Debunking Tony Northrup.

[Edge100]

I don't know how to explain this any more simply. If a MFT sensor and a FF sensor have the same total number of megapixels, the MFT's sensor sites are 1/4 the size. They are NOT THE SAME SIZE.

I agree. I never said ANYTHING about the # of megapixels. Again, consider a FF sensor and m4/3 sensor *cut from the same silicon*.

Obviously, this is hypothetical. An 18MP m4/3 sensor will not have the same noise performance as an 18MP FF sensor, but that doesn't have anything WHATSOEVER to do with the sensor size. It has to do with the size of the photosites collecting the light.

The effect of sensor size ALONE is only down to the amount the image must be enlarged.

Again, why does large format film show less grain than 35mm film for the same print size? It's because the LF film required less enlargement. Now, if you shoot ISO50 film in 35mm and ISO 1600 film in LF (akin to changing the photosite size in digital), then the grain might be similar in the end. But this doesn't have anything AT ALL to do with the size of the film.

And so it is with sensors.

 

[winkosmosis]

I agree. I never said ANYTHING about the # of megapixels. Again, consider a FF sensor and m4/3 sensor *cut from the same silicon*.

Obviously, this is hypothetical. An 18MP m4/3 sensor will not have the same noise performance as an 18MP FF sensor, but that doesn't have anything WHATSOEVER to do with the sensor size. It has to do with the size of the photosites collecting the light.

The effect of sensor size ALONE is only down to the amount the image must be enlarged.

Again, why does large format film show less grain than 35mm film for the same print size? It's because the LF film required less enlargement. Now, if you shoot ISO50 film in 35mm and ISO 1600 film in LF (akin to changing the photosite size in digital), then the grain might be similar in the end. But this doesn't have anything AT ALL to do with the size of the film.

And so it is with sensors.

Yes the light collected by each pixel would be the same in such a scenario because it's effectively a crop. But what is the relevance of that?

You might as well say "14mm on MFT and 14mm on FF are the same focal length. Yes they're the same focal length but that is irrelevant. That's why we convert to *equivalent* focal lengths, so we can compare different formats.

MFT companies will tell you that the 30mm lens is equivalent to 60mm, but they won't tell you that the f2.8 aperture is really equivalent to f5.6. Because they don't want you to realize how bad the system is in low light simply due to the lens and sensor being smaller.

 

[Edge100]

Yes the light collected by each pixel would be the same in such a scenario because it's effectively a crop. But what is the relevance of that?

You might as well say "14mm on MFT and 14mm on FF are the same focal length. Yes they're the same focal length but that is irrelevant. That's why we convert to *equivalent* focal lengths, so we can compare different formats.

The 'relevance' is that it's true. It's especially true for those of us who shoot film, since in that case we actually do have the same 'sensor' in multiple formats.

The bottom line is this: sensor size affects noise performance only because images captured on smaller sensors need more enlargement to achieve the same print size.

If you want to discuss the effect of photosite size on noise, then that's a different conversation.

Also, 'equivalent' focal lengths is also misnomer. What you mean is "equivalent angle of view". Focal length is a property of a lens, not of a sensor.

 

[juanm]

Also, 'equivalent' focal lengths is also misnomer. What you mean is "equivalent angle of view". Focal length is a property of a lens, not of a sensor.

Welcome to the fight against misunderstood concepts. The person who invented the term "equivalent focal length" deserves prison. Its negative effect on mankind will never wear off.

 

[Meister]

Welcome to the fight against misunderstood concepts. The person who invented the term "equivalent focal length" deserves prison. Its negative effect on mankind will never wear off.

I actually think this thread is quite educational.
You are right about the term equivalent focal length. It should be called equivalent FOV.

 

[tomnavratil]

Are there any YouTube personalities that are just rock solid in their knowledge of the craft? That's a legit question. I watch Kelby, Fro, Tony.. even DigitalRev strictly for entertainment, and don't think of them as serious instructors. I also don't try and work on my Honda following YouTube DIY videos, but that's just me.

