Home

Please support mflenses.com if you need any graphic related work order it from us, click on above banner to order!
SearchSearch MemberlistMemberlist RegisterRegister ProfileProfile Log in to check your private messagesLog in to check your private messages Log inLog in

What is the sharpest 24mm lens you know of?
Post new topic   Reply to topic
Home   Manual Focus Lenses
View previous topic :: View next topic  


PostPosted: Fri Jul 01, 2022 8:58 pm    Post subject: Reply with quote
RokkorDoctor wrote:
Gerald wrote:
RokkorDoctor wrote:
I am not so sure MTF's are all that good/useful a measure for sensor performance for your everyday photographer. The discrete nature of the photo-sites (and the particular configuration of colour filters) causes all sorts of issues if you try and determine the MTF of a sensor near the lp/mm close to the spacial frequency of the photo-sites. It gets even worse in absence of an anti-aliasing filter.

In fact, since RCA's development of the picture tube-based TV shortly after World War II, MTF is routinely used by engineers working with image sensors. Both an image tube (iconoscope, orthicon, vidicon, etc.) and solid state sensors (CCD, MOS) are basically spatial samplers of the image projected by the lens. Sampling theory is solidly established and there is no fundamental difficulty in measuring the MTF of an image sensor.


I didn't suggest it isn't used or shouldn't be used for image sensors, but rather that the MTF for image sensors requires great care in its interpretation. Something engineers are aware of but the average photographer will likely not be.

It is certainly possible to measure the MTF of an image sensor, but for a sensor without an AA filter, near the Nyquist frequency (or slightly beyond) the resulting MTF values become difficult to interpret. On a sensor without an AA filter, a little beyond the Nyquist frequency the MTF reads non-zero, but what it is measuring is a heavily aliased signal, hardly a good indicator for sensor resolving power. The colour filter pattern (e.g. bayer) + associated processing makes the issue even more complicated, again especially in the absence of an AA filter.

http://www.normankoren.com/Tutorials/MTF2.html

Bottom line is that optical engineers will be aware of these issues and interpret the MTF measurement with due caution, but an average photographer will look at an MTF value of say 0.3 of an aliased sensor output, and mistakenly think the sensor can still resolve some relevant detail there.

Gerald wrote:
As an example, the figure below shows the MTF of a CCD sensor used in Olympus microscopes. By the way, 6.7 micrometer is slightly greater than the 6 micrometer pixel pitch of a 24MP full-frame sensor.


https://www.olympus-lifescience.com/en/microscope-resource/primer/digitalimaging/digitalimagingdetectors/


The Olympus article doesn't make any mention of it, but that graph to me looks like the sensor has an aggressive AA filter, and they are probably processing multiple photo-sites samples into a lower number of output pixels to ensure sufficient quality resolution (which is something they do mention in their article). Of course then the MTF becomes well-behaved, but drops to zero faster than you at face value would theoretically expect from a 6.7 micrometer pixel size sensor.


Good points. I agree with you. Like 1 small


PostPosted: Fri Jul 01, 2022 9:11 pm    Post subject: Reply with quote
Gerald wrote:
About the mtf_mapper program, it is free and relatively easy to use. I use an older version, so I believe there is a more up-to-date version today, but the version I use works well for my purposes. The most critical in using mtf_mapper is the chart with the black rectangles:




OK, that chart plays some weird tricks on my eyes! Laugh 1


PostPosted: Fri Jul 01, 2022 9:31 pm    Post subject: Reply with quote
Gerald wrote:
I used the mtf_mapper program to measure the MTF50 of all my lenses, so now I know exactly how they perform.


iangreenhalgh1 wrote:

Prove it.

Post some images, otherwise, why should anyone believe a word you say?

Especially as lately you have been caught out telling a LOT of deliberate falsehoods.

Methinks you might be a false expert and full of it....




I'm going to present some MTF50 measurements I've taken on my lenses just because I believe they are of general interest, not to "prove" something to some ill-mannered fool.

The figures below show MTF50 2D for a CZJ Flektogon 20mm F2.8 lens, for apertures F2.8, F5.6 and F11.

