[pschlute]

My argument is that exactly the same quantity of photons are entering both lenses, and that the APSC design actually concentrates the light more than an FF design would

If the aps-c lens "concentrated" the light more than a FF lens, then it should project a brighter image , no ?

The fact that f-stops are identical across formats and lenses show that this does not happen.

But "f-stops" are relative to the lens and body combination they happen to be on

That is wrong..... f-stops relate to the lens only, and a f2.8 FF lens provides the same illumination as a f2.8 aps-c lens. The format of the body is irrelevant to f-stops. Show me an incident light meter that requires you to enter the camera format

An analogy..... imagine a room with a floor to ceiling window that stretches the entire room width. Open the curtains half way and place a light meter directly in front the middle of the window. Now open the curtains fully and check the light meter.....will it read the same or more than before ?

---

Last edited by pschlute; 08-13-2022 at 09:19 AM.

[mikesbike]

An analogy..... imagine a room with a floor to ceiling window that stretches the entire room width. Open the curtains half way and place a light meter directly in front the middle of the window. Now open the curtains fully and check the light meter.....will it read the same or more than before ?

A good analogy!

The comparison given between performance of the FF FA 35mm f/2 and the DA 35mm f/2.8 macro Ltd provides support for better edge-to-edge results from the FF lens due to using only its central area, although this DA 35mm Ltd lens does deliver very good edge performance. It is also true that the FA 31mm Ltd has a high degree of vignetting wide open, which is worse when used on a FF camera. This is easy to see between FF and APS-C usage. Fortunately, the vignetting diminishes very rapidly when stopping down even a little bit, but needing a little less stopping down with APS-C. Good edge performance is always a difficult aspect for lens designers, and this is reflected in lab testing of lenses. So naturally if the edge of the lens is not being used in an image, there will more likely be a better result at the edges of the image frame.

But this is not always the case. One example is comparing the edge performance between the FA HD 77mm f/1.8 and the DA HD 70mm f/2.4 Limiteds. Edge performance of the DA 70mm at its wide open aperture of f/2.4 is superior to that of the FA 77mm at f/2.4, in fact the performance of the DA 70mm is quite phenomenal from wide open through f/11! That said, not to forget the FA 77mm has a significantly longer FL reach, and it is typical for the edge performance of faster lenses to need more stopping down to reach peak levels. And, the vignetting performance of the FA 77mm is actually better than that of the DA 70mm, though still very good with this lens. The FA 77mm (actual) FL difference over the DA 70mm (actually 65mm), is great enough for me to justify getting the DA HD 70mm also. I use it when I need shorter tele shots with more in the frame, and with marvelous edge-to-edge results. I use the FA HD 77mm when I need more reach, am facing lower light or faster action, want to blur the background more, and with great bokeh (although bokeh from the 70mm is very good).

Many of the FF lenses in use are either of original film-use design, or are DA digital lenses having a design carried over from film use lenses. I remember reading of differences in the way digital sensors respond to light compared to film, so lenses designed specifically for digital usage can have an advantage in avoiding some unwanted abnormalities. So this seems to be the main difference in favor of APS-C lenses over FF lenses that are available. Perhaps the most common such abnormality is purple fringing, a problem cropping up more frequently in FF lenses from the film era. This is again shown as a difference between the FA 77mm and DA 70mm Limiteds. However, I have rarely, if ever, encountered this issue when actually using my FA HD 77mm Ltd in the capacities I have indicated. It happens only when confronting certain circumstances. When this is the case, the DA HD 70mm Ltd will triumph.

[Lowell Goudge]

But "f-stops" are relative to the lens and body combination they happen to be on, and do not represent either the amount of light or the scope of the viewing angle, right?

Wrong. F stops relate to the amount of light projected per square mm irrespective of the format. Body has nothing to do with it, it really all relates to the diameter of the lens (aperture) and focal length, specifically for a simple lens the maximum aperture is focal length / diameter

The camera you are on, coverage etc is all managed separately by baffles in the lens

[Wheatfield]

But "f-stops" are relative to the lens and body combination they happen to be on, and do not represent either the amount of light or the scope of the viewing angle, right?

