Wildlife, long lenses, and sensor format

[Steve Spencer]

Advertisement (gone after registration)

The amount of light available to be converted into an image is not converted into an image. Only the light falling on a sensor cell is converted into an image and this amount of light is the same regardless of how many cells there are. Bigger cells will have more light falling on them than smaller ones but if they are the same size then whether there are ten or a hundred doesn't change the amount of light falling on each.

 

You almost have it but you mix it up in the end. A crucial question is how much of the light available to be converted into an image is actually converted into an image. I am not talking about sensor cells or pixels or mega pixels here, but the whole sensor. Now the amount of light to which a sensor is exposed is the amount of light per square unit of measurement multiplied by the the number of square units. So, if a smaller sensor has 4 times more light per square unit of measurement but only has a 1/4 of the area of the larger sensor then both sensors are exposed to the same amount of light. This is simple math. Amount of light per square unit times number of units equals the total amount of light. That can no more be wrong than 2 + 2 = 4 or more precisely 4 X 1/4 = 1. Now the implications of this mathematical fact I will discuss below but I haven't even got to that yet, because Jaap has derailed the discussion by trying to put words in my mouth and failing to acknowledge a simple mathematical fact. And that mathematical fact is that if we take the same 100-400mm lens and use it without a teleconverter on an M4/3rds sensor and with a 2X teleconverter on a FF 35mm sensor, the two sensors will both have the same amount of light available to convert to an image. The M4/3rds sensor will have 4 times more light per square millimetre, but have 1/4th the number of square millimetres 4 X 1/4 = 1. If we can agree on this simple mathematical fact, then we can discuss what the implications of it are. I haven't said anything about that yet, because this basic premise has not been understood or acknowledged, and I would hope that Jaap and anyone else could wait until the argument has been made before they state with certainty that the argument is wrong. One should at least understand and at least be presented with an argument before they state with certainty that an argument is wrong.

[Steve Spencer]

I'm sorry, I cannot make myself more clear than I did in my previous posts, but you have it wrong if you equate cropping by sensor size to using an extender.

Your argument only explains why a FF lens must be larger than a 4/3rds one.

 

Why don't you wait until I actually make the argument before you declare that it is wrong. All I have argued at this point is that the total amount of light to which a sensor is exposed is the amount of light per square unit multiplied by the number of units and that a 100-400mm lens attached to a camera with an M4/3rds sensor will have 4 times the amount of light per square millimetre but 1/4 the number of square millimetres as that same lens with a 2X teleconverter attached to a camera with a FF 35mm lens. This is simple math 4 X 1/4 = 1. If we can acknowledge that basic premise without telling me I am wrong, I can then go on to make an argument about the implications of this mathematical fact with which you may or may not agree, but I think simple logic and basic courtesy requires that you actually wait until the argument is made before you pronounce it is wrong and I haven't even made the actual argument yet. I was just trying to establish a simple mathematical premise with which I would hope everyone who can multiply 4 X 1/4 could agree.

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[pop]

Why don't you wait until I actually make the argument ...

Perhaps it would be useful if you started telling someone why you found that simple mathematical statement so important, i.e. what its impact on the photograph might be in your opinion. I don't think anyone has disagreed with x/x = 1, so far. 

[jaapv]

Why don't you wait until I actually make the argument before you declare that it is wrong. All I have argued at this point is that the total amount of light to which a sensor is exposed is the amount of light per square unit multiplied by the number of units and that a 100-400mm lens attached to a camera with an M4/3rds sensor will have 4 times the amount of light per square millimetre but 1/4 the number of square millimetres as that same lens with a 2X teleconverter attached to a camera with a FF 35mm lens. This is simple math 4 X 1/4 = 1. If we can acknowledge that basic premise without telling me I am wrong, I can then go on to make an argument about the implications of this mathematical fact with which you may or may not agree, but I think simple logic and basic courtesy requires that you actually wait until the argument is made before you pronounce it is wrong and I haven't even made the actual argument yet. I was just trying to establish a simple mathematical premise with which I would hope everyone who can multiply 4 X 1/4 could agree.

Because you started this whole argument in response to this post:

Except for, of course, the light loss of converters.

Let's get back to the origin of the argument. I am saying that using an 400/6.3 lens on a 4/3rds sensor is better than using a 400/4.0 on full frame with extender because

1. The 4/3rds lens will remain a 6.3 lens with the angle of view of an 800 lens on FF

2. The FF lens will turn into an 8.0 lens and in practical use (because you have to stop down) an 11.2 lens.

Which is unpractical in use.

