why not f/1.2 by Sony?

Started 11 months ago | Discussions
GaryW
Veteran MemberPosts: 6,430Gear list
Like?
Re: why not f/1.2 by Sony?
In reply to tomtom50, 11 months ago

tomtom50 wrote:

GaryW wrote:

My concern is what tomtom50 posted about - past about f2, it just won't make much difference, at least for light-gathering power, on APS-C sensors. I guess that leaves DOF, or is that reduced as well?  If an f1.2 lens does no better than a f1.8 lens, I'd rather pay less for the 1.8.

Having said that, how does f1.2 work on the Nikon1 if it doesn't work on APS-C?  (Tomtom?)

-- hide signature --

Gary W.

That is the question, isn't it?

The Sony 35mm f1.4 lens works pretty well on the a900 (T1.6) and not so well on the a380 (T2.2). Yet Sony is happy to sell it as an alpha lens to all alpha owners (and NEX., since they sell adapters).

I guess it is buyer beware, except how is a typical buyer to know?

As for DOF, the same problem applies. The more angled rays that are lost come from the outside of the lens, so not only do you get less light you get deeper DOF. Oh well.

How does the Nikon f1.2 work on the 1? Who knows. Maybe as poorly as te Sony 35mm f1.4 works on the a380. Maybe better on the 10MP Nikon1 than the 14MP or vice versa. Maybe Nikon took this into account in the sensor design and it will work well.

One problem is, the manufacturers (according to articles) actually spoof the ISO, so you can't easily tell from the settings.  You actually have to analyze the noise level of the resulting photo.  (Hmm, maybe DOF could be analyzed as well.)  There are going to be those who will never think there's a problem.

But, I suspect that lens design should be able to mitigate this to some extent, if not sensor design.  Why not?  Part of the problems are adapting FF lenses to APS-C, right?

The RX-100 has a pretty wide aperture on a smaller sensor, and that does seem to help produce photos with less noise than the Nex.

Perhaps the f1.2 will work fine on the N1, but one has to wonder whether or not if it is diminishing returns before that point....

The problem is that we cannot count on the manufacturers to tell us about this. Sony seems perfectly happy to sell 35mm f1.4 lenses to a380n owners without warning them. Would Nikon be more careful?

They work as lenses.  

-- hide signature --

Gary W.

 GaryW's gear list:GaryW's gear list
Sony Alpha NEX-6 Sony E 16-50mm F3.5-5.6 PZ OSS Sony Cyber-shot DSC-V3 Sony Cyber-shot DSC-HX5 Sony Alpha DSLR-A100 +9 more
Reply   Reply with quote   Complain
EinsteinsGhost
Senior MemberPosts: 8,973Gear list
Like?
Re: why not f/1.2 by Sony?
In reply to forpetessake, 11 months ago

forpetessake wrote:

tomtom50 wrote:

forpetessake wrote:

You need to keep in mind that we are not talking about really fast apertures here. The f/1.2 on APS-C format is equivalent to just f/1.8 on FF -- hardly a superfast lens. Testing with focal reducers also seem to indicate there isn't a problem going up to f/1.0 on APS-C.

Can you cite a source? I had the opposite impression.

I'm not sure what you are asking about and what your impression is. It's a trivial math, equivalent lenses are scaled by the crop factor. The 50/1.8 lens on FF has the same properties as 34/1.2 lens on APS-C, provided all other things being equal. Focal reducers are a living demonstration of that.

Focus reducers are also concentrating light onto a smaller area, which should (theoretically at least) result in faster shutter speeds than would be achieved on FF.

The f-stop, however, is not an accurate representative of exposure (light transmission), and more so at larger apertures. Unfortunately, manufacturers get away with simply quoting f-stops. I wish t-stop would be included, as they do for cine lenses.

 EinsteinsGhost's gear list:EinsteinsGhost's gear list
Sony Cyber-shot DSC-F828 Sony SLT-A55 Sony Alpha NEX-6 Sigma 18-250mm F3.5-6.3 DC OS HSM Sony 135mm F2.8 (T4.5) STF +12 more
Reply   Reply with quote   Complain
tomtom50
Senior MemberPosts: 1,955Gear list
Like?
Re: why not f/1.2 by Sony?
In reply to EinsteinsGhost, 11 months ago

The f-stop, however, is not an accurate representative of exposure (light transmission), and more so at larger apertures. Unfortunately, manufacturers get away with simply quoting f-stops. I wish t-stop would be included, as they do for cine lenses.