The MTF50 measurements are actually the combination of the MTF50's lens, 24MP sensor and anti-aliasing filter. The scales are in cycles/mm.

The F5.6 aperture provides the best lens performance in the center, but the performance in the corners is already very good. For F11, the performance in the center drops a little due to diffraction, but the performance in the corners is the best it can be.

The difference between meridional and sagittal MTF50 is due to the existence of residual laCA, which is common in practically all wide-angle lenses.













PostPosted: Sat Jul 02, 2022 8:52 am    Post subject: Reply with quote
According to DXO-Mark it's the Canon EF 24mm f/2.8 IS USM

https://www.dxomark.com/Lenses/Canon/Canon-EF-24mm-F28-IS-USM-mounted-on-Canon-EOS-5DS-R---Measurements__1009


PostPosted: Sat Jul 02, 2022 8:59 am    Post subject: Reply with quote
[quote="Gerald"]
Gerald wrote:
The figures below show MTF50 2D for a CZJ Flektogon 20mm F2.8 lens, for apertures F2.8, F5.6 and F11.

The MTF50 measurements are actually the combination of the MTF50's lens, 24MP sensor and anti-aliasing filter. The scales are in cycles/mm.

The F5.6 aperture provides the best lens performance in the center, but the performance in the corners is already very good. For F11, the performance in the center drops a little due to diffraction, but the performance in the corners is the best it can be.

The difference between meridional and sagittal MTF50 is due to the existence of residual laCA, which is common in practically all wide-angle lenses.






Interesting utility. Not perfect obviously given the residual diagonal block pattern in the results, but I'm sure Fourier theory would show that using the test chart with relatively large block squares it is inevitable that you will get some localised windowing and ringing artefacts in the output. An inevitable compromise trying to get the MTF over the entire field of view with just one single test chart.

Unrelated to the results, I'm not convinced about the program(er's) choice of heatmap scale colours. They are not in natural wavelength order, and it looks like they used linear interpolation of the primary colours in RGB space, which makes the results a bit ambiguous to interpret as very similar colours appear at different lpm values (e.g on the top-left chart the colours for 18 lpm and 32 lpm are both desaturated due to linear interpolation in RGB space and not easy too distinguish.)

IMO, having numerical values monotonously coloured in line with wavelengths (like in a rainbow, thus BCGYOR, not BYRG as in the charts shown) helps interpretation.

Interpolation of colour scales in RGB space is usually a bad idea as it results in desaturated intermediate colours, which are harder to read/distinguish. Interpolation is best done on Hue in HSL space. Essentially linear interpolation of colours in RGB space results in cutting across the colour wheel, through the centre area of desaturation, whereas linear interpolation on Hue in HSL space involves interpolating along the edge of the colour wheel , which keeps colours saturated and in natural order of wavelength).

(a bit of a rant but the "lazy" coding of linear interpolation in RGB space (Ms Excel color scales conditional formatting Rolling Eyes ) is something that bugs me; code for conversion between RGB and HSL is easily found publically available, and it makes "scientific" heatmap charts so much easier on the eye and less ambiguous to read...)


PostPosted: Sat Jul 02, 2022 9:13 am    Post subject: Reply with quote
Gerald wrote:
Ernst Dinkla wrote:
MTF becomes more sensible if it would be done on the whole process to an output format, so including RAW conversion, upsampling and sharpening routines, 4or8K output, print size and print/paper quality, MTF measurements on prints is hardly done.

Undoubtedly, if you know the MTF of each individual element in the chain of an optical-electronic system, you can calculate the global MTF of the system. However, this fact does not diminish the importance of measuring the MTF of an element simply to characterize it properly, regardless of where it will be used. This explains, for example, why lens manufacturers measure MTF of their lenses, even without knowing exactly how these lenses will be used in practice.