The only way to get close to the 4/3rds image is to crop the FF image, but then you well get into motion blur problems for three reasons:

You need a 80 MP FF sensor.

Your I.S. effect (if at all present) will be quartered.

The whole combo will be larger and heavier.

[Exodies]

<snip>

if a smaller sensor has 4 times more light per square unit of measurement but only has a 1/4 of the area of the larger sensor then both sensors are exposed to the same amount of light. <snip>

Assume both sensors have the same cell size. Take "square unit of measurement" to be this cell size.

You are proposing the small sensor receives 4 times more light per cell.

The small sensor picture will therefore be 4 times brighter.

[gyoung]

Of course and I never said otherwise. I am not talking about exposure. I am talking about the amount of light that is available to convert into an image and that is important in its own right.

The ability of a lens to cover a larger format is governed by its designed circle of illumination, which also dictates sizes of elements and thus overall size of the lens. Amounts of light are generally measured specified by area, lumens per sq metre etc (in my youth 'foot candles'). If a lens has a larger circle of illumination it does not have a lower 'brightness' over a given area of the focal plane. All other things being equal such as aperture, transmission efficiency etc.

 

Gerry

[Steve Spencer]

Advertisement (gone after registration)

Because you started this whole argument in response to this post:

 

Let's get back to the origin of the argument. I am saying that using an 400/6.3 lens on a 4/3rds sensor is better than using a 400/4.0 on full frame with extender because

1. The 4/3rds lens will remain a 6.3 lens with the angle of view of an 800 lens on FF

2. The FF lens will turn into an 8.0 lens and in practical use (because you have to stop down) an 11.2 lens.

Which is unpractical in use.

The only way to get close to the 4/3rds image is to crop the FF image, but then you well get into motion blur problems for three reasons:

You need a 80 MP FF sensor.

Your I.S. effect (if at all present) will be quartered.

The whole combo will be larger and heavier.

 

No, Jaap, your, "except for the light loss comment" was in response to what I had written in which I said that a 100-400 f/5.6 lens with a 2X teleconverter would have the same angle of view as a 100-400 f/5.6 lens on M4/3rds, it would also have the DOF capabilities, and it could have very similar weight, and in those ways it could be a very similar option. To that you replied, "except for the light loss of the tele extender." So let's not further confuse things by talking about f/4 lens, etc. This is about comparing a 100-400 f/5.6 lens on M4/3rds with a 100-400 f/5.6 lens with a 2 teleconverter on FF 35mm. I had also argued that with longer lenses image circle is typically not a limiting factor because longer lenses naturally typically have much wider image circles than are needed for the format. We can discuss this point separately, but let's not get derailed there either. So let's start with the same 100-400 f/5.6 lens which has a wide enough image circle to cover both M4/3rds and FF 35mm and let's use it on the M4/3rds sensor without a teleconverter and on the FF sensor with a 2X teleconverter. The first thing I would like to note is that in both cases the sensors will have the same amount of light available to convert into an image and I think I have established that mathematically at this point.

 

Just so we don't get derailed again, please note I am not saying they will have the same exposure parameters. I know they won't so let's not go down that road again. Instead, I think it is useful to think about this situation in comparison to a common situation that many wildlife photographer's face regardless of the format they use. That situation is they would like more reach to photograph something in the distance with more detail but they wonder should I use a teleconverter or should I just crop the image. In practice it is actually easy to do this comparison. How does a 4X crop compare to using a 2X teleconverter?` Here you are using the same camera with of course the same sensor, and in practice sometimes one result is better and sometimes the other result is better. Now let's talk about exposure in this example. Well if you want to keep the same exposure to the two shots, which you would, and you want to keep the same shutter speed, which you often would, then when you added the 2X teleconverter you would need to turn the ISO up 2 stops. Does this mean that image with the 2X tele extender will have more noise, and less dynamic range basically automatically have worse image quality? If you have done such comparisons, then you will know that the answer is no. The image with the 2X tele extender that was shot at two stop higher ISO looks remarkably similar in noise and dynamic range to the shot with two stops lower ISO that has a 4X crop. Why is that?  Because cropping increases the prominence of noise and decreases the dynamic range, but fundamental to what I have been arguing it does that because cropping reduces the amount of light that is used when an image is created. Here as in the example I was discussing above the image created with the 2X converter and two stops higher ISO has exactly the same amount of light to convert into the image as the shot created by the 4X crop. Further when we compare the shot using the 2X converter at two stops higher ISO with the shot using a 4X crop it is the same sensor that is converting the same amount of light, so of course we get a very similar looking image in noise and dynamic range. The basic lesson is amount of light being converted into an image multiplied by how well a sensor converts that light into an images gives us how much noise and dynamic range and other properties of the image we will see. 