Agreed. It is like the 60's when a car advertised at 250HP could only produce that air cleaner and exhaust removed, etc.

Now HP is SAE net and meas something. A similar regulation for lenses would be nice.

 tomtom50's gear list:tomtom50's gear list
Sony RX100 Canon EOS M Sony Alpha NEX-3N
Reply   Reply with quote   Complain
TiagoReil
Senior MemberPosts: 1,474Gear list
Like?
Re: why not f/1.2 by Sony?
In reply to EinsteinsGhost, 11 months ago

EinsteinsGhost wrote:

forpetessake wrote:

tomtom50 wrote:

forpetessake wrote:

You need to keep in mind that we are not talking about really fast apertures here. The f/1.2 on APS-C format is equivalent to just f/1.8 on FF -- hardly a superfast lens. Testing with focal reducers also seem to indicate there isn't a problem going up to f/1.0 on APS-C.

Can you cite a source? I had the opposite impression.

I'm not sure what you are asking about and what your impression is. It's a trivial math, equivalent lenses are scaled by the crop factor. The 50/1.8 lens on FF has the same properties as 34/1.2 lens on APS-C, provided all other things being equal. Focal reducers are a living demonstration of that.

Focus reducers are also concentrating light onto a smaller area, which should (theoretically at least) result in faster shutter speeds than would be achieved on FF.

The f-stop, however, is not an accurate representative of exposure (light transmission), and more so at larger apertures. Unfortunately, manufacturers get away with simply quoting f-stops. I wish t-stop would be included, as they do for cine lenses.

There was a resaon why in photography f number was more important than T number, and that for cine lenses T number was more important.

Actually, in theory, the t number and f number should be equal with a lens and coating that looses nothing of light. All lenses loose a bit, some loose more,cause of elements, goups (when 2 elements are sticked) and with coatings. IT is a balancing act. You add more elements, for instance a zoom, gives you more flexibility, you add coating, better contrast and less flare, but you loose a bit of light. Generally is not much, and actually primes loose in general less light than zooms, cause zooms generally have more elements.

In photography, before Autofocus, or even manual focus through the lens, people focused with distance and area in focus. The area in focus is given by the f number, not by the t number. On the other hand, lenses where simple, so the f number was close to the T number (from loose from elements and coating). So photographers learned to use f numbers and those where close to the T number to be acceptable. The magnitude that film got you was enough to make a bit of a correction when printing (add a bit of light) and thats it.

Cine lenses needed to be precise on the T number cause as you change them, you need to be exactly on the same exposure for a different reel. IF not, on screen, one after the other, a very small change was perceived and ruined the film. Thats why cine lenses are categorized that way. Cause for Movies, it is more important to have the exposure perfect than to have the focus perfect.

Now, with digital sensors, and the phenomenon we are describing here, the difference from f to t number is bigger because of limitations on digital sensors. Only on fast apertures.

All this that I mention about lenses is well known for a long time, and people that work in this field knows it pretty well. Only people new to this are surprise to realize that actually f is not a value for lighg, but as an aprozimation, it used to work pretty well. (like one would say certain scientific theories, though not exact, are good cause for day to day calculations give results good enough, ex, newtons formula of gravity). The problem is that digital sensors ruined this for fast lenses. Different manufacturers do different things. Some cook the files. Jpgs and even raws, others limit the lenses, others are more open to this. But this is a very well known problem.

Also, a lens is considered fast cause you can get a very fast shutter speed with it, not because the DoF is shallow. Those people that say that a fast lens is not fast on an aps-c camera have it wrong. It is still a fast lens (ignoring the sensor problem) it has less shallow depth of field than the same lens on a FF camera, but it is as fast in terms of light.

