Crazy disinformation about DOF, 4/3 f-stop vs 35mm, microlenses

Great Bustard · Nov 23, 2011

Kai Griffin said:
Great Bustard said:

As for "complete make-believe nonsense", I guess you missed every single link demonstrating the validity of what I was saying in the beginning of this subthread:

http://forums.dpreview.com/forums/read.asp?forum=1022&message=39908807

Click to expand...

That points to a post about pixel density (and I can't disagree with any of it), but that doesn't support the contention which set me off, which was: "When you crop the photo, you also crop the amount of light that was used to make the photo, so it becomes more noisy."

The link above is Part 2 of a two part post, and contains links to all the photos demonstrating the theory. Part 1:

http://forums.dpreview.com/forums/read.asp?forum=1022&message=39908491

discusses Total Light vs Exposure.

Kai Griffin said:
You either miswrote that... or you're crazy! How does cropping reduce the transmitted light... or make it noisier (you did not mention pixel densities in that statement - did you mean to?).

Cropping doesn't reduce transmitted light (or course) -- it reduces the amount of light the final photo is made from.

Kai Griffin said:
You leave me confused, because the two statements - particularly the first one - that I identified above make you sound like a crank... though you don't seem to be elsewhere. If you could reword it, would you?

Specifically, reword what?

Kai Griffin said:
I can't help but feel you meant to refer to pixel density/size, and not solely cropping as an influence on noise.

The total amount of light that makes up the photo is the primary source of noise in the photo (more light means less apparent noise), and cropping reduces the amount of light that the photo is made from.

Kai Griffin said:
Because so many cranks shoot off their mouths here about how sensors change physical lens properties, I thought you were another of them, and I apologize if I misidentified you and shouted at you

Read Part 1:

http://forums.dpreview.com/forums/read.asp?forum=1022&message=39908491

Let me know if you disagree or want further explanation.

Steen Bay · Nov 23, 2011

Great Bustard said:
The total amount of light that makes up the photo is the primary source of noise in the photo (more light means less apparent noise), and cropping reduces the amount of light that the photo is made from.

And cropping also reduces the MP count by the same factor as the area is reduced. Sensor area and MP count in combination determines the amount of noise in the image. Both the pixel level noise and the 'total noise power' remains the same when cropping an image.

Great Bustard · Nov 23, 2011

Steen Bay said:
Great Bustard said:

The total amount of light that makes up the photo is the primary source of noise in the photo (more light means less apparent noise), and cropping reduces the amount of light that the photo is made from.

Click to expand...

And cropping also reduces the MP count by the same factor as the area is reduced. Sensor area and MP count in combination determines the amount of noise in the image. Both the pixel level noise and the 'total noise power' remains the same when cropping an image.

...but that just ain't so. If I take a pic of a scene at, say, 50mm f/2 1/200 ISO 3200 and display it at 1200 x 900 (or printed it at 16x20 inches), it will look less noisy then if I crop out the middle 25% of the pic and display it at 1200 x 900 (or printed it at 16x20 inches), given, of course, that the average scene luminance is roughly the same for the overall pic and the crop.

Great Bustard · Nov 23, 2011

BillZM said:
Where's joemama when you need him?

'Sup?

Steen Bay · Nov 23, 2011

Great Bustard said:
Steen Bay said:

Great Bustard said:

The total amount of light that makes up the photo is the primary source of noise in the photo (more light means less apparent noise), and cropping reduces the amount of light that the photo is made from.

Click to expand...

And cropping also reduces the MP count by the same factor as the area is reduced. Sensor area and MP count in combination determines the amount of noise in the image. Both the pixel level noise and the 'total noise power' remains the same when cropping an image.

Click to expand...

...but that just ain't so. If I take a pic of a scene at, say, 50mm f/2 1/200 ISO 3200 and display it at 1200 x 900 (or printed it at 16x20 inches), it will look less noisy then if I crop out the middle 25% of the pic and display it at 1200 x 900 (or printed it at 16x20 inches), given, of course, that the average scene luminance is roughly the same for the overall pic and the crop.

