Lens speed and crop factor

Eric Sorensen said:

a 50 f1.4 on a 300D, you might need a 1/60 second. Everything else
equal, a 50 f1.4 on a 1DsII you might need 1/90?

Click to expand...

Not really, because the 1DsII has more pixels than the 300D. But given you had an equivalent full frame 6 MP, this would exactly be the case.

In the analog world, it would be like that:

Take an analog full frame EOS. Take ISO 100 film, it has a certain resolution (grain). in order to get the same resolution when you crop the picture by a factor of 1.6, you need a finer grain, i.e. "smaller ISO" - I can only guess that it would have to be ISO 60. And guess what - you would have to use 1/60 second for the cropped picture, compared to 1/100 second for the ISO100 full frame.

Regards,
Martin

Either you presume that the different size sensors have the same amount of pixels, i.e. the larger sensor has larger photodetectors, or I´m in bad need of some sleep

MArtin64 said:

Having understood the "amount of light issue" - does this mean that
the kit lens 18-55/3.5-5.6 for the EOS 300D produces equal pictures
with regard to noise as a 29-88/5.6-9.0 on my "virtual full frame"?

Click to expand...

No, completely wrong. It produces equal pictures to
a 1/2 area crop from the full frame camera's image.
Exactly the same exposure, exactly the same noise
(assuming equal sensor technolog).

MArtin64 said:

Even further, all those fast EF-L lenses used on 20D or 300D would
equal lenses with a 1.6 times longer focal length and a 1.6 times
smaller maximum aperture on a full frame, e.g. EF-L 24-70/2.8 would

Click to expand...

No no no no no: aperture doesn't change regardless of
what the sensor size is. An f4 EF-S lens is an f4 lens, period.
It'll deliver exactly the same exposure as an f4 full-frame lens.

MArtin64 said:

be equal to EF-L 38-112/4.5 on full frame?
How about the other parameters, like DOF?

Click to expand...

Again, nothing changes : the lens is a physical object
and these things are defined by its construction. They
don't change magically when you put a different type
of film behind it.

What you will see is that for a given FIELD OF VIEW,
and aperture, the depth of field will be deeper with
a smaller sensor. This is because you are using a shorter
focal length to achieve the same field of view. The
shorter focal length achieves a deeper depth of view
(for the same subject distance).

MArtin64 said:

In the analog world, it would be like that:

Take an analog full frame EOS. Take ISO 100 film, it has a certain
resolution (grain). in order to get the same resolution when you
crop the picture by a factor of 1.6, you need a finer grain, i.e.
"smaller ISO" - I can only guess that it would have to be ISO 60.
And guess what - you would have to use 1/60 second for the cropped
picture, compared to 1/100 second for the ISO100 full frame.

Click to expand...

Basically you've just re-stated this:

A larger sensor, with equal technology and photosite density,
has higher resolution than a smaller one.

But you've confused the whole discussion by
introducing lenses and apertures into the picture (sic).

I think you need to remember that for a large proportion
of all images, the resolution provided by today's 8M
1.6x crop bodies is more than adequate, even after cropping.

So someone looking for very high resolution certainly
would benefit today from a full frame body, but the
proportion of photographers that fall into that category
is quite small (IMHO).

dboreham said:

Member said:

I think you are wrong. This would mean I can have the same image on
a 130/2.0 using an EOS 20D as I would get with an EOS 1DMkII using
something like an imaginary 208/2.0, with Photoshop resampling it
to 8 MP? More than that, by using an even smaller sensor at a crop
factor of, e.g., 10, I could replace a 500/1.4 lens by a 50/1.4?

Click to expand...

Well, he's right (and so are all the other folks trying to set
you straight here). But what you say above is roughly correct.
This is for example why digicams with small sensors need
only very short focal length lenese (relative to 35mm lenses).
It's also why it's cheap and easy to make an f2.0 lens for those
digicams.

What you haven't taken into account above is the difference
in depth of field (different because the focal length is different).