Finally, can we kill this!!!! and stop miss informing? The reason sony doesn't offer a faster than 1.8 lens for crop sensor is that it is not worth it, that the cost would be too big for a small return in light, and they consider their customers are mostly consumers, amateurs, advance amateurs (at least for crop sensor, FF is a bit different, but still there is a limit there, 1.4, they think the same there, not worth it to get that extra f1.2). Accept it , and keep on taking pictures.

 TiagoReil's gear list:TiagoReil's gear list
Sony Alpha NEX-5 Sony Alpha NEX-6 Sony E 18-55mm F3.5-5.6 OSS Sony 85mm F2.8 SAM Sigma 70mm F2.8 EX DG Macro +3 more
Reply   Reply with quote   Complain
DtEW
Senior MemberPosts: 1,624Gear list
Like?
Re: why not f/1.2 by Sony?
In reply to TiagoReil, 11 months ago

Actual content to improve the forum SNR.  Thank you.

 DtEW's gear list:DtEW's gear list
Canon EOS 6D Canon EF 100mm f/2.0 USM Canon EF 75-300mm f/4.0-5.6 IS USM Canon EF 24-105mm f/4L IS USM Samyang 14mm F2.8 IF ED MC Aspherical +14 more
Reply   Reply with quote   Complain
franzel
Contributing MemberPosts: 752
Like?
Re: why not f/1.2 by Sony?
In reply to TiagoReil, 11 months ago

Great Bustard wrote:

They are related in that, for equally efficient sensors, the same total amount of light falling on the sensor results in the same image noise.

The reason larger sensor systems are less noisy than smaller sensor systems is because more light falls on the sensor for a given exposure.

Negative - it's pixel density; or enlargement factor , if the sensors only differ in size but havethe same resolution.
In real life, it is more complex, but not by much .

There is no such thing as measuring any 'total' amount of light, based on the area it covers .
The "amount of light" - which isn't exactly proper terminology - is the exposure any object of a predetermined brightness creates on any given part of a sensor, no matter its size .
It's physically impossible to collect more light from a lit object or scene (aka light source) by increasing the area it is projected on .

TiagoReil wrote:

Also, a lens is considered fast cause you can get a very fast shutter speed with it, not because the DoF is shallow. Those people that say that a fast lens is not fast on an aps-c camera have it wrong. It is still a fast lens (ignoring the sensor problem) it has less shallow depth of field than the same lens on a FF camera, but it is as fast in terms of light.

Amen .

Reply   Reply with quote   Complain
GaryW
Veteran MemberPosts: 6,430Gear list
Like?
Re: why not f/1.2 by Sony?
In reply to franzel, 11 months ago

franzel wrote:

Great Bustard wrote:

They are related in that, for equally efficient sensors, the same total amount of light falling on the sensor results in the same image noise.

The reason larger sensor systems are less noisy than smaller sensor systems is because more light falls on the sensor for a given exposure.

Negative - it's pixel density; or enlargement factor , if the sensors only differ in size but havethe same resolution.
In real life, it is more complex, but not by much .

There is no such thing as measuring any 'total' amount of light, based on the area it covers .
The "amount of light" - which isn't exactly proper terminology - is the exposure any object of a predetermined brightness creates on any given part of a sensor, no matter its size .
It's physically impossible to collect more light from a lit object or scene (aka light source) by increasing the area it is projected on .

You don't collect more light per unit area, but since the area is larger, then more photons were collected (assuming the same shutter speed, obviously).  So, you can think of it in those terms if you really want to (collecting more light), but that's not how the f ratio works -- it works as you describe, per unit area.

TiagoReil wrote:

Also, a lens is considered fast cause you can get a very fast shutter speed with it, not because the DoF is shallow. Those people that say that a fast lens is not fast on an aps-c camera have it wrong. It is still a fast lens (ignoring the sensor problem) it has less shallow depth of field than the same lens on a FF camera, but it is as fast in terms of light.

Amen .

-- hide signature --

Gary W.