Well, we've discussed this before.. and the way I see it, then a 32x40" print isn't more noisy than a 16x20" print, it's just, well, bigger (assuming that the viewing distance is close enough to see the actual amount of noise and detail in the 16x20" print). You'll probably agree that a 12mp FF image has less noise than a 12mp FT image (same exposure, efficiency, etc.), but how about a 48mp FF image vs. a 12mp FT image? I'd say the they have the same noise (well, plus/minus 4%, or so).

Great Bustard · Nov 23, 2011

Steen Bay said:
Steen Bay said:

Steen Bay said:

Steen Bay said:

The total amount of light that makes up the photo is the primary source of noise in the photo (more light means less apparent noise), and cropping reduces the amount of light that the photo is made from.

Click to expand...

And cropping also reduces the MP count by the same factor as the area is reduced. Sensor area and MP count in combination determines the amount of noise in the image. Both the pixel level noise and the 'total noise power' remains the same when cropping an image.

Click to expand...

...but that just ain't so. If I take a pic of a scene at, say, 50mm f/2 1/200 ISO 3200 and display it at 1200 x 900 (or printed it at 16x20 inches), it will look less noisy then if I crop out the middle 25% of the pic and display it at 1200 x 900 (or printed it at 16x20 inches), given, of course, that the average scene luminance is roughly the same for the overall pic and the crop.

Click to expand...

Well, we've discussed this before...

You lie.

Steen Bay said:
...and the way I see it, then a 32x40" print isn't more noisy than a 16x20" print, it's just, well, bigger (assuming that the viewing distance is close enough to see the actual amount of noise and detail in the 16x20" print). You'll probably agree that a 12mp FF image has less noise than a 12mp FT image (same exposure, efficiency, etc.), but how about a 48mp FF image vs. a 12mp FT image? I'd say the they have the same noise (well, plus/minus 4%, or so).

So, you're saying that if I post a photo downsized for web display, and a 100% crop of that photo at that same size, that they'll both have the same noise? 'Cause I have more than a few examples of that which will well support what I've said.

Steen Bay · Nov 23, 2011

Great Bustard said:
Steen Bay said:

Great Bustard said:

Great Bustard said:

Great Bustard said:

The total amount of light that makes up the photo is the primary source of noise in the photo (more light means less apparent noise), and cropping reduces the amount of light that the photo is made from.

Click to expand...

And cropping also reduces the MP count by the same factor as the area is reduced. Sensor area and MP count in combination determines the amount of noise in the image. Both the pixel level noise and the 'total noise power' remains the same when cropping an image.

Click to expand...

...but that just ain't so. If I take a pic of a scene at, say, 50mm f/2 1/200 ISO 3200 and display it at 1200 x 900 (or printed it at 16x20 inches), it will look less noisy then if I crop out the middle 25% of the pic and display it at 1200 x 900 (or printed it at 16x20 inches), given, of course, that the average scene luminance is roughly the same for the overall pic and the crop.

Click to expand...

Well, we've discussed this before...

Click to expand...

You lie.

;-)

Great Bustard said:
Great Bustard said:

...and the way I see it, then a 32x40" print isn't more noisy than a 16x20" print, it's just, well, bigger (assuming that the viewing distance is close enough to see the actual amount of noise and detail in the 16x20" print). You'll probably agree that a 12mp FF image has less noise than a 12mp FT image (same exposure, efficiency, etc.), but how about a 48mp FF image vs. a 12mp FT image? I'd say the they have the same noise (well, plus/minus 4%, or so).

Click to expand...

So, you're saying that if I post a photo downsized for web display, and a 100% crop of that photo at that same size, that they'll both have the same noise? 'Cause I have more than a few examples of that which will well support what I've said.

I said "assuming that the viewing distance is close enough..". A down sizing/sampling will of course reduce the amount of noise and detail in the image. What about my 48mp FF vs. 12mp FT example? Won't the two images have the same noise, and isn't the 12mp FT image just like a 12mp crop of the 48mp FF image?

noirdesir · Nov 23, 2011

Kai Griffin said:
but that doesn't support the contention which set me off, which was: "When you crop the photo, you also crop the amount of light that was used to make the photo, so it becomes more noisy."

You either miswrote that... or you're crazy! How does cropping reduce the transmitted light... or make it noisier

Before you accuse others of being crazy, you should take a deep breath and think what light [impacting on a surface] actually is. You can see it as a flux of energy and at the same as a stream of particles (photons) 'raining' down on a surface. Half the surface (ie, crop the sensor) and only half of the photons actually reach the part of sensor used to create the image.