AdoramaTV is a very decent channel where I like to watch videos from Gavin Hoey or Mark Wallace who are able to provide very clear and precise knowledge. Also Chase Jarvis has very good level of knowledge and I like his channel.

 

[winkosmosis]

The 'relevance' is that it's true. It's especially true for those of us who shoot film, since in that case we actually do have the same 'sensor' in multiple formats.

The bottom line is this: sensor size affects noise performance only because images captured on smaller sensors need more enlargement to achieve the same print size.

If you want to discuss the effect of photosite size on noise, then that's a different conversation.

Also, 'equivalent' focal lengths is also misnomer. What you mean is "equivalent angle of view". Focal length is a property of a lens, not of a sensor.

You're thinking of "enlarging" as if it's film. But it's not. We have pixels, and to some degree resolution cancels out noisiness of individual pixels as you increase in megapixels. So it makes the most sense to compare absolute sensor sizes, IE total light collected by the sensor. A 30mm f2.8 lens on MFT equivalent to a 60mm f5.6 lens on FF in that 1) it's gathering the same amount of light and b) it's producing the same DOF c) same field of view

In what way is it not equivalent?

Whatever you call it, equivalent field of view is a useful number. Likewise, equivalent aperture is useful, but it isn't used. The only reason for that is that giving equivalent aperture makes the smaller formats look bad, so it's all about marketing. They'll tell you the 200mm MFT lens is 400mm equivalent but they won't tell you the f5.6 aperture is equivalent to f11 for light gathering and DOF

 

[Edge100]

Whatever you call it, equivalent field of view is a useful number. Likewise, equivalent aperture is useful, but it isn't used. The only reason for that is that giving equivalent aperture makes the smaller formats look bad, so it's all about marketing. They'll tell you the 200mm MFT lens is 400mm equivalent but they won't tell you the f5.6 aperture is equivalent to f11 for light gathering and DOF

They won't tell you it's equivalent for light gathering because that isn't true.

 

[simonsi]

That's what I said, intensity of light is the same. But the TOTAL LIGHT hitting the 1/1.7" sensor is far lower. That's why small sensors perform worse in low light. For the same ISO on a tiny sensor and a full frame one, the tiny sensor is being fed more voltage to reach the same exposure

Total light is totally irrelevant, even in sensors it is the light per photosite that matters, each photosite doesn't care how big the sensor is or how much total light is received by all sites. When you come to amplifying the signal it is the difference per photosite that makes the design difference, not any aspect of total light hitting the sensor.

On any aspect of this, the total light hitting the sensor really doesn't matter, its all about the light per pixel or photosite or sq mm (for film), its all about intensity.

----------

I actually think this thread is quite educational.
You are right about the term equivalent focal length. It should be called equivalent FOV.

Hmmm - but a particular lens/sensor combo will have an actual FoV, its measurable and isn't equivalent to anything so might as well be used directly.

"Equivalent focal length" is just a means to describe an effect that isn't actually a focal length, if you don't want to measure and quote the actual field of view then quoting it as the same as another focal length on a differently sized sensor has its uses and obvious pitfalls.

I could describe the DX view I have as "3 chickens" it is just a descriptive term and only useful when I tell you that the FX view I would describe as "4.5 chickens".

This is where TN's wheels fall off, the physics of a lens don't change with sensor size - and that includes the exposure of a scene viewed through it.

----------

They won't tell you it's equivalent for light gathering because that isn't true.

LOL - a valid point, well made

 

[winkosmosis]

They won't tell you it's equivalent for light gathering because that isn't true.

Yes it is true. A MFT sensor is gathering 1/4 the light compared to a FF sensor with the same aperture, because the sensor is 1/4 the size. For a given aperture it's getting the same brightness, but over a 1/4 smaller area.