F-stops are a mathematical result from dividing the diameter of the aperture into the focal length of the lens.
For example a 50mm lens with an aperture diameter of 25mm is a lens set to f/2.

[brofkand]

But "f-stops" are relative to the lens and body combination they happen to be on, and do not represent either the amount of light or the scope of the viewing angle, right?

Incorrect, f/8 is f/8 is f/8.

[martin42mm]

A good analogy!

The comparison given between performance of the FF FA 35mm f/2 and the DA 35mm f/2.8 macro Ltd provides support for better edge-to-edge results from the FF lens due to using only its central area, although this DA 35mm Ltd lens does deliver very good edge performance. It is also true that the FA 31mm Ltd has a high degree of vignetting wide open, which is worse when used on a FF camera. This is easy to see between FF and APS-C usage. Fortunately, the vignetting diminishes very rapidly when stopping down even a little bit, but needing a little less stopping down with APS-C. Good edge performance is always a difficult aspect for lens designers, and this is reflected in lab testing of lenses. So naturally if the edge of the lens is not being used in an image, there will more likely be a better result at the edges of the image frame.

But this is not always the case. One example is comparing the edge performance between the FA HD 77mm f/1.8 and the DA HD 70mm f/2.4 Limiteds. Edge performance of the DA 70mm at its wide open aperture of f/2.4 is superior to that of the FA 77mm at f/2.4, in fact the performance of the DA 70mm is quite phenomenal from wide open through f/11! That said, not to forget the FA 77mm has a significantly longer FL reach, and it is typical for the edge performance of faster lenses to need more stopping down to reach peak levels. And, the vignetting performance of the FA 77mm is actually better than that of the DA 70mm, though still very good with this lens. The FA 77mm (actual) FL difference over the DA 70mm (actually 65mm), is great enough for me to justify getting the DA HD 70mm also. I use it when I need shorter tele shots with more in the frame, and with marvelous edge-to-edge results. I use the FA HD 77mm when I need more reach, am facing lower light or faster action, want to blur the background more, and with great bokeh (although bokeh from the 70mm is very good).

Many of the FF lenses in use are either of original film-use design, or are DA digital lenses having a design carried over from film use lenses. I remember reading of differences in the way digital sensors respond to light compared to film, so lenses designed specifically for digital usage can have an advantage in avoiding some unwanted abnormalities. So this seems to be the main difference in favor of APS-C lenses over FF lenses that are available. Perhaps the most common such abnormality is purple fringing, a problem cropping up more frequently in FF lenses from the film era. This is again shown as a difference between the FA 77mm and DA 70mm Limiteds. However, I have rarely, if ever, encountered this issue when actually using my FA HD 77mm Ltd in the capacities I have indicated. It happens only when confronting certain circumstances. When this is the case, the DA HD 70mm Ltd will triumph.

Glad someone brought up the difference between the design requirements for film and digital sensor lenses, which is why some full-frame lenses do not perform as well as they would on a film camera when put on a digital body, FF or APS-C. I have never quite got my head around how the designers manage to maximise the effectiveness of the micro-lenses that overlay digital sensors since light travels in a straight line, and so how can it incident each micro-lens to optimise the resolution of the sensor? Anyway, however they work, the design requirements of the digital sensor often preclude film-era lenses from working at their best and why dedicated APS-C lenses will often better them from an image quality perspective.

There is no cut-and-dried method of deciding which FF film lenses are satisfactory that I have found other than trying them, and if they work, they stay, if they don't , they go! Which sadly is why most of my M42 Takumars of various focal-lengths and eras ended up being sold to mainly film-users who could get the best out of them. Better they were used than sitting around being admired as beautiful pieces of Pentax optical engineering and gathering dust.....

[GUB]

Glad someone brought up the difference between the design requirements for film and digital sensor lenses, which is why some full-frame lenses do not perform as well as they would on a film camera when put on a digital body, FF or APS-C. I have never quite got my head around how the designers manage to maximise the effectiveness of the micro-lenses that overlay digital sensors since light travels in a straight line, and so how can it incident each micro-lens to optimise the resolution of the sensor? Anyway, however they work, the design requirements of the digital sensor often preclude film-era lenses from working at their best and why dedicated APS-C lenses will often better them from an image quality perspective.