 

Ok, now let's apply this lesson to the case I was originally discussing with you , Jaap. We have a M4/3rds camera with a 100-400 f/5.6 lens and a FF 35mm camera with a 100-400 f/5.6 lens with a 2X converter. Now let's further assume that the two camera have identical sensors except the FF 35mm sensor is of course 4X bigger. Now we have exactly the same situation we had in the cropping vs. 2X teleconverter on the same camera example above. If we are going to to have same exposure and shutter speed we will have to shoot the FF camera at two stops higher ISO. Will this mean that the M4/3rds image will have better image quality? No, because in this example just as above the 2 sensors are identical except in size. They would have remarkably similar imaging properties because the amount of light that is being converted by how well the sensor converts light is the same.  

 

Now let's apply what we have learned here to real world comparisons of M4/3rds cameras to FF 35mm cameras. I am not disputing that M4/3rds cameras often can produce better images in these instances, what I am disputing is the inherent advantage of the format or more specifically I want to be crystal clear on what the advantage of the format actually is. I think right now M4/3rds cameras have two advantages and one of these advantages is not format dependent and in not inevitable and the other advantage is format dependent but is small for long lenses. First, let's start with the advantage that is not format dependent. If we think of the example described directly above, M4/3rds sensors and FF 35mm sensors are not exactly the same except for size. M4/3rds sensors are actually better if we compare the modern 20mp sensor with even the modern 42mp BSI sensor in the Sony A7r II, and they are better in a very specific way--they convert the amount of light to which they are exposed more efficiently with less noise and dynamic range. This advantage, however, is not format dependent and is not inevitable, and there are reasons to expect it to get smaller. As more and more of the market shifts to FF 35mm and as one of the chief players in FF 35mm (Sony) becomes more invested in FF 35mm lenses, then we can expect the gap in the ability of M4/3rds sensors to convert the light to which they are exposed well to shrink and perhaps even reverse. So, this advantage, although real and important right now is not an advantage of the format. It is not inevitable and it should not be treated as such.

The second advantage which I avoided talking about earlier is that in the examples above I assume that the two lens systems (the 100-400 f/5.6 without a 2X teleconverter and with a 2X teleconverter) will be equally easy to design and will produce equal IQ. This assumption is in fact not true and here in lies the real advantage of M4/3rds. It will be easier and cheaper to design good lenses for the format, but the advantage should not be overstated either. As I reviewed earlier and you can easily see if you examine MF and LF lenses, at longer focal lengths it is fairly easy to make lenses with wider image circles, which negates the advantage somewhat. There is also the problem that tele extenders have often been made poorly in the past, but that does not mean they have to be in the future. The bottom line for me is that if you want to hold up a lens like the Panny Leica 100-400 f/4.5-6.3 as a standard (which is a very good lens), I still think it is perfectly reasonable to expect that a FF 35mm lens that is about 3 times the cost could be produced with an excellent 2X converter which has similar quality in terms of creating an image. The Sony 100-400 f/4.5-5.6 with their 2X converter may not be far off and as we compare images we will know more. The Panny/Leica is smaller but the Sony is all that big, and as reviewed earlier much of the smaller size of the Panny/Leica is the choice to make the max aperture a half stop smaller, which allows the lens to be considerably smaller and the smaller size only has a little bit to do with the smaller sensor size.

[jaapv]

For one thing you are wrong about the DOF, and about the aperture change The Sony lens you hold up as an example will be a 200-800 9.0-11.2  with a 2x converter and far too slow with inferior stabilization (because of the converter).  I already responded to the rest of you arguments elaborately. I summarized in post # 104 and won't argue this any further.

[Exodies]

@Steve Spencer - please explain the difference between cropping by having a smaller sensor and cropping in pp. The latter certainly changes nothing about the exposure and noise and the former only changes exposure and noise if the cell size is different.

If the point you are making is due to different cell sizes then why won't you state this clearly?

[Steve Spencer]

For one thing you are wrong about the DOF, and about the aperture change The Sony lens you hold up as an example will be a 200-800 9.0-11.2  with a 2x converter and far too slow. I already responded to the rest of you arguments elaborately. I summarized in post # 104 and won't argue this any further.