Think of it like a solar cell. Let's say you have 60 square feet of solar cells on your house that provides all the power you need. A hurricane comes and blows away 45 square feet, leaving you with 15. Your argument is basically "The 15 can power my house just as well because the sun is just as bright".

Practically, at the same resolution that means 1/4 light hitting each pixel, assuming same resolution for simplicity's sake. Hence more noise at any given ISO. I feel like I'm repeating myself and you aren't getting it!

----------

Total light is totally irrelevant, even in sensors it is the light per photosite that matters, each photosite doesn't care how big the sensor is or how much total light is received by all sites. When you come to amplifying the signal it is the difference per photosite that makes the design difference, not any aspect of total light hitting the sensor.

On any aspect of this, the total light hitting the sensor really doesn't matter, its all about the light per pixel or photosite or sq mm (for film), its all about intensity.

Actually... It turns out the light per photosite isn't what matters that much for resulting image quality. You know how the Sony A7S and A7R have the same sensor size but one has 12MP and one has 36MP? Well if you view the pixels at 1:1 yes the 36MP one is noisier, but if you look at both images at the same size, they are about the same. They're both collecting the same total amount of light, just using it differently.

That doesn't hold true if you compare the same resolution between sensor sizes. For any given generation of tech, an APS-C has an equivalent noise level at half the ISO as a FF. And MFT has an equivalent noise level at 1/4 the ISO as FF.

Hmmm - but a particular lens/sensor combo will have an actual FoV, its measurable and isn't equivalent to anything so might as well be used directly.

"Equivalent focal length" is just a means to describe an effect that isn't actually a focal length, if you don't want to measure and quote the actual field of view then quoting it as the same as another focal length on a differently sized sensor has its uses and obvious pitfalls.

I could describe the DX view I have as "3 chickens" it is just a descriptive term and only useful when I tell you that the FX view I would describe as "4.5 chickens".

This is where TN's wheels fall off, the physics of a lens don't change with sensor size - and that includes the exposure of a scene viewed through it.

OK, so you want to describe equivalent field of view in degrees? But then you have to retrain people who already think in terms of focal length. The simplest method is the one we use now... Equivalent focal length.

THE pitfall of equivalen focal length is that equivalent aperture is not given along with it, so people end up with inflated expectations of their f1.8 MFT lens.

The only thing that really makes sense is, as Northrup argues, is to give the consumer the equivalent focal length plus the equivalent aperture. No it's not the actual focal length and actual aperture (which are also given). Duh. That's why it's called **equivalent**.

 

[Edge100]

Yes it is true. A MFT sensor is gathering 1/4 the light compared to a FF sensor with the same aperture, because the sensor is 1/4 the size. For a given aperture it's getting the same brightness, but over a 1/4 smaller area.

The light per unit area is the same, irrespective of sensor size.

Think of it like a solar cell. Let's say you have 60 square feet of solar cells on your house that provides all the power you need. A hurricane comes and blows away 45 square feet, leaving you with 15. Your argument is basically "The 15 can power my house just as well because the sun is just as bright".

You have to integrate the light over the area in this situation, of course. But that's not what's happening in photography. The exposure is what it is for the exposed area.

In your example, the power output per unit area is unaffected.

Practically, at the same resolution that means 1/4 light hitting each pixel, assuming same resolution for simplicity's sake. Hence more noise at any given ISO. I feel like I'm repeating myself and you aren't getting it!

No. It. Does. Not. The same amount of light hits each pixel. You just have fewer pixels (or smaller pixels, but we've already agreed that that is a different issue).

Actually... It turns out the light per photosite isn't what matters that much for resulting image quality. You know how the Sony A7S and A7R have the same sensor size but one has 12MP and one has 36MP? Well if you view the pixels at 1:1 yes the 36MP one is noisier, but if you look at both images at the same size, they are about the same. They're both collecting the same total amount of light, just using it differently.

That's because you're effectively binning pixels at 3:1 when you go from 36MP to 12MP.