 GaryW's gear list:GaryW's gear list
Sony Alpha NEX-6 Sony E 16-50mm F3.5-5.6 PZ OSS Sony Cyber-shot DSC-V3 Sony Cyber-shot DSC-HX5 Sony Alpha DSLR-A100 +9 more
Reply   Reply with quote   Complain
TiagoReil
Senior MemberPosts: 1,474Gear list
Like?
Re: why not f/1.2 by Sony?
In reply to GaryW, 11 months ago

GaryW wrote:

franzel wrote:

Great Bustard wrote:

They are related in that, for equally efficient sensors, the same total amount of light falling on the sensor results in the same image noise.

The reason larger sensor systems are less noisy than smaller sensor systems is because more light falls on the sensor for a given exposure.

Negative - it's pixel density; or enlargement factor , if the sensors only differ in size but havethe same resolution.
In real life, it is more complex, but not by much .

There is no such thing as measuring any 'total' amount of light, based on the area it covers .
The "amount of light" - which isn't exactly proper terminology - is the exposure any object of a predetermined brightness creates on any given part of a sensor, no matter its size .
It's physically impossible to collect more light from a lit object or scene (aka light source) by increasing the area it is projected on .

You don't collect more light per unit area, but since the area is larger, then more photons were collected (assuming the same shutter speed, obviously).  So, you can think of it in those terms if you really want to (collecting more light), but that's not how the f ratio works -- it works as you describe, per unit area.

TiagoReil wrote:

Also, a lens is considered fast cause you can get a very fast shutter speed with it, not because the DoF is shallow. Those people that say that a fast lens is not fast on an aps-c camera have it wrong. It is still a fast lens (ignoring the sensor problem) it has less shallow depth of field than the same lens on a FF camera, but it is as fast in terms of light.

Amen .

-- hide signature --

Gary W.

Exactly. Some people think that you will actually get more speed with a FF lens on FF than on a crop sensor, and it will give you the same (more or less) shutter speed. So, if you receive more light, how is it that you have the same shutter speed value?

IF you receive more light, why a photometer doesnt have sensor size as a parameter? why you use the same photometer for a P&S, for a crop sensor, for a FF or for a medium format camera?

Of course when you compare cameras there are more difficult things, for instance, different manufacturers grade ISO numbers differently, or even different technologies from same manufacturer. There are many factors, but when you compare with an Ideal sensor, that you could consider the same, a lens is not more bright in a FF camera than in an aps-c camera. IT is the same actually. You dont get a faster shutter speed. You get the same (approximately, again, depending on other things)

If not, lets through all our photometers and lets get one (imaginary) that besides ISO, you have to enter the sensor size. Where do I get one?

 TiagoReil's gear list:TiagoReil's gear list
Sony Alpha NEX-5 Sony Alpha NEX-6 Sony E 18-55mm F3.5-5.6 OSS Sony 85mm F2.8 SAM Sigma 70mm F2.8 EX DG Macro +3 more
Reply   Reply with quote   Complain
ilza
Forum MemberPosts: 55
Like?
Re: why not f/1.2 by Sony?
In reply to tomtom50, 11 months ago

tomtom50 wrote:

The Sony 35mm f1.4 lens works pretty well on the a900 (T1.6) and not so well on the a380 (T2.2).

!!!

As for DOF, the same problem applies. The more angled rays that are lost come from the outside of the lens, so not only do you get less light you get deeper DOF. Oh well.

!!!

Sony seems perfectly happy to sell 35mm f1.4 lenses to a380n owners without warning them.

I especially enjoy this one!

Focus reducers are also concentrating light onto a smaller area, which should (theoretically at least) result in faster shutter speeds than would be achieved on FF.

Well, this one at least looks like an honest misunderstanding, not a blatant misunderstanding…

I’m afraid you need to get your knowledge of basic physics of photography up a little bit.  What I would agree with is that the whole subject of comparing different form-factors it too obfuscated; too many people talk about things they do not have any clue about.

It helps if one thinks in correct terms:

  • What part of the subject plane is captured;
  • number of captured photons per an image of some area on the subject;
  • Diffraction circle mapped back to the subject plane;
  • Circle of confusion mapped back to the subject plane(s);
  • (Averaged) readout noise of photocells per an image of some area on the subject, in electrons [*];
  • Full well capacity of photocells per an image of some area on the subject, in electrons;
  • QE: fraction of photons converted to electrons.