Click to expand...

Different because the sensor size (and final image magnification) is different. Depth of field is yet another thing focal length has no direct effect on.

dboreham said:

What you will see is that for a given FIELD OF VIEW,
and aperture, the depth of field will be deeper with
a smaller sensor. This is because you are using a shorter
focal length to achieve the same field of view. The
shorter focal length achieves a deeper depth of view
(for the same subject distance).

Click to expand...

OK, let's try DOF to get me understand it.

My view at the moment is: I get the same picture (same DOF) when I use
1. an EOS 300D and a 50/1.4
2. an EOS 1DMk2 and a 50/1.4 with a 1.6 factor crop in photoshop
2. an EOS 1DMk2 and an 80/2.24

when printing the picture on the same size. Of course, resolution would be different, but I don't care.

Anything wrong?

Regards,
Martin

MArtin64 said:

Doug,

you are absolutely right - the focal length is not different
between an EF-S lens and an EF lens. However, the EFFECTIVE focal
length is different, and therefore the EFFECTIVE lens speed should
be different.

Click to expand...

The answer is quite simple, really. First you've defined something that simply does not exist. Then you base your theory on this non-existing definition. There is no such thing as an "effective focal length", not even in theory. What you refer to is the difference in angle of view for the same focal length when used on different formats. That has been around a long as photography has been around.

Let's step away from pixels and sensors to the old times of film photography. Maybe that will make you understand. I have a 45mm/4.0 for Pentax 67. With a special adapter I can use this lens on any 35mm camera. What do you think this lens will become on a 35mm camera? A 90mm or so? Wrong, it stays a 45mm/4.0, just what it was. The only difference is that "45mm" probably means "standard lens" to you because you think 35mm format. In reality, a 45mm is a wide angle lens on 6x7 cm.

--
Johan

They are used in astronomy quite a lot (manual mode), but I have yet to see one for Canon EOS.

The "S" in EF-S stands for short-back focus.

http://www.canon.com/camera-museum/tech/report/200401/200401.html

Some lens groups are moved inside the camera, passed the EF lens mount. The lens is optimized for the 22.7x15.1mm sensor, and projects an image, through the lens, that covers the sensor.

As EF lenses were designed for 36x24mm film size, when mounted on a camera with APS-C size sensor, the image projected by the EF lens onto the film plane would more than cover the sensor.

To me, optimization in EF-S means no over-coverage. The EF-S lenses are designed to project an image onto the APS-C size sensor only. My thinking is, as the image does not need to cover 36mmx24mm, moving the lens elements back and insider the mirror chamber would reduce the distance between the rear element and the sensor. Projection of the image would be smaller, and, there, no over-coverage.

You see, f/number is a ratio. That ratio is calculated by dividing Focal Length by Maximum Aperture Diameter. Focal length (distance between the optical center of the lens and the principal focus) is as stated.

The concept of "Focal Length Multiplier" is misleading. I don't know why people still use that term. "Field of View Crop" is more accurate, and, to me, causes less confusion. Lens manufacturers should explain that a lens that is originally designed to project an image on 36x24mm film dimensions has a certain field of view. Fit the same lens onto a camera will a smaller film dimensions, you would end up not using the full field of view that the lens provides. That lens gives a wider coverage than what the sensor can capture. The lens-sensor pair is not "optimized".

MArtin64 said:

The lens speed when using EF lenses on e.g. a EOS 300D is something
I haven't understood. Let's take a F50, 1.4 lens. The 1.4 is at
full frame. Lens speed is maximum aperture divided by focal length.
If I now use the 50/1.4, the maximum aperture obviously stays the
same (it's still the same lens), however the effective focal length
is 80mm, resulting in an effective lens speed of 1.6 x 1.4 = 2.2

When Canon would build an EF-S 50/1.4 - what would that 1.4 mean
then? Would it mean that effectively more light reaches the smaller
sensor (i.e. an EF-S 50/1.4 would be effectively equal to an EF
50/0.88 on an EOS 300D), or is it simply smaller, same amount of
light on the sensor, but cropped? In order to do so, however, a
smaller aperture would be sufficient, which would again mean that
the lens speed would be less than at EF 50/1.4

I am confused. Maybe someone can explain.

