[Class A]

DoF on FF is a blessing and a curse.

It is just a blessing, as you can simply stop down, if you prefer a deeper DOF.

You don't lose out by stopping down. Shooting a 50mm lens at f/8 on FF provides the same amount of light as a 33mm lens at f/5.8 on APS-C.
In short, don't worry about losing light. You don't. The only real difference between FF and APS-C is that there are some FF lenses (e.g., a 50mm f/1.4) which have no equivalent on APS-C. Apart from those "holes" in the APS-C lens line up, you can always emulate one camera with another by just choosing the appropriate lens and settings. Ergo, FF gives you more opportunities (with the right lenses), but you don't have to pay a price (only in the sense that you have to "pay" for more light than you'd be able to get on APS-C with a shallower DOF).

FF + f2.8 zoom is heavy;

Yes, but it also provides more options than an f/2.8 zoom on APS-C.
The equivalent zoom on FF is an f/4 zoom, so for instance the 60-250/4 lens would be a good FF lens for you, if you don't want to have more options than you have with an f/2.8 zoom on APS-C.

[Class A]

ISO invariance is not an advantage, it's a disadvantage that most digital camera have.

It most certainly is not a disadvantage. There are no downsides to ISO invariance whatsoever.

It is generally considered to be an advantage as shooting with an ISO setting that is nominally too low, allows one to guard against blown out highlights without having to pay the price of increased noise due to underexposure.

A sensor that is not ISO invariant requires one to always shoot at the edge of ISO settings, in order to avoid any avoidable noise. This carries the risk of blowing out highlights. An ISO invariant sensor is more flexible in that regard.

[biz-engineer]

It most certainly is not a disadvantage. There are no downsides to ISO invariance whatsoever.

ISO invariant sensor means there is not signal conditioning between the pixel and the A/D conversion, meaning when ISO increases, exposure decreases and the signal gets smaller and smaller relative to ADC full scale. That's a disadvantage because as the exposure signal gets smaller, the read noise (quantization noise+digital noise from substrate) remains constant, which significantly lower S/N and also decrease linearity and resolution for coding tones. Ideally, gain would increase 6dB for every stop of increased ISO in order to have the pixel output level match to ADC full scale, but while AGC (Automatic Gain Control) is using in radio receivers, it's not implemented in CMOS imagers in order to keep the designs simple. Some have implemented dual gain, which shows better IQ at higher ISO.

[Class A]

First of all, I understand "ISO invariant" or "ISO less" as a property of a sensor, not a description of a particular implementation.

AFAIC, it does not matter whether there is amplification in the analogue domain or not, the defining principle of an "ISO less" sensor is that it doesn't matter at which ISO setting one captures the image, as amplification in the digital domain after the fact is just as good.

I don't know any sensor that actually achieves any significant advantage by amplifying in the analogue domain before A/D conversion compared to a sensor with extremely low read noise, but I'm happy to learn.

As I understand, analogue amplification was used when read noise levels still suggested that analogue amplification should be used before A/D conversion in order to maintain an improved SNR.

Since the advent of extremely low read noise sensors, there is no need anymore to artificially boost the signal before conversion with all the problems analogue amplification entails.

meaning when ISO increases, exposure decreases and the signal gets smaller and smaller relative to ADC full scale.

The actual signal also gets "smaller and smaller" when analogue amplification ("signal conditioning" as you call it) is used. Amplifying a signal does not increase the information content in a signal. On the contrary, the amplification process itself will deteriorate the signal by introducing noise, potential non-linearities, etc. All these problems were a good trade-off when sensors still had very high read noise levels (and thus struggled with very low signals), but with modern sensors that is no longer the case.

...while AGC (Automatic Gain Control) is using in radio receivers, it's not implemented in CMOS imagers in order to keep the designs simple.

Do you maintain that is also true for modern medium format sensors, such as the one used in the Fuji GFX 50S?

My understanding is that the low read noise of modern CMOS sensors implies that it is better to amplify in the digital domain rather than in the analogue domain. In other words, "ISO less" sensors are not a result of cost-saving as you present it, but a result of advances in A/D technology that ultimately not only give photographers cleaner images to start with but also the ability to avoid overexposure without having to pay a (increased noise) price.