Now, take a good physics book (or just Wikipedia) and look up the term shot noise. IN short, the photons in light coming from an uncorrelated light source (which includes most light sources including the sun), do not come in a perfectly even stream but in noisy stream with fluctuations. And the relative amount of these fluctuations, the variation around the mean, decrease with an increasing total number of photons. Thus more light, more photons, less variations (in relative terms), less noise in the image.

Jeff · Nov 23, 2011

(Following corrections from the Great Bustard)

I'll take my stab at a simple explanation of how to create similar images on cameras using different formats.

Suppose you take an image with a FF 35mm camera and wish to make the same image with in 4/3's format. By same image, we mean (0) same resolution, (1) same perspective and framing, (2) same motion artifacts, (3) same DOF, and (4) same image brightness. Then (assuming lenses are sharp) --

-- for the same resolution, use cameras with the same number of pixels

-- for the same perspective and framing, use a lens with half the focal length (e.g., from 50mm to 25mm).
-- for the same motion artifacts, use the same shutter speed.

-- for the same DOF, open the lens by the 2 f-stops. (e.g., from f/4 to f/2).

-- for the same image brightness, decrease the ISO by 2 f-stops (e.g., from ISO 400 to ISO 100) (to keep the same exposure)

That's it. The same prescription applied in the 1930's if you were moving from 8x10 film to 4x5 film in your view camera. This is about camera format, optics and exposure, and has nothing to do with using a digital sensor.

If your photographic situation doesn't require you to maintain the same perspective, the same shutter speed, or the same DOF, then you're working with fewer constraints and can use the extra degrees of freedom to achieve other effects.

There. Simple, and to the point. This works until diffraction effects raise their ugly head, which is largely a non-issue for current generation DSLRs.
--
Jeff

http://www.flickr.com/photos/jck_photos/sets/
http://jeffkantor.zenfolio.com/

noirdesir · Nov 23, 2011

Steen Bay said:
I said "assuming that the viewing distance is close enough..".

When you say this, you implicitly say that the viewing distance is close enough to see individual pixels (ie, you would notice the difference between four individual pixels and a merged four-pixel area). Which means you see more pixels on the larger print, or said differently you view the two prints at different resolutions (you view more pixels on the larger print).

And that is the crux of this argument, GB and a lot of other people always say that when you view two images from two sensors at the same resolution , the sensor resolution does not affect noise (if we leave actual NR out of it). But you always insist on viewing images from sensors with different resolutions at different output resolution. But that is comparing apples with oranges. It is like viewing prints from two different sensors, where you use a magnifying glass for one of the prints.

Steen Bay said:
What about my 48mp FF vs. 12mp FT example? Won't the two images have the same noise, and isn't the 12mp FT image just like a 12mp crop of the 48mp FF image?

But what is the point of comparing a print on wall with another one that is four times the area?

Great Bustard · Nov 23, 2011

Steen Bay said:
Steen Bay said:

Steen Bay said:

...and the way I see it, then a 32x40" print isn't more noisy than a 16x20" print, it's just, well, bigger (assuming that the viewing distance is close enough to see the actual amount of noise and detail in the 16x20" print). You'll probably agree that a 12mp FF image has less noise than a 12mp FT image (same exposure, efficiency, etc.), but how about a 48mp FF image vs. a 12mp FT image? I'd say the they have the same noise (well, plus/minus 4%, or so).

Click to expand...

So, you're saying that if I post a photo downsized for web display, and a 100% crop of that photo at that same size, that they'll both have the same noise? 'Cause I have more than a few examples of that which will well support what I've said.

Click to expand...

I said "assuming that the viewing distance is close enough..". A down sizing/sampling will of course reduce the amount of noise and detail in the image. What about my 48mp FF vs. 12mp FT example? Won't the two images have the same noise, and isn't the 12mp FT image just like a 12mp crop of the 48mp FF image?

OK, from the top. If you crop a photo, and display it at the same size as the original photo, the crop will be more noisy (assuming, of course, that the average luminance of the scene in the crop is about the same as the original photo).

The reality of the situation is well known (photos look more noisy a 100% than downsampled for web display), but the reason is not well known (because the crop is made from less light than the original photo).

odl · Nov 23, 2011

Great Bustard said:
Because it is all completely, and entirely, wrong.