There is no cut-and-dried method of deciding which FF film lenses are satisfactory that I have found other than trying them, and if they work, they stay, if they don't , they go! Which sadly is why most of my M42 Takumars of various focal-lengths and eras ended up being sold to mainly film-users who could get the best out of them. Better they were used than sitting around being admired as beautiful pieces of Pentax optical engineering and gathering dust.....

Could you identify the visible symptoms you are attributing to this "micro lens" issue that precludes you from using film designed lenses?
ie show us an example of a degraded image due to the "micro lenses.

[UncleVanya]

Incorrect, f/8 is f/8 is f/8.

Mostly true. In practice the f8 ratio is generally accurate but in real use the t stops can vary making one lenses f8 brighter than another. This has nothing to do with the body or the focal length or the format of the lens - just the optical design and the efficiency of light transmission. In a TTL type metering scenario it doesn't matter what your T stops are since you meter wide open and the deviation of T stop vs. F stop applies at wide open and then as you stop down the ratio is presumed to stay constant.

[martin42mm]

Could you identify the visible symptoms you are attributing to this "micro lens" issue that precludes you from using film designed lenses?
ie show us an example of a degraded image due to the "micro lenses.

Simple answer is....no. All I know is the micro lenses in front of the digital sensors mean that different optical calculations are involved compared with film lenses, which essentially focus on a flat plane. As I said in my original posting , I don't understand the maths involved in calculating the optimum design characteristics required for digital sensors with micro-lenses in front of them. Which is why I said my tried an tested method is suck it and see. If it works fine, if it doesn't, it won't.. That includes CA, which I realise can be fixed post processing, but I don't use post processing other than minor tweaks, and CA is not one of them. I reckon every correction has it's downside, and I prefer my images 'as-is;.
So to answer your question more comprehensively, it boils down to CA and lack of sharpness, as well as a certain missing pictorial quality that is present with the lens when it is used with film. The problem with describing of what works, or doesn't work, with images is that it is subjective. Leica nailed the problem when they admitted to including aberrations in the computation of their lenses back in the 60's, because they realised that some aberrations actually add to the pictorial quality of images, something we need to remember when the main question that seems to be asked of any new lens these days is 'how sharp is it'?

[martin42mm]

It is very unfortunate that your correct intuition about using FF lenses on APS-C bodies was met by many responses that are misleading at best.

There are several facts -- which will be denied by some but are nevertheless true -- concerning the use of an FF lens on an APS-C body:

1. A proper analysis requires keeping the output size constant. Regardless of whether the image was taken with an APS-C or FF camera, the output size should be kept constant (e.g., same 8x10 print, or same screen magnification).
   
   
2. The above implies a different enlargement factor. An APS-C image will require a 1.5x enlargement before it reaches the same output size as an FF image. The often used metaphor of the FF image circle just being cropped would result in a smaller output image. In practice, however, the smaller crop is blown up to have the same size as if the image had been taken with an FF camera. No one looks at APS-C images with their size reduced by a factor of 1.5
   
   
3. This above enlargement represents a stress test for the lens, as it was not designed for this use case. It's resolution was calculated for regular enlargement, not the 1.5x case. It's optical aberrations (CA, astigmatism, etc.) will be magnified by the factor 1.5 as well.
   
   
4. The effective light gathering abilities of the lens are compromised as part of the image circle is not used for imaging. Apart from the potential light scattering effects inside the mirror box (unlikely) the overly large image circle means that the FF-equivalent f-ratio of the lens is not the same as if it had projected its image circle to an APS-C-sized sensor. Many will dispute this, but this is how speed boosters work. They decrease the effective f-ratio of a lens by reducing its large image circle so that it fits on the target sensor size. This is how a metabones speed booster can turn an f/1.4 lens into an f/0.93 lens. Note that f/0.93 is expressed in terms of "APS-C money". The FF-equivalent f-ratio of the lens and speedbooster combination is still just f/1.4.
   