[*] Well, another useful measure of noise is electrons/√Hz; but this affects only the technological part: how many ADC do you need.

When you take all this into account, it turns out that everything but the full-well is absolutely trivial: there are only two relevant pieces of data: angle of view, and entry pupil (forgetting for a moment about the lens transparency!).  This means that if you compare 8×10in film with a ⅔" sensor, if your lenses have similar optical quality [**], and the same angle of view and entry pupil, you can get identical images if your sensors are of similar quality [***] and the full wells do not matter.

[**] One should keep in mind that it is progressively harder and harder to make a lens with given optical quality, given angle of view, and given entry pupil when the focal length goes down.  And: at some moment the Abbe's sine law will strike back, and this will not be physically possible.

[***] My back-of-envelop calculations show that it is not possible to match 8×10in film by a ⅔" sensor (about ¹⁄₂₇ difference in linear size).  With QE=1, digital would match the best film at about ¹⁄₁₆ difference of formfactor.  And AFAIK, current QE of digital cameras (considered as systems) is about 0.12 nowadays. So today it is ⅙ difference in formfactor: digital-APC ≈ MF-film. [And what am I talking about here?  Essentially, it was the discussion of the effective-QE of film; in my estimates, it is about ¹⁄₂₅₆. But this topic is completely orthogonal to the discussion of digital-to-digital comparison.]

And since full-well is determined by technology, not by physics, it is natural to expect that this should not matter — at least when one compares bodies of different generations.

Reply   Reply with quote   Complain
tomtom50
Senior MemberPosts: 1,955Gear list
Like?
Re: why not f/1.2 by Sony?
In reply to ilza, 11 months ago

ilza wrote:

tomtom50 wrote:

The Sony 35mm f1.4 lens works pretty well on the a900 (T1.6) and not so well on the a380 (T2.2).

!!!

As for DOF, the same problem applies. The more angled rays that are lost come from the outside of the lens, so not only do you get less light you get deeper DOF. Oh well.

!!!

Sony seems perfectly happy to sell 35mm f1.4 lenses to a380n owners without warning them.

Have you read F stop blues at Dxomark?

http://www.dxomark.com/index.php/Publications/DxOMark-Insights/F-stop-blues

Very worthwhile when considering the usefulness of fast lenses on a digital camera.

 tomtom50's gear list:tomtom50's gear list
Sony RX100 Canon EOS M Sony Alpha NEX-3N
Reply   Reply with quote   Complain
tomtom50
Senior MemberPosts: 1,955Gear list
Like?
Re: why not f/1.2 by Sony?
In reply to ilza, 11 months ago

ilza wrote:

I’m afraid you need to get your knowledge of basic physics of photography up a little bit.  What I would agree with is that the whole subject of comparing different form-factors it too obfuscated; too many people talk about things they do not have any clue about.

It helps if one thinks in correct terms:

Yes. It is important to realize when matching a lens to a digital sensor that the optics do not end at the lens.

Unlike film digital sensors have their own optical systems, including lenses. The photons need to follow a path through the sensor, past wires (DSLR sensors are not BSI)  before they can be counted at the photodiode, and that path is sensitive to angle of incidence, more on some sensors that on others. Quantum efficiency is not a single number; it is dependent of the angle the light strikes the sensor/microlens system.

Dxomark rates T stops for tested lenses, and their transmission testing includes the sensor. That is why Dxomark reports different full-open T stops for the same lens.

Look up the Sony 36mm f1.4 and check the empirically obtained T stop for various cameras. You will find significant variation.

 tomtom50's gear list:tomtom50's gear list
Sony RX100 Canon EOS M Sony Alpha NEX-3N
Reply   Reply with quote   Complain
ilza
Forum MemberPosts: 55
Like?
Re: why not f/1.2 by Sony?
In reply to tomtom50, 11 months ago
Have you read F stop blues at Dxomark?

http://www.dxomark.com/index.php/Publications/DxOMark-Insights/F-stop-blues

Very worthwhile when considering the usefulness of fast lenses on a digital camera.

