mm macro the lens seems to outresolve the 60MP sensor of my Sony A7R5 since pixelshift images show considerably more detail. So what is the resolution of this lens and how could we measure it?
This is precisely the point at issue:
Let's take a practical example: a Mitutoyo HR 10x resolves (according to the manufacturer's specifications) approximately 160 lp/mm. The sensor of a Nikon Z8 camera resolves approximately 160 lp/mm with pixel shift. The followers of the equivalence lens resolution = sensor resolution say: "the combination is perfect, and the result will be a resolution of 160 lp/mm of your image". The followers of the sensor resolution > lens resolution say: "no, the image will not have a resolution of 160 lp/mm, but less. To obtain this resolution you must use a sensor with a lower pitch". Yes, but by how much lower? It is clear that you cannot decrease the pitch infinitely in an attempt to capture all the resolution of the lens. There must be a limit. This limit can vary depending on the needs of the photographer: a professional, who must take into account costs, the amount of work and so on, can be satisfied with a lower limit; those who photograph for pleasure can expect more. The point is that, apparently, there is not a study, a research that has developed a curve that represents the trend of the resolution transfer of a lens as a function of the sensor pitch. Indeed, some say, this curve does not exist because there is no alternative to perfect coupling. Perhaps a law that, without confirmation, seems dictated by a capricious god. The minimum pitch enthusiasts have a calculation formula that dates back to the analog era, and that worked well. But is it applicable to digital? That is the question...
Last edited by IMAGOμM on Tue Jan 21, 2025 6:48 am, edited 1 time in total.

chris_ma
Posts: 591
Joined: Fri Mar 22, 2019 2:23 pm
Location: Germany

Re: Reduce magnification

Post by chris_ma »

IMAGOμM wrote:
Mon Jan 20, 2025 10:58 am
The point is that, apparently, there is not a study, a research that has developed a curve that represents the trend of the resolution transfer of a lens as a function of the sensor pitch.
Actually I'd be very surprised if there was no research which came up with pretty good modelling of how lens resolution and sensor resolution results in total system resolution.
There is so much talent and money in scientific and industrial processes that would need to know.

The reason we don't have these is probably because those are either trade secrets or in rather complicated papers.
they would also be system depended, and most of them won't deal with sensors using bayer pattern interpolation (which is a mess for calculating true resolution on complex patterns)
chris

Lou Jost
Posts: 6322
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Re: Reduce magnification

Post by Lou Jost »

The minimum pitch enthusiasts have a calculation formula that dates back to the analog era, and that worked well. But is it applicable to digital? That is the question...
I think that while the formula may not be exact in the digital case, we still know for certain that adding any noisy component to a system in series will lower the quality of the system's output. There is no doubt in my mind that the "equalized" solution is wrong. The only question is "How wrong is it?"

rjlittlefield
Site Admin
Posts: 24209
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Re: Reduce magnification

Post by rjlittlefield »

chris_ma wrote:
Mon Jan 20, 2025 5:19 pm
IMAGOμM wrote:
Mon Jan 20, 2025 10:58 am
The point is that, apparently, there is not a study, a research that has developed a curve that represents the trend of the resolution transfer of a lens as a function of the sensor pitch.
Actually I'd be very surprised if there was no research which came up with pretty good modelling of how lens resolution and sensor resolution results in total system resolution.
There is so much talent and money in scientific and industrial processes that would need to know.

The reason we don't have these is probably because those are either trade secrets or in rather complicated papers.
they would also be system depended, and most of them won't deal with sensors using bayer pattern interpolation (which is a mess for calculating true resolution on complex patterns)
https://www.edmundoptics.com/knowledge- ... and-lenses provides some hint of issues, approaches, and difficulties.

Their end goal is different from Lou's, but even a quick reading of the text should amply confirm Lou's position that "matching" the sensor and lens gives a result that is not as good as either alone.
EdmundOptics wrote:Since system-level MTF is a product of the MTFs of all of the components of a system, the lens and the sensor MTFs must be multiplied together to provide a more accurate conclusion of the overall resolution capabilities of a system.
--Rik

ray_parkhurst
Posts: 3636
Joined: Sat Nov 20, 2010 10:40 am
Location: Santa Clara, CA, USA
Contact:

Re: Reduce magnification

Post by ray_parkhurst »

At the "conclusion" of this interesting discussion, did the OP's questions get answered?