OK, so you want to describe equivalent field of view in degrees? But then you have to retrain people who already think in terms of focal length. The simplest method is the one we use now... Equivalent focal length.

Why do you want to describe differences between systems anyway? Just remember that on FF, a 50mm is a normal, while on 4x5, it's ultra wide angle, while on m4/3, it's a short tele.

THE pitfall of equivalen focal length is that equivalent aperture is not given along with it, so people end up with inflated expectations of their f1.8 MFT lens.

No, they don't, if they expect it to require the same shutter speed and ISO as they would use on FF, to give the same exposure. If what you were saying was true, you'd need to tell an external light meter what sensor size you're using, in order for it to determine correct exposure. Clearly, you don't.

The only thing that really makes sense is, as Northrup argues, is to give the consumer the equivalent focal length plus the equivalent aperture. No it's not the actual focal length and actual aperture (which are also given). Duh. That's why it's called **equivalent**.

It's not equivalent.

 

[Meister]

No. It. Does. Not. The same amount of light hits each pixel. You just have fewer pixels (or smaller pixels, but we've already agreed that that is a different issue).

Actually, when pixels are smaller, less light hits them. So he is correct that with a smaller sensor, but at the same resolution, the amount of light gathered by each pixel is less.

 

[Edge100]

Actually, when pixels are smaller, less light hits them. So he is correct that with a smaller sensor, but at the same resolution, the amount of light gathered by each pixel is less.

Yes. Which is why I said "but that is a different issue". That has nothing to do with sensor size, per se.

If resolution is the same, m4/3 sensors have smaller pixels than FF sensors. Which means that each pixel collects less light, and consequently noise is increased. On the other hand, if you bin pixels on a higher MP camera of the same physical size (as per the A7S/A7R example above), you essentially average the noise out over those binned pixels. This is why we refer to pixel-level noise (i.e. at 1:1) vs. output-level noise (i.e. at your intended print size). If generation B's camera has the same pixel level noise as generation A, but has twice the resolution, then the OUTPUT level noise will be much lower. At the end of the day, this is what matters.

None of that is in dispute.

What IS in dispute is that this is the reason that SENSOR SIZE affects noise. It's not. It's why PIXEL SIZE affects noise.

Ceteris paribus, smaller sensors (or smaller film) affect noise (or grain) because they need to be magnified more than larger sensors (or film) to obtain the same output. That's it.

 

[simonsi]

Yes it is true. A MFT sensor is gathering 1/4 the light compared to a FF sensor with the same aperture, because the sensor is 1/4 the size. For a given aperture it's getting the same brightness, but over a 1/4 smaller area.

Think of it like a solar cell. Let's say you have 60 square feet of solar cells on your house that provides all the power you need. A hurricane comes and blows away 45 square feet, leaving you with 15. Your argument is basically "The 15 can power my house just as well because the sun is just as bright".

Practically, at the same resolution that means 1/4 light hitting each pixel, assuming same resolution for simplicity's sake. Hence more noise at any given ISO. I feel like I'm repeating myself and you aren't getting it!

----------



Actually... It turns out the light per photosite isn't what matters that much for resulting image quality. You know how the Sony A7S and A7R have the same sensor size but one has 12MP and one has 36MP? Well if you view the pixels at 1:1 yes the 36MP one is noisier, but if you look at both images at the same size, they are about the same. They're both collecting the same total amount of light, just using it differently.

That doesn't hold true if you compare the same resolution between sensor sizes. For any given generation of tech, an APS-C has an equivalent noise level at half the ISO as a FF. And MFT has an equivalent noise level at 1/4 the ISO as FF.




OK, so you want to describe equivalent field of view in degrees? But then you have to retrain people who already think in terms of focal length. The simplest method is the one we use now... Equivalent focal length.

THE pitfall of equivalen focal length is that equivalent aperture is not given along with it, so people end up with inflated expectations of their f1.8 MFT lens.