 

Again you are attributing things to me that I am not saying that I agree are wrong and then you say I am wrong. That sort of disrespectful behaviour is  pretty lame. I agree that the Sony lens is a 200-800 f/9 to 11.2 with the 2X teleconverter. When did I say otherwise? The answer is never. I can't even tell what you think I said about DOF, but I said very little and that certainly wasn't my main point. If you want to interact with what I actually said that would  be fine, but please stop saying I made arguments that I clearly did not and then say those arguments are wrong. You might as well say, you are wrong. You said pigs can fly and they most certainly cannot.

[Steve Spencer]

@Steve Spencer - please explain the difference between cropping by having a smaller sensor and cropping in pp. The latter certainly changes nothing about the exposure and noise and the former only changes exposure and noise if the cell size is different.

If the point you are making is due to different cell sizes then why won't you state this clearly?

Actually cropping in post processing has a big effect on noise and dynamic range because it affects magnification. How noticeably the noise is, certainly is affected by magnification and that is why downsizing an image dramatically reduces noise, and by the same token cropping an image and viewing it at the same size dramatically makes noise more prominent. My argument has nothing to do with cell size, but rather about sensor size.

Edited May 3, 2017 by Steve Spencer

[Steve Spencer]

For one thing you are wrong about the DOF, and about the aperture change The Sony lens you hold up as an example will be a 200-800 9.0-11.2  with a 2x converter and far too slow with inferior stabilization (because of the converter).  I already responded to the rest of you arguments elaborately. I summarized in post # 104 and won't argue this any further.

Oh, and by the way you have not already responded to the rest of my argument, because I had not made it until now. What you have displayed is a clear misunderstanding of my argument. If you take the time to understand what I am saying and not pretend I am saying things that I am not then maybe you can respond to it, but unless you are clairvoyant you didn't respond to the argument I just made earlier because I didn't make it until now. That you think you responded just shows how little attention you are actually paying to the arguments that have been made.

[jaapv]

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

 

 

 

 

used courtesy of Africa Check, a non-profit organisation which promotes accuracy in public debate www.africacheck.org
 

[Exodies]

Actually cropping in post processing has a big effect on noise and dynamic range because it affects magnification. How noticeably the noise is, certainly is affected by magnification and that is why downsizing an image dramatically reduces noise, and by the same token cropping an image and viewing it at the same size dramatically makes noise more prominent. My argument has nothing to do with cell size, but rather about sensor size.

Ah. I think we are getting closer. You include resizing in the act of cropping; I don't.

So the discussion isn't about sensor size, it's about magnification of the image.

[Steve Spencer]

Ah. I think we are getting closer. You include resizing in the act of cropping; I don't.

So the discussion isn't about sensor size, it's about magnification of the image.

 

Yes, we are getting closer. The discussion includes consideration of magnification, but the it is related to sensor size, so it is about sensor size as well. 

[jmahto]

Because you started this whole argument in response to this post:

 

Let's get back to the origin of the argument. I am saying that using an 400/6.3 lens on a 4/3rds sensor is better than using a 400/4.0 on full frame with extender because

1. The 4/3rds lens will remain a 6.3 lens with the angle of view of an 800 lens on FF

2. The FF lens will turn into an 8.0 lens and in practical use (because you have to stop down) an 11.2 lens.

Which is unpractical in use.

The only way to get close to the 4/3rds image is to crop the FF image, but then you well get into motion blur problems for three reasons:

You need a 80 MP FF sensor.

Your I.S. effect (if at all present) will be quartered.

The whole combo will be larger and heavier.

I picked this response (#104) since you quoted it later as summarized response. If I may, let me add few thoughts to this....

 

You said in second point: "The FF lens will turn into an 8.0 lens and in practical use (because you have to stop down) an 11.2 lens. Which is unpractical in use."

 

Which is correct. However, you can bump up the ISO of FF sensor to 2 stops in order to keep the same shutterspeed as of MFT+400mm f/6.3 lens. Since FF sensor has fatter pixels with better signal to noise ratio compared to MFT, the noise level on the final picture from both the sensors (FF@ 800 ISO and MFT@ 200 ISO)will be kind of similar. I am assuming similar advancement in sensor technology in both sensor sizes. Basically what I am pointing out is that there is no free lunch in MFT sensor with more densely packed pixels.