Not really.

Great Bustard said:
If the framing is not the same, then it doesn't perform the same, does it?

Semantics, the lens doesnt change. If i point a lens up, then point it down, does its performance change? No, obviously not.

Great Bustard said:
If you step back, you change the perspective, and, once again, it does not perform the same?

No. It performs exactly the same. The image has changed, the lens hasnt.

Great Bustard said:
Or are framing and perspective unimportant photographic qualities?

They are photographic qualities, but they dont change lens performance. If i take a picture with the 50mm f2, then step back and take another picture, the lens performed exactly the same. Neither framing nor perspective are lens qualities, they are photographers tools, framing is your choice, and perspective is entirely about your relationship to the subject.

Great Bustard said:
None of the photo is actually perfectly in focus -- the focal "plane" has no depth whatsoever.

Okay okay, you know what i mean.

Great Bustard said:
However, how much of the scene appears to be in focus depends on how much the image is enlarged, and a photo on a 4/3 sensor has to be enlarged twice as much as a photo on a FF sensor for the same display size.

Okay, we know this, but...

Great Bustard said:
It is this different enlargement that results in the different DOF.

This is a terrible way of putting your point. If I use a compact, I would find it very difficult to see any subject isolation using DoF, unless i get close to them. So for the photographer using the same lens on different formats, which side by side would show exactly the same OOF areas, ends up stepping back to get the same framing. He therefore loses the proximity to the subject which gives shallower DoF... OR you could say he is using a shorter FL from the same position with the smaller aperture blah blah blah... Your way of putting it is awful, and to be honest would be of no help to anyone outside theory, as it works well for a crop of 2, but very badly for a crop of 1.3 and the plethora of compact sensor sizes.

Great Bustard said:
I'd like to know what you've seen, because that statement is demonstrably false.

Was all the noise in the D40 image photon shot? Did the amount of photon shot noise decrease dramatically? No sensor tech improved, NR improved and it continues to do so. When olympus simply reduced the strength of the AA allowing stronger NR for the same detail in High ISO settings they increased noise performance.

Plus you keep dodging the NEX7 NEX5n example, you can clearly see the better performance of one over the other. Sensor performance. Now while people say 57% efficiency is very good, we mean "for current technology".

Great Bustard said:
Hint: the 5D collects 2.5x more light for a given exposure.

Why would the newer, more expensive NEX7 perform worse than the NEX5n. Same release date, same sensor size, and same tech. This is a benchmark for me, as everythign else has far too many variables for a fair comparison.

Great Bustard said:
Demonstrations always work better for me than "gut feelings":

Dont they just. NEX7 vs NEX5n.

Great Bustard said:
Great Bustard said:

...Olympus will get there, the G3 already looks pretty good.

Click to expand...

The G3 is Panasonic, however.

I would like to see Olympus efforts on that sensor. I dont like panasonics output and i dont like PP...

Great Bustard said:
Great Bustard said:

So i enjoy my lenses and pictures. Nuff said.

Click to expand...

In the end, that's all that matters.

At least everyone can agree here. As usual I will let you have the last say here, but I know I believe my lying eyes

--
alatchinphotography.com

“Imagination is more important than knowledge. For
knowledge is limited to all we now know and
understand, while imagination embraces the entire
world, and all there ever will be to know and
understand.” - Albert Einstein

Great Bustard · Nov 23, 2011

Jeff said:
(Following corrections from the Great Bustard)

I'll take my stab at a simple explanation of how to create similar images on cameras using different formats.

Below, I've modified your text -- please let me know if you approve:

:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::

Jeff said:
Suppose you take an image with a FF 35mm camera and wish to make the same image with in 4/3's format. By same image, we mean (0) same resolution, (1) same perspective and framing, (2) same motion artifacts, (3) same DOF, and (4) same image brightness. Then if the lenses have the same relative sharpness (resolve equally well on their respective formats) --

-- for the same resolution, use cameras with the same number of pixels

-- for the same perspective and framing, take a pic from the same position using a lens with half the focal length (e.g., from 50mm to 25mm).
-- for the same motion artifacts, use the same shutter speed.

-- for the same DOF, open the lens by the 2 f-stops. (e.g., from f/4 to f/2).