[peterh337]

"It is generally considered to be an advantage as shooting with an ISO setting that is nominally too low, allows one to guard against blown out highlights without having to pay the price of increased noise due to underexposure."

I agree. The K1's low noise means that one can use the TAV mode for most shots, which is how it should be. After all, photographers choose the focal length, aperture and (to a much lesser extent, unless shooting action) shutter speed for the desired effect, and worrying about the ISO is just a hassle which has no advantage whatever. With a noiseless sensor and an infinite-resolution A-D converter, ISO would not exist.

[victormeldrew]

one can use the TAV mode for most shots, which is how it should be.

Mmm...what?

ISO is just a hassle which has no advantage whatever.

Personally, I like using noise creatively. It's much more realistic and random than fake film grain in post, and much less hassle than using film. I also like having control over sensitivity, rather than letting a dumb processor decide for me. Not everybody wants an arbitrary 'correct' exposure decided for them by a Japanese engineer's algo.

[peterh337]

Well, yes, that's true, you want exposure control, but that's not the same thing as ISO which is basically irrelevant to photography (unless you like to create noise).

TAV is the mode where you set the shutter speed, set the F number, and the camera sets the exposure (by tweaking what we call the ISO).

[keos]

It is just a blessing, as you can simply stop down, if you prefer a deeper DOF.

You don't lose out by stopping down. Shooting a 50mm lens at f/8 on FF provides the same amount of light as a 33mm lens at f/5.8 on APS-C.
In short, don't worry about losing light. You don't. The only real difference between FF and APS-C is that there are some FF lenses (e.g., a 50mm f/1.4) which have no equivalent on APS-C. Apart from those "holes" in the APS-C lens line up, you can always emulate one camera with another by just choosing the appropriate lens and settings. Ergo, FF gives you more opportunities (with the right lenses), but you don't have to pay a price (only in the sense that you have to "pay" for more light than you'd be able to get on APS-C with a shallower DOF). .

I don't think that's quite right.
F-stop is f-stop, regardless of the lens and sensor, given the same optical transparency of the glass and the absolute sensitivity of the light sensor. Use this thought experiment using film.
If you take an exposure with a certain f-stop and shutter speed, at a set iso, it will meter the image in a certain way, regardless of how much of the film you cut. Let's say for the the sake of this experiment that this is iso-400 film, metered at f2.8 with 1/60 shutter. If you cut out the sides of the developed film so you've cropped by a factor equivalent to an angle of view that is equivalent of 1.6x your focal length, you've created an aps-c crop. The image won't magically darken.

The same happens if your frame of film only has sensitivity in the area that equates to an aps-c sized film. When metered at f2.8 with a 1/60 shutter, it will give the same image brightness as your "full frame" 35mm film, only with a cropped picture. This is all that a smaller sensor does, at the same f-stop and shutter, given the same sensor properties, it will meter the scene the exact same way on the same lens, only with a cropped field of view.

Now if you stand back to get the same field of view, as long as your light levels do not change in relation to your new relative distance to subject, the scene will still be shot at the same brightness. Camera distance does not affect exposure

[normhead]

What you are missing is that you are changing the Depth of Field by moving the camera back. Depth of field increases as you mover further and further from the subject.

To keep the same FoV and DoF constant, you must shoot with the ISO the same, and stop down the FF to create a roughly equivalent DoF.

But now your exposure is 1 stop under, so you need to either increase the ISO on the FF, or reduce the shutter speed on the FF.

The scenario you post is the illusion that was bantered about by FF enthusiasts for years, but it was smoke and mirrors. The only way you keep the same shutter speed and ISO is to have narrower DoF on the FF. For small birds etc close to the camera, the APS-c camera actually gives you a faster shutter speed than the FF, to help freeze action, using the same ISO and compensating for the same DoF.

The only time the FF actually has an advantage is in the one stop better low ISO performance, meaning you can just raise the ISO one more stop and get an equivalent images, in every respect. Noise, shutter speed, equivalent ISO.