Macro_Cosmos
Posts: 1609
Joined: Mon Jan 15, 2018 9:23 pm
Contact:

Re: Reduce magnification

Post by Macro_Cosmos »

The bellows shortening method to find the subject is certainly one of the many valid approaches. On a microscope, people would often call a 2x or 4x the "finder objective" because of its large field of view, and then slowing switch to higher magnification objectives to obtain the needed detail.
What you can do is 3D print a small clamp to add to the bellows' rail, using it to mark the exact, correct position where the tube lens is ideally focused at infinity. Another clamp can be added at the desired shortening point, think of these as "focus limiters". Any half-decent microscope and higher end stages have such features.
Lou Jost wrote:
Sun Jan 19, 2025 8:17 am
I'm relieved that you agree with me about the nonsense of matching pixel resolution to lens resolution.
This statement is bound to rub people the wrong way. The approach is certainly not nonsense.
There are many practical reasons to this approach. Unlike astrophotography, the main target users of microscopes are researchers on the higher end with fancy rigs and histopathologists using it as a routine. Most of them do not care about whatever out-resolving, they just want a good image to showcase or good ergonomics.

Once you consider widefield fluorescence microscopy, this approach makes total sense. You want bigger pixels for better SNR but you still need a half-decent brightfield image to accommodate it. Between the two, the compromise is often just that. We match the camera to the optical system instead of trying to out-resolve it. We typically care less about brightfield because good fluorescence signal is far more important. That is why very low resolution cameras (0.3 or 1 megapixel) are still relevant. You can have all the pixels in the universe, it is meaningless if the sensor cannot see the needed faint signals. Admittedly this is an extreme use case.

On the other end, histopathologists want clear, clean and tone-true images at the click of a button. That is it. I had to explain to someone that their inadequate H&E staining protocol resulted in the colour shift and the images from our camera (red tinted) was more faithful than Evident's expensive DP28 ($5000, lol). They did not care, CellSens automatically corrected the red hue, our camera software doesn't do that kind of deceptive garbage. Instead of simply adding a washing protocol to their slide preparation, they thought our stuff was defective. Many modern cameras already deceptively bake the raw files that cannot be turned off, this is a trend I deem despicable. However, once again, their target audience are ordinary people who wants a good image, I just hope they make these options opt-in.
You can see the clear disconnect even among the target groups.

It is quite unfortunate. Amateur microscopists is a tiny market. Personally, I would always try to go far into the empty magnification range. I have moved up to a Nikon Z8 and I am very dissatisfied with its fluorescence performance, there is so much more noise compared to my Z6. At ISO 3200, I deem it painful whereas I would shoot at 12800 without thinking on my Z6. For brightfield, I love the extra detail. I would push the objective far into empty magnification range with a magnification changer to get the most out of it, as you suggested. It is all about tradeoffs.

Instead of calling something nonsense, I think it is better to see the forest instead of one tree.

mjkzz
Posts: 1745
Joined: Wed Jul 01, 2015 3:38 pm
Location: California/Shenzhen
Contact:

Re: Reduce magnification

Post by mjkzz »

Lou Jost wrote:
Mon Jan 20, 2025 6:43 am
Let me explain what I meant by "matching lens and sensor resolutions", since mjkzz may be misunderstanding this, and we don't want to be arguing about words, we want to argue about optical phenomena.

When people (on this forum at least) talk about matching sensor resolution to lens resolution, they usually seem to mean choosing a sensor whose resolution equals the resolution of the lens.

My point is that this always causes a significant loss of resolution in the final image. To recover most of the resolution of the lens, you must use a sensor that significantly outresolves the lens.

If this were in film days, the formula I gave would be correct and would show you exactly the trade-offs involved. Economics, not optical laws, would decide how you make those tradeoffs.

I do not know how to modify the formula for digital sensors. But in general, a chain of noisy information channels will always produce a noisier result than any element of the chain. A lens + detector channel in which both components are equally noisy will produce a final image much noisier than either channel alone.