The only thing that really makes sense is, as Northrup argues, is to give the consumer the equivalent focal length plus the equivalent aperture. No it's not the actual focal length and actual aperture (which are also given). Duh. That's why it's called **equivalent**.

So to prove the aperture debate. Consider an FX camera that has a DX mode - so the pixels are the same size, the ISO-amplification algorithms are the same, for a given lens the intensity of the light is the same (ignore any corner falloff defects). If you switch the camera between FX and DX modes in Auto does the camera choose a different exposure setting? No it doesn't.

Yes I was agreeing that "equivalent focal length" is quite valid (as is chickens if we all adopted that), but to use "equivalent FoV" is a bit of nonsense as there is a perfectly good actual FoV to be measured...and using "equivalent focal length" also has benefits in guiding minimum shutter speed (ignoring VR/VC), assessments for handholding long lenses.

TN giving the consumer an "equivalent aperture" is a nonsense, it makes the lens description far worse (ie more misguiding), in practical use than the existing physical description ever is - and that is ignoring the fact that the physical description is still correct.

 

[Edge100]

So to prove the aperture debate. Consider an FX camera that has a DX mode - so the pixels are the same size, the ISO-amplification algorithms are the same, for a given lens the intensity of the light is the same (ignore any corner falloff defects). If you switch the camera between FX and DX modes in Auto does the camera choose a different exposure setting? No it doesn't.

Precisely. And this is akin to the situation with different film sizes, in which the 'sensor' has the same characteristics, but is simply larger or smaller.

When I shoot 4x5 (Wista VX, FWIW) or 35mm (Leica MP, FWIW), the exposure is the same. If I select f/16, 1/125 on Portra 400 with my Leica or on the Wista, I get the same exposure on the film (using different focal lengths, of course... ). But when I make prints from those negatives, the grain is a LOT smaller on the 4x5, because the enlargement is a LOT less.

 

[simonsi]

Precisely. And this is akin to the situation with different film sizes, in which the 'sensor' has the same characteristics, but is simply larger or smaller.

When I shoot 4x5 (Wista VX, FWIW) or 35mm (Leica MP, FWIW), the exposure is the same. If I select f/16, 1/125 on Portra 400 with my Leica or on the Wista, I get the same exposure on the film (using different focal lengths, of course... ). But when I make prints from those negatives, the grain is a LOT smaller on the 4x5, because the enlargement is a LOT less.

Yep, people (and TN), are being confused by sensor aspects changing between FX and DX but if you have identical pixels and photosites, the answer is clear.

 

[paolo-]

Actually... It turns out the light per photosite isn't what matters that much for resulting image quality. You know how the Sony A7S and A7R have the same sensor size but one has 12MP and one has 36MP? Well if you view the pixels at 1:1 yes the 36MP one is noisier, but if you look at both images at the same size, they are about the same. They're both collecting the same total amount of light, just using it differently.

On this very topic http://www.dpreview.com/articles/4613822764/high-iso-compared-sony-a7s-vs-a7r-vs-canon-eos-5d-iii Even knowing this, I'm surprised how similar the images look. I would have expected the space between the photosites to hurt the 36Mpx sensor much more. But I think that space has been reduced greatly in recent sensors.

Anyhow, I'd love to see a website with a 'nutritional fact' style sheet giving equivalence in a meaningful way. A 50mm f/1.8 lens is a 50mm f/1.8 lens no matter what format it's for but getting those equivalencies can be useful. Stating that manufacturers are misleading by providing that information is preposterous. You need to break it down to:

Equivalent field of view
Equivalent DoF
Equivalent Noise

 

[Toltepeceno]

Since the FBI won't do anything about it, I think it should be corrected.

If the FBI went after liars they would never do anything else again. A good part of people on this forum would probably tie them up for some time.