But shooting below 800 ISO on APS-c where there is no clear noise advantage, the advantage goes to APS-c, and even more so at 400 and 200 and 100 ISO. If you can shoot both the APS-c and FF at 400 ISO there won't be a clear noise advantage, the APS-c will just be shooting at a faster shutter speed and freeze the action better.

So yes, f-stop is f-stop, but the same ƒ-stop produces a different depth of field on the different sensor sizes. All the bogus "equivalence, total light" nonsense posted on the internet ignores that. It's a slight of hand thing they do to sound knowledgeable.

[Class A]

F-stop is f-stop, ...

"F-stop is f-stop" is true in the same sense as "Focal length is focal length". Correct, from a certain perspective (physical attributes of a lens don't change, regardless of sensor size), however, incorrect when considering the effect of an attribute, such as f-stop or focal length, on the image, depending on format size.

If you take an exposure with a certain f-stop and shutter speed, at a set iso, it will meter the image in a certain way, regardless of how much of the film you cut.

Correct.

Let's say for the the sake of this experiment that this is iso-400 film, metered at f2.8 with 1/60 shutter. If you cut out the sides of the developed film so you've cropped by a factor equivalent to an angle of view that is equivalent of 1.6x your focal length, you've created an aps-c crop. The image won't magically darken.

Yes, the image will not "magically darken".

However, remember that your film will be the negative (once developed). Now imagine to produce the same image on photographic paper, once using the full negative (FF) and once using a cropped negative (APS-C). You won't be able to. Larger negatives yield better quality images (less noise). That's the whole point of large format photography.

Note that the "cutting sides off" approach is not helpful in making useful comparisons. If you cut sides off a negative and then develop with an unchanged enlargement factor then you'll get a crop of the original image. This crop will be identical in quality to the crop you'd get by cutting sides off the photographic paper, but you can hardly compare two photographic prints where one has twice the size and shows more of the scene, can you?

If you attempt to get the same/similar information into the smaller negative (by using a wider lens, or "standing back") then you will have to use a bigger enlargement factor in order to get the same photographic print as you can get from the original FF negative. The bigger enlargement factor results in more noise in the photographic print. N.B., only using a wider lens (with the f-stop adjusted by a divisor of 1.5) will give you the same information. "Standing back" will result in a different perspective and different DOF.

The enlargement factor (which is higher when using APS-C compared to FF) is easily forgotten about when talking about digital photography, but as soon as you print the same image to a given output format (say 16x20) then you'll notice that capture format (film size or sensor size) does matter.

This is not a straightforward topic. In order to not derail the thread, I'll only say that "f-stop is f-stop" is not a correct statement in every sense of the phrase, just like "focal length is focal length" is not a correct statement in every sense of the phrase.

[Rondec]

I don't think that's quite right.
F-stop is f-stop, regardless of the lens and sensor, given the same optical transparency of the glass and the absolute sensitivity of the light sensor. Use this thought experiment using film.
If you take an exposure with a certain f-stop and shutter speed, at a set iso, it will meter the image in a certain way, regardless of how much of the film you cut. Let's say for the the sake of this experiment that this is iso-400 film, metered at f2.8 with 1/60 shutter. If you cut out the sides of the developed film so you've cropped by a factor equivalent to an angle of view that is equivalent of 1.6x your focal length, you've created an aps-c crop. The image won't magically darken.

The same happens if your frame of film only has sensitivity in the area that equates to an aps-c sized film. When metered at f2.8 with a 1/60 shutter, it will give the same image brightness as your "full frame" 35mm film, only with a cropped picture. This is all that a smaller sensor does, at the same f-stop and shutter, given the same sensor properties, it will meter the scene the exact same way on the same lens, only with a cropped field of view.

Now if you stand back to get the same field of view, as long as your light levels do not change in relation to your new relative distance to subject, the scene will still be shot at the same brightness. Camera distance does not affect exposure

f stops are f stops and focal length is focal length and iso is iso. If 85mm f4, 1/150 sec, and iso 200 give you a good exposure on a K-1, it will do the same on a K3 -- but it won't give you the same picture. And if you back up enough with the K3 that your 85 is giving you the same framing it will still be really different. The only way to get a similar framing/depth of field is to shoot with a shorter lens on APS-C 55mm and to open the aperture to f2.8 and set the iso to 100.