Regards,
Martin

Click to expand...

--
Bill
----------
Vancouver Snow Days: http://www.pbase.com/bill_ting/snowdays
People Shots: http://www.pbase.com/bill_ting/people
Miscellany: http://www.pbase.com/bill_ting/miscellany

Hi, Martin,

MArtin64 said:

Doug,

thanks again for taking your time explaining this matter in such a
competent way to a fool like me

Having understood the "amount of light issue" - does this mean that
the kit lens 18-55/3.5-5.6 for the EOS 300D produces equal pictures
with regard to noise as a 29-88/5.6-9.0 on my "virtual full frame"?

Click to expand...

I've said nothing about the noise implications of the lumnous flux per pixel issue.

MArtin64 said:

I am not impressed ;-)

Click to expand...

I didn't ask you to be impressed. I don't even ask you to believe what I say.

MArtin64 said:

Even further, all those fast EF-L lenses used on 20D or 300D would
equal lenses with a 1.6 times longer focal length and a 1.6 times
smaller maximum aperture on a full frame, e.g. EF-L 24-70/2.8 would
be equal to EF-L 38-112/4.5 on full frame?

Click to expand...

No.

Any lens used on a 20D or 300D will exhibit the same field of view that a lens of 1.6 times the lens' focal length would exhibit on a full-frame 35-mm camera.

An f/2.8 lens will have the same effect on exposure on a full-frame 35-mm camera as a lens of that aperture will have on a 20D. The fact that the pixel size is different may affect the noise performance, but not the exposure (assuming a given ISO sensitivity). This is is part because the ISO sensitivity of the sensor relates to luminous flux per unit area, not luminous flux per pixel.

Thus, if, for a particlar scene, on a 20D at ISO 100, a nice exposure would be given by 1/400 sec at f/5.6 then on a 1Ds Mk II, at ISO 100, a nice exposure would be given by 1/400 sec at f/5.6.

MArtin64 said:

How about the other parameters, like DOF?

Click to expand...

Suppose we have a 50 mm lens on a 20D and shoot a subject 100 meters away at an aperture of f/4.

On a IDs Mk II, to get the same field of view, we would need to use a lens with a focal length of 80 mm.

Having done so, to have the same depth of field as in the 20D setup, on the 1Ds II we would need to use an aperture of f/6.4.

At an aperture of f/4, a subject distance of 100 meters, and a focal length of 80 mm, the depth of field would be less on the 1Ds II than on the 20D with an aperture of f/4, a subject distance of 100 meters, and a focal length of 50 mm.

All this is based on our adoption of a circle of confusion diameter limit (COCDL) that is a consistent fraction of the frame size. If another approach for adopting a COCDL on cameras of differing format size were adopted, the result will be different.

Best regards,

Doug

no text

If this is "yet another one of these discussions" - could you please point me to either a discussion thread or a web site that totally explains the issue?

I am still convinced that using a 1.6 crop camera makes me lose a factor of 1.6 on the lens speed considering DOF and ISO. This is particularly bad for wide angle lenses, but also for tele - a 85 1.2 L, worth 2000$, becomes an equivalent of a 136/1.9 L - you can get the 135/2 L at half the price, producing the same picture on a full frame camera.

Regards,
Martin

MArtin64 said:

If this is "yet another one of these discussions" - could you
please point me to either a discussion thread or a web site that
totally explains the issue?

I am still convinced that using a 1.6 crop camera makes me lose a
factor of 1.6 on the lens speed considering DOF and ISO. This is
particularly bad for wide angle lenses, but also for tele - a 85
1.2 L, worth 2000$, becomes an equivalent of a 136/1.9 L - you can
get the 135/2 L at half the price, producing the same picture on a
full frame camera.