   
5. Because of the above, not only the FF-equivalent focal length of an FF lens changes on an APS-C body (a 50mm FF lens behaves on an APS-C body like a 75mm lens behaves on an FF body), but also the FF-equivalent f-ratio (an f/1.4 FF lens on an APS-C body behaves like a f/2.1 lens on an FF body. Note that a speed booster "undoes" the cropping effect in terms of both f-ratio and focal length. A 50mm lens will become a 35.7mm lens on a speed booster, again 35.7mm being expressed in "APS-C money". The FF-equivalent focal length of the lens and speed booster combination is still 50mm, of course.

To be clear: The physical properties of a lens, i.e., its focal length and its widest aperture, do not change when it is moved from an FF body to an APS-C body. Posters writing something like f/8 = f/8 are correct in that regard.

However, as you correctly pointed out, an f-ratio only has meaning in a context and f/8 on APS-C produces the same DOF as f/12 on FF (assuming equivalent focal length). Most people are happy to refer to an FF 50mm lens as "working like a 75mm on APS-C", meaning that on an FF-camera one would have to choose 75mm to get the same AOV, but most people are (incorrectly) unhappy to follow this through and acknowledge that the f-ratio currency changes as well.

It is correct that the exposure (a measure of relative light intensity) does not change when a lens is moved from an FF-body to an APS-C body, but the the amount of total light it sends on to the sensor (an absolute measure of light available to the sensor) does change due to the fact that quite a bit of it is lost to the image circle that is outside the APS-C sensor.

Due to the higher enlargement factor needed for APS-C (the crop factor 1.5), the same exposure, i.e., the same relative light intensity, will lead to a slightly noisier image (just like you'd expect a slightly noisier image from an f/12 exposure, compared to an f/8 exposure, everything else being the same).

The standard "sweet spot" argument is therefore not the full story, as it is not just a "sweet spot" situation but also a stress test of a lens' ability to perform under a 1.5x magnifying glass. Better lenses will not break much sweat but lesser ones may show sign of weaknesses.

In most circumstances, nothing of the above will matter. One will just open up the lens a bit more (only fails at the widest apertures), the resolution reserves built into (the better) lenses easily cope with the extra 1.5x magnification, and at normal viewing conditions lens aberrations will still be insignificant. So, yes, using FF-lenses on APS-C is almost always completely fine, but your intuitions about there being consequences to part of the image circle being lost were completely right.

BTW, comparisons between different lenses, one designed for FF, the other designed for APS-C are not helpful, since the optical designs will typically be rather different. FWIW, it is easier (cheaper) to design a lens for FF than it is for APS-C, when aiming at certain performance targets. The FF lens can afford to be a little less good due to the lower enlargement factor. This is one of the main attractions of larger formats as, in principle, one can design small format lenses that are every bit as good as large format lenses (including DOF), it is just a lot harder and therefore expensive to achieve the same image quality.

No mention of the sensor's impact on the design characteristics of the FF lens depending on whether it was intended for film or sensor use? Or in your opinion does this have zero effect , and is therefore irrelevent in this debate?

[Rondec]

No mention of the sensor's impact on the design characteristics of the FF lens depending on whether it was intended for film or sensor use? Or in your opinion does this have zero effect , and is therefore irrelevent in this debate?

We have these discussions a lot. The equivalence formulas take care of the mathematical differences between different size sensors. So, if you are shooting a 50mm f1.4 lens at iso 100 on an APS-C, you should get similar results shooting a 75mm lens at f2 and iso 200 on a full frame camera.

As for differences in lens design, they don't come into these calculations. My experience with lenses designed for film is that they have more issues with things like flare, chromatic aberration and purple fringing, but there are lenses that were designed for digital that have some issues with these things as well.

High quality lenses from the film era tend to do well now, although they may need to have some fixing of aberrations. Low quality lenses of the past continue to be low quality today and should probably be left by the wayside.