You would need the OTF (Optical Transfer Function) of the systems to do that. The MTF is a subset of the OTF, in which phase information is discarded. You would need the phase information contained in the OTF for Fourier theory to do its magic when daisy-chaining transfer functions Wink


PostPosted: Sat Jul 02, 2022 10:34 am    Post subject: Reply with quote
D1N0 wrote:
According to DXO-Mark it's the Canon EF 24mm f/2.8 IS USM

https://www.dxomark.com/Lenses/Canon/Canon-EF-24mm-F28-IS-USM-mounted-on-Canon-EOS-5DS-R---Measurements__1009

The Zeiss Batis 25mm F2 will be sharper but only Sony FE. Wink


PostPosted: Sat Jul 02, 2022 11:05 am    Post subject: Reply with quote
calvin83 wrote:
D1N0 wrote:
According to DXO-Mark it's the Canon EF 24mm f/2.8 IS USM

https://www.dxomark.com/Lenses/Canon/Canon-EF-24mm-F28-IS-USM-mounted-on-Canon-EOS-5DS-R---Measurements__1009

The Zeiss Batis 25mm F2 will be sharper but only Sony FE. Wink


I didn't include 25mm in my search. In comparison the zeiss has a sharper center, the Canon has a more even sharpness. They are pretty close. Other factors like CA rendering, Bokeh and price would be a more decisive factor when choosing between them.


PostPosted: Sun Jul 03, 2022 12:46 pm    Post subject: Reply with quote
Only for the not faint of heart... Twisted Evil

Gerald wrote:
Undoubtedly, if you know the MTF of each individual element in the chain of an optical-electronic system, you can calculate the global MTF of the system. However, this fact does not diminish the importance of measuring the MTF of an element simply to characterize it properly, regardless of where it will be used. This explains, for example, why lens manufacturers measure MTF of their lenses, even without knowing exactly how these lenses will be used in practice.


Ernst Dinkla wrote:
MTF becomes more sensible if it would be done on the whole process to an output format, so including RAW conversion, upsampling and sharpening routines, 4or8K output, print size and print/paper quality, MTF measurements on prints is hardly done.


RokkorDoctor wrote:

You would need the OTF (Optical Transfer Function) of the systems to do that. The MTF is a subset of the OTF, in which phase information is discarded. You would need the phase information contained in the OTF for Fourier theory to do its magic when daisy-chaining transfer functions Wink



The global response of a system formed by linear systems connected in cascade is the product of the individual responses. MTF is simply the amplitude of the complex response of an imaging system, so the global MTF is the product of the individual MTFs:




PostPosted: Sun Jul 03, 2022 1:10 pm    Post subject: Reply with quote
Gerald wrote:







RokkorDoctor wrote:
Unrelated to the results, I'm not convinced about the program(er's) choice of heatmap scale colours. They are not in natural wavelength order, and it looks like they used linear interpolation of the primary colours in RGB space, which makes the results a bit ambiguous to interpret as very similar colours appear at different lpm values (e.g on the top-left chart the colours for 18 lpm and 32 lpm are both desaturated due to linear interpolation in RGB space and not easy too distinguish.)


My main complaint about mtf_mapper scales is that they are not fixed, which sometimes makes a little difficult to compare the performance of different lenses. Even the performance comparison for the various apertures of the same lens is not so easy.

Certainly mtf-mapper isn't perfect, but considering what it does with just a digital camera, a tripod and a chart, it is a real miracle. In comparison, measuring MTF of a lens in an optical laboratory requires extensive technical experience and an investment of tens or hundreds of thousands of dollars in equipment.

My motivation for using the mtf_mapper program was to more objectively evaluate the performance of my lenses. The tests and assessments you see around are generally not very reliable. With the internet, anyone can say anything about any lens. A situation has been created in which there is a lot of "information" of very low reliability.


PostPosted: Sun Jul 03, 2022 3:22 pm    Post subject: Reply with quote
Gerald wrote:
Only for the not faint of heart... Twisted Evil


The global response of a system formed by linear systems connected in cascade is the product of the individual responses. MTF is simply the amplitude of the complex response of an imaging system, so the global MTF is the product of the individual MTFs:




I don't understand why we can discard this. Of course if we only care about the amplitude this is mathematically totally fine. But wouldn't a translation of the pattern (so a phase shift in the OTF) also lead to perceived loss of sharpness / resolution when the abberation is not symmetrical (such as is the case with coma)?