Typically, you will see a one stop difference between APS-C and full frame when it comes to noise control and dynamic range, meaning that shooting iso 800 on a K-1 will look like iso 400 on a K5. This allows you to stop down on your full frame camera, if you need to for adequate depth of field without a penalty.

[keos]

"F-stop is f-stop" is true in the same sense as "Focal length is focal length". Correct, from a certain perspective (physical attributes of a lens don't change, regardless of sensor size), however, incorrect when considering the effect of an attribute, such as f-stop or focal length, on the image, depending on format size.

Without regard to DoF, 50mm at f1.4 taken on a "FF" sensor oran aps-c sensor by walking back to get the same FoV will meter the scene exactly the same way, assuming no light source enters/leaves the scene and the light source is not the camera.

Yes, the image will not "magically darken".

However, remember that your film will be the negative (once developed). Now imagine to produce the same image on photographic paper, once using the full negative (FF) and once using a cropped negative (APS-C). You won't be able to. Larger negatives yield better quality images (less noise). That's the whole point of large format photography.

Note that the "cutting sides off" approach is not helpful in making useful comparisons. If you cut sides off a negative and then develop with an unchanged enlargement factor then you'll get a crop of the original image. This crop will be identical in quality to the crop you'd get by cutting sides off the photographic paper, but you can hardly compare two photographic prints where one has twice the size and shows more of the scene, can you?

If you attempt to get the same/similar information into the smaller negative (by using a wider lens, or "standing back") then you will have to use a bigger enlargement factor in order to get the same photographic print as you can get from the original FF negative. The bigger enlargement factor results in more noise in the photographic print. N.B., only using a wider lens (with the f-stop adjusted by a divisor of 1.5) will give you the same information. "Standing back" will result in a different perspective and different DOF.

The enlargement factor (which is higher when using APS-C compared to FF) is easily forgotten about when talking about digital photography, but as soon as you print the same image to a given output format (say 16x20) then you'll notice that capture format (film size or sensor size) does matter.

This is not a straightforward topic. In order to not derail the thread, I'll only say that "f-stop is f-stop" is not a correct statement in every sense of the phrase, just like "focal length is focal length" is not a correct statement in every sense of the phrase.

a 50mm f1.4 on an K3-II when using the same FoV AND DoF as a "FF" (K-1) will result in a lens equivalence to a 75mm at f2.1. If using an ISO of 100 on the K3-II, when using this lens on a K-1, the ISO has to be increased by a factor equal to the crop factor^2. Meaning to get this exact FoV and DoF on the K1 at that same exposure given the same shutter speed, you must use 75mm@f2.1 @ ISO 225. Why? Because your f-stop increased by 1.5, which results in 2.25 times less total light.

---------- Post added 02-22-18 at 09:09 PM ----------

f stops are f stops and focal length is focal length and iso is iso. If 85mm f4, 1/150 sec, and iso 200 give you a good exposure on a K-1, it will do the same on a K3 -- but it won't give you the same picture. And if you back up enough with the K3 that your 85 is giving you the same framing it will still be really different. The only way to get a similar framing/depth of field is to shoot with a shorter lens on APS-C 55mm and to open the aperture to f2.8 and set the iso to 100.

Typically, you will see a one stop difference between APS-C and full frame when it comes to noise control and dynamic range, meaning that shooting iso 800 on a K-1 will look like iso 400 on a K5. This allows you to stop down on your full frame camera, if you need to for adequate depth of field without a penalty.

You get about one stop of noise advantage at the same ISO; but to get the same FoV and DoF, if you're constrained to the same shutter speed, your ISO has to be pushed one stop higher on the FF. so you lose that advantage.

[peterh337]

"Without regard to DoF, 50mm at f1.4 taken on a "FF" sensor oran aps-c sensor by walking back to get the same FoV will meter the scene exactly the same way, assuming no light source enters/leaves the scene and the light source is not the camera."

That's true because the intrinsic brightness of an object cannot be modified by any optical system. All that optics can do is change the field of view i.e. the magnification