 

However, the 400mm f/4 lens on FF has to be extremely good to be used with 2x extender projecting a much bigger image circle (compared to MFT's 400mm f/6.3 lens). This will force FF lens to be bigger in size. Basically, FF lens with extender is wasting it's bigger potential. In contrast, MFT lens is being used to it's full potential and can remain compact.This is where I think MFT has advantage.

I stand to be corrected.

Edited May 4, 2017 by jmahto

[jaapv]

Yes, you would want to use a Telyt 280/4.0 APO because it is diffraction limited and still acceptable in size-weight (1750 gram IIRC) plus the extender, of course, or the  400 of  the Modular system. Not only real heavy and large, but serious $$$$$ as well.

And you would still miss stabilization, or, if the camera/lens does support IS, the effectiveness will be diminished by an extender.

[Steve Spencer]

However, you can bump up the ISO of FF sensor to 2 stops in order to keep the same shutterspeed as of MFT+400mm f/6.3 lens. Since FF sensor has fatter pixels with better signal to noise ratio compared to MFT, the noise level on the final picture from both the sensors (FF@ 800 ISO and MFT@ 200 ISO)will be kind of similar. I am assuming similar advancement in sensor technology in both sensor sizes. Basically what I am pointing out is that there is no free lunch in MFT sensor with more densely packed pixels.

However, the 400mm f/4 lens on FF has to be extremely good to be used with 2x extender projecting a much bigger image circle (compared to MFT's 400mm f/6.3 lens). This will force FF lens to be bigger in size. Basically, FF lens with extender is wasting it's bigger potential. In contrast, MFT lens is being used to it's full potential and can remain compact.This is where I think MFT has advantage.

I stand to be corrected.

 

The first point that you made is essentially one of the points I was trying to make. That is if an M4/3rds camera and a FF 35mm camera have the same technology, you can turn up the ISO 2 stops on the FF camera and shoot a lens that has a 2 stop slower aperture. The reason, however, is not because of fatter pixels but rather because the FF sensor if it has the same technology will have 4 times as many pixels and you can then downsize the image to a quarter of its captured size and it will be equal in size and noise to the M4/3rds image. 

 

I don't agree with your second point. Although any lens will need a 4 times larger image circle to cover FF 35mm than M4/3rds, at 400mm it does not require a large lens to cover the 43mm image circle needed for FF 35mm. For example, the Rodenstock 360mm f/9 APO-Ronar covers a 318mm image circle (over 7 times larger than in needed for FF 35mm) and weighs just 550g. Now of course large format lenses are different and they need a bellow, etc., but they demonstrate that large image circles at long focal lengths are easy to obtain. The much more difficult task for lens design at long focal lengths is create wider apertures and keep the size small. You can see the too with large format lenses. For example, the Rodenstock 360mm F/6.8 APO-Sironar-N weighs 1560g. That one stop in wider aperture basically triples the weight. 

 

I think we see the same thing when we compare the three mirrorless 100-400mm lenses. The Sony lens is a 100-400 f/4.5-5.6 and weighs 1395g. The Fuji lens is also a 100-400 f/4.5-5.6 lens and only covers the much smaller APS-C image circle. Yet is weighs only 20g lighter at 1375g. Both lenses are image stabilized and both claim 5-stop image stabilization. If image circle were central to lens size then you would think the Fuji would be much smaller compared to the Sony than it is. Interestingly, we can compare both of these lenses with the Panny/Leica 100-400 f/4-6.3. The Panny Leica weighs just 985g and only needs to cover the much smaller yet M4/3rds image circle, but is it the smaller image circle or the smaller max aperture at the long end that allows the Panny/Leica to be smaller? Both probably contribute some, but I am arguing the weight reduction is primarily about the slower max aperture at the long end (just like we see in comparing long LF lenses such as the APO-Ronar f/9 and the APO-Sironar-N f/6.8) and not very much about the smaller image circle.

[jaapv]

What FF sensor has 80 MP?

[Advertisement]

Hallo,

Look here: Interesting Leica gear

[Steve Spencer]

What FF sensor has 80 MP?

 

None right now, but Sony is rumored to have one at almost that resolution (70mp) ready to be introduced. And the point that I tried to make above was that although M4/3rds clearly do have an advantage in their current sensors, that is not inherent in the format and is dependent on which sensors get the newest technology. Right now M4/3rds sensors have been getting the new tech. It isn't that hard to imagine, however, that Sony who makes the majority of sensor might reverse that trend in the not too distant future and start introducing new tech with FF 35mm sensors (and they have already with the stacked sensor in the A9). So yes, right now M4/3rds sensors have a tech advantage, but that is not inevitable.