This got highly technical but it's very interesting, even tho i don't quite get it.


PostPosted: Sun Jul 03, 2022 3:45 pm    Post subject: Reply with quote
Gerald wrote:
RokkorDoctor wrote:

You would need the OTF (Optical Transfer Function) of the systems to do that. The MTF is a subset of the OTF, in which phase information is discarded. You would need the phase information contained in the OTF for Fourier theory to do its magic when daisy-chaining transfer functions Wink



The global response of a system formed by linear systems connected in cascade is the product of the individual responses. MTF is simply the amplitude of the complex response of an imaging system, so the global MTF is the product of the individual MTFs:




Ah yes, you are right. My memory was faulty here Rolling Eyes . I was thinking in terms of the product of complex transfer functions the way I used to study them way back at Uni in context of feedback systems. There the phase is crucial to be observed in the product, but here there is no feedback involved of course Wink. Assuming no aliasing happens along the way the cycles/mm of the system is maintained and only the modulation-dept info needs to be tracked (which is what the MTF does, pretty much...)


PostPosted: Sun Jul 03, 2022 3:53 pm    Post subject: Reply with quote
There is aboslutely no need to get this technical in order to be able to use a lens effectively to take a good photograph, which is what is important and we shouldn't lose sight of.


PostPosted: Sun Jul 03, 2022 3:59 pm    Post subject: Reply with quote
marius.zaech wrote:
I don't understand why we can discard this. Of course if we only care about the amplitude this is mathematically totally fine. But wouldn't a translation of the pattern (so a phase shift in the OTF) also lead to perceived loss of sharpness / resolution when the abberation is not symmetrical (such as is the case with coma)?

This got highly technical but it's very interesting, even tho i don't quite get it.


Possibly you may be right; I'm not familiar enough with the complex transfer function math in 2D image processing here (mostly used to use it in feedback systems), but would a difference in magnitude between the meridional and sagittal MTF's indicate a degree of astigmatism, and a difference in phase between the meridional and sagittal MTF's be an indicator of coma?


PostPosted: Sun Jul 03, 2022 4:06 pm    Post subject: Reply with quote
iangreenhalgh1 wrote:
There is aboslutely no need to get this technical in order to be able to use a lens effectively to take a good photograph, which is what is important and we shouldn't lose sight of.


I absolutely agree, but in this topic we are talking about sharpness (the sharpest 24mm lens) and a technical discussion about metrics of sharpness is therefore interesting to some in this context.

This is why some time ago I proposed a dedicated sub-forum here for technical/engineering aspects of photography. In absence of such a forum these type of topics unfortunately will end up being discussed where not everyone is expecting (or appreciating) it...


PostPosted: Sun Jul 03, 2022 4:11 pm    Post subject: Reply with quote
Absolutely Mark, it is a perfectly valid discussion, just pointing out that sharpness isn't everything.

Last edited by iangreenhalgh1 on Sun Jul 03, 2022 4:40 pm; edited 1 time in total


PostPosted: Sun Jul 03, 2022 4:28 pm    Post subject: Reply with quote
iangreenhalgh1 wrote:
Absolutely Mark, it is a perfectly valid discussion, just pointing out that sharpness isn't everything.

Stitching shots with a 50mm lens was mentioned and it occurred to me I actually shot an example the other day.

A single frame with a 50mm lens.

...

Four shots with the same 50mm lens stitched together.

...


That was the other topic I think ("is it possible to do landscape with 50mm FL on APS-c").

In any case, please don't take this the wrong way but this shot illustrates exactly some of the problems of doing stitching. With things like moving waves it is near impossible to avoid the blurry transition areas illustrated here, which is why I usually prefer the extreme wideangle + panoramic crop route, even if that sacrifices a bit of overall resolution...


PostPosted: Sun Jul 03, 2022 4:39 pm    Post subject: Reply with quote
Oh bugger, you're right, it was the other topic, sorry, I'll remove my post.

You're also right about the problems with stitching, it works best when nothing in the frame is moving.