It is interesting, but their choice of using T-word where F-word is due is abysmal.  When T-number is mentioned, one assumes that the light loss is discussed (as opposed to the angle of incoming cone of light hitting a point on a sensor). T-number affects the exposure, F-number affects also the diffraction and the circle of confusion due to misfocus.

While what they measure (I assume it is the photon noise) may be, technically speaking, the T-number, the observed difference between the cameras cannot be attributed to difference in light transmission in the lens.  My immediate conjecture is that what they observe is due to vignetting in the microlenses; then this would affect the diffraction circle and the circle of confusion in a similar way to a mechanical limiter of the aperture ring.

(Well, the difference with mechanical limiter is that the bokeh must be significantly improved with such vignetting… :-))

Reply   Reply with quote   Complain
ilza
Forum MemberPosts: 55
Like?
Re: why not f/1.2 by Sony?
In reply to tomtom50, 11 months ago

The photons need to follow a path through the sensor, past wires (DSLR sensors are not BSI)  before they can be counted at the photodiode, and that path is sensitive to angle of incidence, more on some sensors that on others. Quantum efficiency is not a single number; it is dependent of the angle the light strikes the sensor/microlens system.

Shadows of the wires will be longer with oblique illumination indeed.  But since wires (as visible from the sensels) form a rectangular skyscraper landscape, would not this rectangularity be visible on the shape of bokeh?  (When everything settles down, bokeh is, AFAIU, only “the shadow of the entry pupil” on the sensor.  So everything which effectively affects the entry pupil should be visible on bokeh)

Until this relation with bokeh is clarified, I like my microlenses conjecture better.

Reply   Reply with quote   Complain
GaryW
Veteran MemberPosts: 6,430Gear list
Like?
Re: why not f/1.2 by Sony?
In reply to ilza, 11 months ago

ilza wrote:

The photons need to follow a path through the sensor, past wires (DSLR sensors are not BSI)  before they can be counted at the photodiode, and that path is sensitive to angle of incidence, more on some sensors that on others. Quantum efficiency is not a single number; it is dependent of the angle the light strikes the sensor/microlens system.

Shadows of the wires will be longer with oblique illumination indeed.  But since wires (as visible from the sensels) form a rectangular skyscraper landscape, would not this rectangularity be visible on the shape of bokeh?  (When everything settles down, bokeh is, AFAIU, only “the shadow of the entry pupil” on the sensor.  So everything which effectively affects the entry pupil should be visible on bokeh)

Until this relation with bokeh is clarified, I like my microlenses conjecture better.

I think Tomtom simply meant the circuitry (wires) in the sensor itself would block light if the angle of the light is too oblique.  This has nothing to do with the subject matter.  As for bokeh, as was mentioned before, if this effect of blocking steep angles is consistent, then it will affect bokeh.  Microlenses could help mitigate the effect if they were offset.  This might make things worse for long tele lenses, I would think.

Maybe if they make larger sensors with backside illumination (as mentioned earlier), it will help with wider apertures.  I think they'll do that eventually, but there's probably not a strong need for it, particularly with consumers not understanding the need -- or rather, few consumers that even would see benefit.  (Probably about .01% of Nex users.   )  Maybe Fuji fans can convince Fuji to go BSI.

-- hide signature --

Gary W.

 GaryW's gear list:GaryW's gear list
Sony Alpha NEX-6 Sony E 16-50mm F3.5-5.6 PZ OSS Sony Cyber-shot DSC-V3 Sony Cyber-shot DSC-HX5 Sony Alpha DSLR-A100 +9 more
Reply   Reply with quote   Complain
tomtom50
Senior MemberPosts: 1,955Gear list
Like?
Re: why not f/1.2 by Sony?
In reply to ilza, 11 months ago

ilza wrote:

Have you read F stop blues at Dxomark?

http://www.dxomark.com/index.php/Publications/DxOMark-Insights/F-stop-blues

Very worthwhile when considering the usefulness of fast lenses on a digital camera.

It is interesting, but their choice of using T-word where F-word is due is abysmal.

T is transmission, f is a ratio. They are measuring transmission. Since their measurement includes transmission loss in the sensor optics they could have chosen a new term, but between T and f they chose the better.