Oversampling is always good, but of course leads to diminishing returns eventually.
Argh, even just for economic reasons to make a tradeoff (matching in my mind), that is a merit, no? it makes sense, no? as for "diminishing returns", there are millions of brilliant people out there who know where to trade off along the chain of noisy channels, to dismiss them as "nonsense" is just too bold of a claim.

mjkzz
Posts: 1745
Joined: Wed Jul 01, 2015 3:38 pm
Location: California/Shenzhen
Contact:

Re: Reduce magnification

Post by mjkzz »

ray_parkhurst wrote:
Mon Jan 20, 2025 6:30 pm
At the "conclusion" of this interesting discussion, did the OP's questions get answered?
argh, get a 70 - 300 zoom lens (focused to infinity) as tube lens for various magnification range, use a micro stage with adequate range. Once all set, swap in the real tube lens.

iconoclastica
Posts: 552
Joined: Sat Jun 25, 2016 12:34 pm
Location: Wageningen, Gelderland

Re: Reduce magnification

Post by iconoclastica »

mjkzz wrote:
Tue Jan 21, 2025 2:03 am
Argh, even just for economic reasons to make a tradeoff (matching in my mind), that is a merit, no? it makes sense, no? as for "diminishing returns", there are millions of brilliant people out there who know where to trade off along the chain of noisy channels, to dismiss them as "nonsense" is just too bold of a claim.
You may be right or you may be woefully wrong, I couldn't possibly tell. But this is a call upon anonymous authority. Arguments may be challenged but statements like this can not be falsified.
--- felix filicis ---

Pau
Site Admin
Posts: 6209
Joined: Wed Jan 20, 2010 8:57 am
Location: Valencia, Spain

Re: Reduce magnification

Post by Pau »

mjkzz wrote:
Tue Jan 21, 2025 3:42 am
argh, get a 70 - 300 zoom lens (focused to infinity) as tube lens for various magnification range, use a micro stage with adequate range. Once all set, swap in the real tube lens.
Seems logical but, do you know any zoom that doesn't vignette a 70mm? My 70-200 vignettes from 140mm downwards so it would be of very limited utility for this task. Maybe a MF zoom... :-k
Pau

mjkzz
Posts: 1745
Joined: Wed Jul 01, 2015 3:38 pm
Location: California/Shenzhen
Contact:

Re: Reduce magnification

Post by mjkzz »

Pau wrote:
Tue Jan 21, 2025 5:06 am
mjkzz wrote:
Tue Jan 21, 2025 3:42 am
argh, get a 70 - 300 zoom lens (focused to infinity) as tube lens for various magnification range, use a micro stage with adequate range. Once all set, swap in the real tube lens.
Seems logical but, do you know any zoom that doesn't vignette a 70mm? My 70-200 vignettes from 140mm downwards so it would be of very limited utility for this task. Maybe a MF zoom... :-k
hahaha, I thought it would only be "searching" . . . this is how I started with a Nikon 70-300 zoom, if it vignettes, use sensor's crop mode, but eventually, when I get used to it, I only use the micro stage.

mjkzz
Posts: 1745
Joined: Wed Jul 01, 2015 3:38 pm
Location: California/Shenzhen
Contact:

Re: Reduce magnification

Post by mjkzz »

iconoclastica wrote:
Tue Jan 21, 2025 4:05 am
mjkzz wrote:
Tue Jan 21, 2025 2:03 am
Argh, even just for economic reasons to make a tradeoff (matching in my mind), that is a merit, no? it makes sense, no? as for "diminishing returns", there are millions of brilliant people out there who know where to trade off along the chain of noisy channels, to dismiss them as "nonsense" is just too bold of a claim.
You may be right or you may be woefully wrong, I couldn't possibly tell. But this is a call upon anonymous authority. Arguments may be challenged but statements like this can not be falsified.
if you could not possibly tell, what is your point? well, you figure it out.

[edit]
OK, let me help:

Whatever the formulas out there, they are also known by millions of brilliant people out there, particularly those in their own field. To dismiss their work, R/D and billions of dollars spent as nonsense, is frankly out right arrogant. In other words, do you think you are the smartest one in this field?

Prove to me if this can be falsified.
[/edit]

Post Reply Previous topicNext topic