-- for the same image brightness, decrease the ISO by 2 f-stops (e.g., from ISO 400 to ISO 100) (to keep the same exposure)

-- the noise will be the same if the sensors are equally efficient (same QE and read noise)

That's it. The same prescription applied in the 1930's if you were moving from 8x10 film to 4x5 film in your view camera. This is about camera format, optics and exposure, and has nothing to do with using a digital sensor.

If your photographic situation doesn't require you to maintain the same perspective, the same shutter speed, or the same DOF, then you're working with fewer constraints and can use the extra degrees of freedom to achieve other effects.

There. Simple, and to the point. In addition, the effects of diffraction softening will be the same for the same DOF.

Jeff / GB

:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::

All in all, not a lot different from:

h ttp: www.josephjamesphotography.com/equivalence/index.htm#superquick

I hope you approve!

Great Bustard · Nov 23, 2011

odl said:
No. It performs exactly the same. The image has changed, the lens hasnt.

OK, well, this is an important point -- I'm talking about the photo. As you may, or may not, have read in my entry into this thread:

http://forums.dpreview.com/forums/read.asp?forum=1022&message=39908491

OK, here's the deal -- 50mm f/2 is 50mm f/2 whether it's on FF, APS-C, or 4/3. That is, neither the focal length nor the f-ratio change as a function of the format.

However, the effect of the focal length and f-ratio do change as a function of format.

So, no, the lens does not change -- no one says it does (I mean, how could it?). But the effect of the lens most certainly does change.

odl said:
odl said:

Or are framing and perspective unimportant photographic qualities?

Click to expand...

They are photographic qualities, but they dont change lens performance. If i take a picture with the 50mm f2, then step back and take another picture, the lens performed exactly the same. Neither framing nor perspective are lens qualities, they are photographers tools, framing is your choice, and perspective is entirely about your relationship to the subject.

Well, yeah, they do -- AOV is one of the primary (if not the primary) attributes of a lens. It really doesn't make sense to compare lenses that have different AOVs for the same job, does it?

odl said:
odl said:

It is this different enlargement that results in the different DOF.

Click to expand...

This is a terrible way of putting your point.

I can put it another, more intuitive way: for the same perspective and framing, two photos displayed at the same size will have the same DOF if the aperture diameters are the same.

odl said:
odl said:

I'd like to know what you've seen, because that statement [the primary source of noise in a photo is not photon noise] is demonstrably false.

Click to expand...

Was all the noise in the D40 image photon shot? Did the amount of photon shot noise decrease dramatically? No sensor tech improved, NR improved and it continues to do so. When olympus simply reduced the strength of the AA allowing stronger NR for the same detail in High ISO settings they increased noise performance.

The D40 has ridiculously poor QE:

http://www.sensorgen.info/NikonD40.html

So, absolutely, in that particular case, the more efficient sensor in the D7000 which has double that amount will result in double the light being recorded for a given exposure (and thus only 71% the apparent photon noise), which is the primary element of the noise in the photo.

In other words, photon noise is dominant, but the QE of the sensor plays a role in the photon noise. However, the maximum possible QE is 100%, and the best modern digicams are hovering around 50%, so we're not going to keep on seeing doublings in performance.

It's not unlike increasing the number of pixels. Going from 6 MP to 12 MP has a more dramatic effect than going from 12 MP to 24 MP. Similarly, an increase in QE from 25% to 50% will be much more dramatic than going from a QE of 50% to 60%.

odl said:
Plus you keep dodging the NEX7 NEX5n example, you can clearly see the better performance of one over the other. Sensor performance. Now while people say 57% efficiency is very good, we mean "for current technology".

I don't keep dodging it -- I keep saying that I have no information on the relative efficiencies between the two sensors. And, even if the NEX7 sensor turns out to be less efficient (rather than unoptimized RAW conversion software and/or jpg engines), it's still not a given that this is due to the smaller pixels, rather than some other cause.

odl said:
Why would the newer, more expensive NEX7 perform worse than the NEX5n. Same release date, same sensor size, and same tech. This is a benchmark for me, as everythign else has far too many variables for a fair comparison.

Remember the D3x and A900? Same sensor, but Nikon's implementation:

http://www.sensorgen.info/NikonD3X.html

was rather more successful than Sony's:

http://www.sensorgen.info/SonyA900.html

So, there you are.

odl said:
odl said:

Demonstrations always work better for me than "gut feelings":

Click to expand...