When T-number is mentioned, one assumes that the light loss is discussed (as opposed to the angle of incoming cone of light hitting a point on a sensor). T-number affects the exposure, F-number affects also the diffraction and the circle of confusion due to misfocus.

The light is being lost, at the sensor rather than the lens. T is traditionally used because it is more accurate for exposure. The Dxomark use carries on that tradition adding a new source of transmission loss that did not exist at the time T was developed.

While what they measure (I assume it is the photon noise) may be, technically speaking, the T-number, the observed difference between the cameras cannot be attributed to difference in light transmission in the lens.  My immediate conjecture is that what they observe is due to vignetting in the microlenses; then this would affect the diffraction circle and the circle of confusion in a similar way to a mechanical limiter of the aperture ring.

Dxomark Uses the sesnsor as the measurement device. They have found some cameras hide transmission loss at wide apertures by boosting ISO surreptitiously. I think they do use noise levels to detecting the ISO spoofing; I don't recall. We can observe the phenomenon by mounting legacy lenses with an adapter. Then the camera does not know the nominal aperture and cannot spoof ISO and exposure can be assessed.

(Well, the difference with mechanical limiter is that the bokeh must be significantly improved with such vignetting… :-))

 tomtom50's gear list:tomtom50's gear list
Sony RX100 Canon EOS M Sony Alpha NEX-3N
Reply   Reply with quote   Complain
RedFox88
Forum ProPosts: 21,232
Like?
Re: bottom line
In reply to forpetessake, 11 months ago

forpetessake wrote:

 That makes sense, especially since the future sensors will be FF.

I could not agree less! The future of image format for photography is smaller, not larger.  Photography format has been shrinking since its inception.  We've gone from large format to medium format to small format (35mm) to digital aps-c.  Where now 35mm is being thought of as the "bigger and better" alternative to aps-c when 35mm is small format!

Aps-c is quickly becoming the "bigger and better" alternative to P&S cameras even though most that take pictures are now using the extremely popular iphone camera.

Reply   Reply with quote   Complain
forpetessake
Senior MemberPosts: 2,798
Like?
Re: why not f/1.2 by Sony?
In reply to ilza, 11 months ago

ilza wrote:

The photons need to follow a path through the sensor, past wires (DSLR sensors are not BSI)  before they can be counted at the photodiode, and that path is sensitive to angle of incidence, more on some sensors that on others. Quantum efficiency is not a single number; it is dependent of the angle the light strikes the sensor/microlens system.

Shadows of the wires will be longer with oblique illumination indeed.  But since wires (as visible from the sensels) form a rectangular skyscraper landscape, would not this rectangularity be visible on the shape of bokeh?  (When everything settles down, bokeh is, AFAIU, only “the shadow of the entry pupil” on the sensor.  So everything which effectively affects the entry pupil should be visible on bokeh)

Until this relation with bokeh is clarified, I like my microlenses conjecture better.

Yep, it's been explained elsewhere that the effect on t-stop is due to heavy vignetting (compensated in software). Vignetting is both natural effect of the large aperture lens as well as microlens design. It's not pronounced in the old sensors without microlenses, nor in the new designs with shifted microlenses.

Since we are discussing Sony NEX design here, there are test pictures that show that there isn't a heavy vignetting even at f/0.85, much less at f/1.2:

Reply   Reply with quote   Complain
Bart Hickman
Veteran MemberPosts: 7,098Gear list
Like?
Re: bottom line
In reply to forpetessake, 11 months ago

forpetessake wrote:

stan_pustylnik wrote:

Why doesn't Sony use smaller sensor surface advantage from NEX system to build f/1.2 35mm lens,  and 85mm f/1.4?

There aren't technical difficulties making those lenses, they already exist in FF equivalent, except the E-mount bodies, they are too small, weak, and uncomfortable for large lenses. The 85mm/1.4 would be heavy (likely about 1.5lb) for small NEX cameras. Rumors say that Sony is heading with larger mirrorless designs using A-mount. That makes sense, especially since the future sensors will be FF.