PostPosted: Sun Jul 03, 2022 5:09 pm    Post subject: Reply with quote
RokkorDoctor wrote:
marius.zaech wrote:
I don't understand why we can discard this. Of course if we only care about the amplitude this is mathematically totally fine. But wouldn't a translation of the pattern (so a phase shift in the OTF) also lead to perceived loss of sharpness / resolution when the abberation is not symmetrical (such as is the case with coma)?

This got highly technical but it's very interesting, even tho i don't quite get it.


Possibly you may be right; I'm not familiar enough with the complex transfer function math in 2D image processing here (mostly used to use it in feedback systems), but would a difference in magnitude between the meridional and sagittal MTF's indicate a degree of astigmatism, and a difference in phase between the meridional and sagittal MTF's be an indicator of coma?


Honestly I'm not sure. Most of my knowledge about transfer functions comes from the context of control systems too. I think optics is a whole other story and i don't feel like reading any papers rn.

Maybe Gerald can clear things up, he seems to have good understanding of the subject matter.


PostPosted: Sun Jul 03, 2022 11:46 pm    Post subject: Reply with quote
So, the short answer is......?

Can we at least agree to a to 5? I'm looking for a different lens in this focal length as well.


PostPosted: Mon Jul 04, 2022 3:03 pm    Post subject: Reply with quote
RokkorDoctor wrote:

Possibly you may be right; I'm not familiar enough with the complex transfer function math in 2D image processing here (mostly used to use it in feedback systems), but would a difference in magnitude between the meridional and sagittal MTF's indicate a degree of astigmatism...

Yes, an imbalance between meridional and sagittal MTF is very common. The cause could be coma and/or lateral CA. In my understanding, pure astigmatism does not produce an imbalance between meridional and sagittal MTF. However, astigmatism can produce an imbalance in the MTFs if it is accompanied by a curvature of the field.





RokkorDoctor wrote:

... a difference in phase between the meridional and sagittal MTF's be an indicator of coma?

MTF by definition has no phase. MTF is the amplitude response of an imaging system.

Last edited by Gerald on Mon Jul 04, 2022 7:39 pm; edited 1 time in total


PostPosted: Mon Jul 04, 2022 3:16 pm    Post subject: Reply with quote
My prefer 24 mm lens is the Sony GM 24/1,4 USED IN MANUAL MODE.


PostPosted: Mon Jul 04, 2022 7:25 pm    Post subject: Reply with quote
Gerald wrote:
marius.zaech wrote:

Possibly you may be right; I'm not familiar enough with the complex transfer function math in 2D image processing here (mostly used to use it in feedback systems), but would a difference in magnitude between the meridional and sagittal MTF's indicate a degree of astigmatism, and a difference in phase between the meridional and sagittal MTF's be an indicator of coma?

Honestly I'm not sure. Most of my knowledge about transfer functions comes from the context of control systems too. I think optics is a whole other story and i don't feel like reading any papers rn.

Maybe Gerald can clear things up, he seems to have good understanding of the subject matter.


Yes, an imbalance between meridional and sagittal MTF is very common. The cause could be coma and/or lateral CA. In my understanding, pure astigmatism does not produce an imbalance between meridional and sagittal MTF. However, astigmatism can produce an imbalance in the MTFs if it is accompanied by a curvature of the field.


You're quoting the wrong person here Wink (the quoting system on mflenese is a bit tricky...)

Come to think of it you are correct, the astigmatism would only be obvious outside the plane of focus, which would indeed occur if there were field curvature.



Gerald wrote:
marius.zaech wrote:

a difference in phase between the meridional and sagittal MTF's be an indicator of coma?

MTF by definition has no phase. MTF is the amplitude response of an imaging system.


Same quotation issue again Wink.

I should have said OTF, I was thinking hypothetically if the MTF did have a phase constituent...


PostPosted: Mon Jul 04, 2022 7:41 pm    Post subject: Reply with quote
RokkorDoctor wrote:
You're quoting the wrong person here Wink (the quoting system on mflenese is a bit tricky...)

I already corrected my previous post. Sorry for the confusion with the quotes. Embarassed
Post new topic   Reply to topic
Home   Manual Focus Lenses
Goto page Previous  1, 2, 3
Page 3 of 3