[martin42mm]

Agreed it is not clear the FF lenses referred to originally were film or digtal era lenses, but if the design of modern lenses for digital takes into account the micro-lenses overlaying the sensors does the size of the sensor dictate any difference in the optical design? As I said in my original response I find it very intriguing how the lens designs can be tailored for what seems like the near impossibel task of the light striking the micro-lenses at just the right angle! Your reference to the telecentricity of modern lens designs seems to bear this out.
Back in the early 50's there were a number of learned books written to help lens designers prior to the advent of computers, and they make fascinating reading when the conflicts of the demands of each factor required to enable a sharp image with minimal aberrations to be achieved. I've not seen any similar book relating to digital lens design, and it would be interesting to compare the criteria should one ever be published. The nearest I have read is some of the explanations given by Edmund Optical on their website, but these are very simplified descriptions compared with the mathematical formulae from those earlier books.

Regarding issues with CA and flare referred to by another poster, these can of course be reduced by careful design, and I don't see how you can exclude these factors from ultimate image quality, although I agree they have little effect on effective aperture calculations. Aperture calculations, though, if you refer to those earlier manuals ,are in fact highly complicated calculations, with some degree of disagreement as to which method is correct! Even the advent of coating, and then multi-coating, has an effect on the amount of light that reaches the film/sensor plane, and so a lens with a calculated aperture of X will in fact allow more light through with each successive coating improvement , assuming the optical design remains unchanged. The ubiquitous Helios f2 58mm is a perfect example, with later improved coating versions having a higher EFFECTIVE aperture if measured by light reaching the film plane.

And fact that CA still exists in modern digital lenses is I suspect merely a by-product of the preponderance of zoom lenses compared with prime lenses, since zoom lenses always involve more complicated compromises to achieve a satisfactory image at all focal lengths. I've yet to encounter any CA on the digital APS-C primes I use, but I may just be lucky.

[pschlute]

This above enlargement represents a stress test for the lens, as it was not designed for this use case. It's resolution was calculated for regular enlargement, not the 1.5x case. It's optical aberrations (CA, astigmatism, etc.) will be magnified by the factor 1.5 as well.

But one could use the same argument for not cropping any image. Whilst you are correct in principle, cropping an image is something we all do.... a lot in my case. Of course I will lose resolution if taken to extremes, but generally I have no issues aggressively cropping images. Lenses are quite capable of resolving detail beyond the standard frame size.

I think the term "aps-c" is also too broad in this context. There is a big difference between the 6MP sensor of my first DSLR, and the 26MP sensor of the latest K3 III. Do we assume that lenses like the DA* 300mm f4 (actually based on a film lens design) is no longer of any use because it was designed at a time when the largest MP count of a Pentax camera was 10MP ?

I have used film lenses on my K10D and my K-1. I have never experienced a situation where they are lacking because I pushed them too far. I am sure a bench test would prove me wrong, but I don't take pictures of benches

I see there being a bit of confusion in the discussion becauss there are at least three different effects, which I see as separate matters. How much light hits the objective lens, how much is transmitted through the lens to the camera's focal plane, how much of that hits the sensor, and then the practical effects of printing and such. But as I see it, the sensor's characteristics make absolutely no difference to what I was asking about - it's only the physical characteristics of the lens that matter because what I'm asking about only has to do with how much light hits the focal plane and whether an APS-C lens sort of concentrates the exact same light into a tighter circle. If the latter be true, then an APS-C lens ought to work better than a full-frame lens on an APS-C sensor. Not that you'd be able to notice when you print 4x6's.

...I have never quite got my head around how the designers manage to maximise the effectiveness of the micro-lenses that overlay digital sensors since light travels in a straight line, and so how can it incident each micro-lens to optimise the resolution of the sensor? ...

Sounds like a calculus problem to me - as the angle of incidence approaches the 90 degree limit...; sort of like the beautiful woman at the end of the long hallway? You can get close enough for all practical purposes?

...There is no cut-and-dried method of deciding which FF film lenses are satisfactory that I have found other than trying them, and if they work, they stay, if they don't , they go! Which sadly is why most of my M42 Takumars of various focal-lengths and eras ended up being sold to mainly film-users who could get the best out of them. Better they were used than sitting around being admired as beautiful pieces of Pentax optical engineering and gathering dust.....

The practical experimental approach may provide the best answer to the theoretical question.