A 35mm/1.2 lens would be no problem on any NEX body.  I personally think an 85/1.4 lens would also be fine on the NEX6 or 7.  Much longer than that and it might be beneficial for Sony to make a slightly larger NEX body for the high end.  But this would be for ergonomics, not strength.  The E-mount itself is as mechanically strong as any on the market.

Bart

-- hide signature --
 Bart Hickman's gear list:Bart Hickman's gear list
Sony Alpha NEX-6 Pentax smc FA 50mm F1.4 Sony E 55-210mm F4.5-6.3 OSS Sony E 10-18mm F4 OSS Sony E 16-50mm F3.5-5.6 PZ OSS +9 more
Reply   Reply with quote   Complain
tomtom50
Senior MemberPosts: 1,955Gear list
Like?
Re: why not f/1.2 by Sony?
In reply to forpetessake, 11 months ago

forpetessake wrote:

ilza wrote:

The photons need to follow a path through the sensor, past wires (DSLR sensors are not BSI)  before they can be counted at the photodiode, and that path is sensitive to angle of incidence, more on some sensors that on others. Quantum efficiency is not a single number; it is dependent of the angle the light strikes the sensor/microlens system.

Shadows of the wires will be longer with oblique illumination indeed.  But since wires (as visible from the sensels) form a rectangular skyscraper landscape, would not this rectangularity be visible on the shape of bokeh?  (When everything settles down, bokeh is, AFAIU, only “the shadow of the entry pupil” on the sensor.  So everything which effectively affects the entry pupil should be visible on bokeh)

Until this relation with bokeh is clarified, I like my microlenses conjecture better.

Yep, it's been explained elsewhere that the effect on t-stop is due to heavy vignetting (compensated in software). Vignetting is both natural effect of the large aperture lens as well as microlens design. It's not pronounced in the old sensors without microlenses, nor in the new designs with shifted microlenses.

Since we are discussing Sony NEX design here, there are test pictures that show that there isn't a heavy vignetting even at f/0.85, much less at f/1.2:

Your picture does not quite answer the question. Loss of oblique rays on the way to the photodiode is not limited to the corners of the frame.

A test can be made pretty simply:

- Mount the 50mm f1.2 with a dumb adapter so lens information is not transferred to the camera (ISO spoofing defeated)

- Take 'equivalent exposures at 1/3 stop from f2.8 to full open at the same st ISO (for example f2.8@1/30, f2.5@1/40, f2.2@1/50, f2@1/60 all the way to wide open.

- The exposure in the center will drop as you get wider.

- Take a few underexposed shots at f2.8 (1/40, 1/50, 1/60). These can be used to benchmark just how ineffective the wide aperture was.

- Once determining that you f1.4 acts more like an f1.8 (or whatever), see if f1.4 actually gives much more bokeh than f1.8.

I will do this myself when I get my Nikon adapter, but i have the 16MP sensor which is less problematic than the sensor in the NEX 7

 tomtom50's gear list:tomtom50's gear list
Sony RX100 Canon EOS M Sony Alpha NEX-3N
Reply   Reply with quote   Complain
ilza
Forum MemberPosts: 55
Like?
Re: why not f/1.2 by Sony?
In reply to ilza, 11 months ago

I wrote:

Shadows of the wires will be longer with oblique illumination indeed.  But since wires (as visible from the sensels) form a rectangular skyscraper landscape, would not this rectangularity be visible on the shape of bokeh?  (When everything settles down, bokeh is, AFAIU, only “the shadow of the entry pupil” on the sensor.  So everything which effectively affects the entry pupil should be visible on bokeh)

Until this relation with bokeh is clarified, I like my microlenses conjecture better.

Thinking more about this, I get completely lost!

Different wavelenghts penetrate on different depth (in doped silicon; see, e.g., the original Foveon's X3 papers on the graphs; IIRC, it is 0.1µm–3µm).  So the shadows (no matter of  what: of wires, or of edges of microlenses) are going to have different areas depending on the depth, so depending on the wavelength.

So the shape of the bokeh should be significantly different for R, G and B.  Is it observed?  Likewise for ISO boost at wide apertures…

Reply   Reply with quote   Complain
Keyboard shortcuts:
FForum MMy threads