Dont they just. NEX7 vs NEX5n.

odl said:

odl said:

...Olympus will get there, the G3 already looks pretty good.

Click to expand...

The G3 is Panasonic, however.

Click to expand...

I would like to see Olympus efforts on that sensor. I dont like panasonics output and i dont like PP...

Ala D3X vs A900, eh?

odl said:
odl said:

odl said:

So i enjoy my lenses and pictures. Nuff said.

Click to expand...

In the end, that's all that matters.

Click to expand...

At least everyone can agree here. As usual I will let you have the last say here, but I know I believe my lying eyes

The final photo is the final word. However, there are many variables on the way to that final photo. Shoot, we all know that just using a different RAW converter can lead to very different results, sometimes favoring one of the other for one situation, and exactly opposite for another.

But don't let the complexity of the situation get in the way of understanding what's going on and how it works.

Jeff · Nov 23, 2011

Great Bustard wrote:
..

Great Bustard said:
All in all, not a lot different from:

h ttp: www.josephjamesphotography.com/equivalence/index.htm#superquick

I hope you approve!

A couple of minor edits follow. The idea is to highlight actions with bullet points, and put the conclusions in the accompanying remarks. If this is works for you, it works for me.

Great Bustard said:
:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::

Suppose you take an image with a FF 35mm camera and wish to make the same image in the 4/3's format. By same image, we mean (0) same resolution, (1) same perspective and framing, (2) same motion artifacts, (3) same DOF, and (4) same image brightness. Then --

-- for the same resolution, use cameras with the same number of pixels and a lens with the same relative sharpness (resolve equally well in their respective formats)

-- for the same perspective and framing, take a pic from the same position using a lens with half the focal length (e.g., from 50mm to 25mm).
-- for the same motion artifacts, use the same shutter speed.
-- for the same DOF, open the lens by the 2 f-stops. (e.g., from f/4 to f/2).

-- for the same image brightness, decrease the ISO by 2 f-stops (e.g., from ISO 400 to ISO 100)

That's it. The same prescription applied in the 1930's if you were moving from 8x10 film to 4x5 film in your view camera. For a digital sensor, the noise will be the same if the sensors are equally efficient (same QE and read noise).

If your photographic situation doesn't require you to maintain the same perspective, the same shutter speed, or the same DOF, then you're working with fewer constraints and can use the extra degrees of freedom to achieve other effects.

The effects of diffraction softening will be the same for the same DOF, and generally negligible for current generation DSLR's except in extreme circumstances.

There. Simple, and to the point.

Jeff / GB

--
Jeff

http://www.flickr.com/photos/jck_photos/sets/
http://jeffkantor.zenfolio.com/

Yin Yang · Nov 23, 2011

you say:

Great Bustard said:
Absolutely correct (given the same shutter speed). Not only that, the DOF will also be the same. So, for example, a photo of the same scene from the same position at 100mm f/4 1/200 ISO 800 on FF and 50mm f/2 1/200 ISO 200 on 4/3 will have the same DOF, put the same total amount of light on the sensor, which will result in the same apparent noise for sensors that are equally efficient. >

Here is what bothers me from the begining: you use a 2 stops difference (ISO 200 and 800) to consider the pictures equivalents which seemed unnatural for me. As if FF needed less light to take the same picture as 4/3.

I think I get what I didn't understand before (at last!), correct me if I'm wrong: for a given scene I need a certain total amount (and not density) of photons to capture the scene. Am I right?

--

L'expérience est une lampe accrochée dans le dos, elle n'éclaire que le chemin parcouru. Lao Tseu

http://yinetyang.deviantart.com/

Great Bustard · Nov 23, 2011

Yin Yang said:
Yin Yang said:

Absolutely correct (given the same shutter speed). Not only that, the DOF will also be the same. So, for example, a photo of the same scene from the same position at 100mm f/4 1/200 ISO 800 on FF and 50mm f/2 1/200 ISO 200 on 4/3 will have the same DOF, put the same total amount of light on the sensor, which will result in the same apparent noise for sensors that are equally efficient. >

Click to expand...