 

[Edge100]

On this very topic http://www.dpreview.com/articles/4613822764/high-iso-compared-sony-a7s-vs-a7r-vs-canon-eos-5d-iii Even knowing this, I'm surprised how similar the images look. I would have expected the space between the photosites to hurt the 36Mpx sensor much more. But I think that space has been reduced greatly in recent sensors.

Why would you expect anything different? You're taking a 36MP image and downsizing it 3:1 to 12MP. So any pixel-level noise is getting reduced 3x, at the cost of decreased potential output resolution. Since most people are only producing 1200px images for Flickr, this seems like an excellent trade off, considering that an A7R can be had for much less than an A7S.

Anyhow, I'd love to see a website with a 'nutritional fact' style sheet giving equivalence in a meaningful way. A 50mm f/1.8 lens is a 50mm f/1.8 lens no matter what format it's for but getting those equivalencies can be useful. Stating that manufacturers are misleading by providing that information is preposterous. You need to break it down to:

Equivalent field of view
Equivalent DoF
Equivalent Noise

Why?

Unless you shoot with the same lenses on multiple formats, why would you care what a 50mm lens looks like on m4/3, APS-C, FF, 6x6, 6x7, 6x9, 4x5, 8x10, 11x14, or anything else? Surely all that matters to you is how a 50mm lens looks on the camera you actually use.

It makes absolutely no sense for a newcomer to learn that the 18-55 lens she uses on her 70D is equivalent to a 29-88 on a 5D Mark III, and even less sense to learn that f/4.5 isn't the same on her camera and a 5D3, except that sometimes it is.

 

[simonsi]

I don't know how to explain this any more simply. If a MFT sensor and a FF sensor have the same total number of megapixels, the MFT's sensor sites are 1/4 the size. They are NOT THE SAME SIZE.

Correct. And irrelevant to assessing the exposure and irrelevant to how the lens projects an image. You are talking about sensor differences, lenses should not be described based on sensor differences. So a 200mm f/2.8 IS a 200mm f/2.8.

 

[Edge100]

Correct. And irrelevant to assessing the exposure and irrelevant to how the lens projects an image. You are talking about sensor differences, lenses should not be described based on sensor differences. So a 200mm f/2.8 IS a 200mm f/2.8.

I have a 135 f/5.6 lens for my 4x5 camera. What is the equivalent on a 1/1.7" sensor? I'm asking for a friend...

 

[simonsi]

You need to break it down to:

Equivalent field of view
Equivalent DoF
Equivalent Noise

You DO understand that lenses don't create noise, don't you?

----------

I have a 135 f/5.6 lens for my 4x5 camera. What is the equivalent on a 1/1.7" sensor? I'm asking for a friend...

That is not what TN was saying.

The equivalent on any camera is a 135 f/5.6 the point is if you could mount it, the same exposure settings will apply and the projected image circle will be correctly exposed on both cameras with the same settings. The field of view will be different of course due to the image capture size of the body, not due to the lens.

So the lens should, and only needs to be, described as a 135 f/5.6, it doesn't "become" anything else.

Sorry but this is <just> physics, not opinion.

 

[Edge100]

That is not what TN was saying.

The equivalent on any camera is a 135 f/5.6 the point is if you could mount it, the same exposure settings will apply and the projected image circle will be correctly exposed on both cameras with the same settings. The field of view will be different of course due to the image capture size of the body, not due to the lens.

So the lens should, and only needs to be, described as a 135 f/5.6, it doesn't "become" anything else.

Sorry but this is <just> physics, not opinion.

http://en.wikipedia.org/wiki/Sarcasm

 

[simonsi]

http://en.wikipedia.org/wiki/Sarcasm

Meaning?

We are talking light and lenses (and sensors). Each is governed by the laws of physics. TN (or anyone else) cannot currently break those, to suggest otherwise is simply misleading.

 

[Edge100]

Meaning?

We are talking light and lenses (and sensors). Each is governed by the laws of physics. TN (or anyone else) cannot currently break those, to suggest otherwise is simply misleading.

Meaning I was joking.