Here is what bothers me from the begining: you use a 2 stops difference (ISO 200 and 800) to consider the pictures equivalents which seemed unnatural for me. As if FF needed less light to take the same picture as 4/3.

No -- it's the exact same amount of light. That's the important point that people are missing. The amount of light that falls on a sensor for a given scene, perspective, and framing is determined solely by the aperture diameter of the lens and the shutter speed.

So, for example, the aperture diameter at 50mm f/2 on 4/3 is 50mm / 2 = 25mm, and the aperture diameter at 100mm f/4 on FF is also 100mm / 4 = 25mm. So, combine that with the same shutter speed, and the same amount of light falls on the sensor. Thus, if the sensors are equally efficient, the noise will be the same. In addition, the DOF is the same for the same perspective, framing, and aperture diameter.

Yin Yang said:
I think I get what I didn't understand before (at last!), correct me if I'm wrong: for a given scene I need a certain total amount (and not density) of photons to capture the scene. Am I right?

Yes -- that's exactly right. If the same total light falls on a 4/3 and FF sensor, then the density of that light will be 1/4 as much on the FF sensor as the 4/3 sensor (two stops lower exposure). The role of the ISO is simply in processing -- the higher ISO setting on the FF camera simply tells the camera to make the photo appear two stops brighter.

The only effect a higher ISO has on noise is that many sensors (such as those in Olympus and Canon DSLRs) are more efficient at higher ISOs. That is, f/2 1/200 ISO 800 is less noisy than f/2 1/200 ISO 200 pushed three stops.

So, most people have got it all wrong: they think that exposure is synonomous with the total light that falls on the sensor (it's not: Total Light = Exposure x Sensor Area), they think that ISO is a factor in exposure (it's not -- ISO is a method of in-camera image processing), and they think that noise is caused by higher ISOs (it's not -- noise is caused by lower exposures -- higher ISOs, for some sensors, actually make them more efficient, and thus less noise for a given exposure).

The misunderstanding of ISO goes to the film era, where the sensitivity of the film did depend on the speed of the film. The sensitivity of a digital sensor, on the other hand, is fixed.

Hope this clears a bunch of things up for you.

Messier Object · Nov 23, 2011

Great Bustard said:
GB wrote:
Noise is an inherent property of the light itself

wow, this is something altogether new for me. I was always taught that light is electromagnetic radiation with only one inherent property, that being its wavelength.

If the noise is in the light does it somehow transfer in the sensor to generate random electrons ?

Or put another way, does the noise 'hide' in the light wave and somehow manifest itself when the wave becomes photons?

Great Bustard · Nov 23, 2011

Messier Object said:
Messier Object said:

GB wrote:
Noise is an inherent property of the light itself

Click to expand...

wow, this is something altogether new for me. I was always taught that light is electromagnetic radiation with only one inherent property, that being its wavelength.

If the noise is in the light does it somehow transfer in the sensor to generate random electrons ?

Or put another way, does the noise 'hide' in the light wave and somehow manifest itself when the wave becomes photons?

The problem is that you don't know what noise is. In the case of incoherent light, it is the random arrival of photons which is described by the Poisson Distribution.

Read all about it:

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

and here's my treatment on the subject, applied to digital photography:

h ttp: www.josephjamesphotography.com/equivalence/index.htm#noise

Messier Object · Nov 24, 2011

No, I do know what Noise is, and Noise Factor, and Noise Temperature and Signal to Noise Ratio.

The problem is that you don't seem to know what light is.

My sub-question about the noise hiding in the light was a joke.

Joking aside, your statement that noise is an inherent property of the light is just wrong
That you make such statements is a real worry.

That you refuse to admit you are wrong - well that's just the way you are I suppose, but unfortunately many readers of your work here take everything you say as true and correct.

Would you like to take the big step and take back your statement that

Great Bustard said:
Noise is an inherent property of the light itself

You can write it off as a typo if it makes it easier

Crazy disinformation about DOF, 4/3 f-stop vs 35mm, microlenses

Veteran Member

Veteran Member

Veteran Member

Veteran Member

Veteran Member

Veteran Member

Veteran Member

Veteran Member

Veteran Member

Veteran Member

Veteran Member

Veteran Member

Veteran Member

Veteran Member

Veteran Member

Leading Member

Veteran Member

Veteran Member

Veteran Member

Veteran Member

Keyboard shortcuts