Regards,
Martin

Click to expand...

--
Bill
----------
Vancouver Snow Days: http://www.pbase.com/bill_ting/snowdays
People Shots: http://www.pbase.com/bill_ting/people
Miscellany: http://www.pbase.com/bill_ting/miscellany

That will be his business. I think this thread is done. No amount of persuasion would work.

So, what Martin was saying was this:
Maximum Aperture Diameter at wide end is not:
18mm / 3.5 = 5.14mm

Instead, it should be: "Effective focal length" 29mm / 5.6 = 5.17mm

Maximum Aperture Diameter at telephoto end is not:
55mm / 5.6 = 9.82mm

Instead, it should be: "Effective focal length" 88mm / 9.0 = 9.78mm

I am seeing how Martin had come up with his flawed logic.

dboreham said:

Member said:

Having understood the "amount of light issue" - does this mean that
the kit lens 18-55/3.5-5.6 for the EOS 300D produces equal pictures
with regard to noise as a 29-88/5.6-9.0 on my "virtual full frame"?

Click to expand...

No, completely wrong. It produces equal pictures to
a 1/2 area crop from the full frame camera's image.
Exactly the same exposure, exactly the same noise
(assuming equal sensor technolog).

Member said:

Even further, all those fast EF-L lenses used on 20D or 300D would
equal lenses with a 1.6 times longer focal length and a 1.6 times
smaller maximum aperture on a full frame, e.g. EF-L 24-70/2.8 would

Click to expand...

No no no no no: aperture doesn't change regardless of
what the sensor size is. An f4 EF-S lens is an f4 lens, period.
It'll deliver exactly the same exposure as an f4 full-frame lens.

Member said:

be equal to EF-L 38-112/4.5 on full frame?
How about the other parameters, like DOF?

Click to expand...

Again, nothing changes : the lens is a physical object
and these things are defined by its construction. They
don't change magically when you put a different type
of film behind it.

What you will see is that for a given FIELD OF VIEW,
and aperture, the depth of field will be deeper with
a smaller sensor. This is because you are using a shorter
focal length to achieve the same field of view. The
shorter focal length achieves a deeper depth of view
(for the same subject distance).

Click to expand...

--
Bill
----------
Vancouver Snow Days: http://www.pbase.com/bill_ting/snowdays
People Shots: http://www.pbase.com/bill_ting/people
Miscellany: http://www.pbase.com/bill_ting/miscellany

MArtin64 said:

I am still convinced that using a 1.6 crop camera makes me lose a
factor of 1.6 on the lens speed considering DOF and ISO. This is
particularly bad for wide angle lenses, but also for tele - a 85
1.2 L, worth 2000$, becomes an equivalent of a 136/1.9 L - you can
get the 135/2 L at half the price, producing the same picture on a
full frame camera.

Click to expand...

It's been pretty well explained already why you do not lose a factor of 1.6 on the speed, but let me try another tack... basically, the APS-C sensor is just a center crop of a 35mm sensor. If you take the center of a 35mm image and blow it up, that doesn't mean you've somehow retroactively lost speed on the new image.

The f number can be thought of as an amount of light hitting the sensor per unit area , not total. Taking just the center of the sensor doesn't change the amount of light per unit area.

Your 85/1.2 becomes the functional equivalent of a 136/1.2 (but the lens is still 85mm). The only difference is DOF - which is not necessarily better or worse, but depends on the shot. For most shots at or around 1.2, I wager you'll probably want more.

You do have a bee in your bonnet about this. It's been explained to you VERY clearly in this thread, by several people. If this isn't enough to penetrate your cranium, I don't hold out much hope.

Here's the "definitive" article on depth of field anyway.

[ http://www.photo.net/learn/optics/dofdigital/ ]

As to the aperture bit, your reasoning is just so plain silly and based on a fundamental misunderstanding about the definition of f-stop that there really isn't an essay that covers it. Once more:

F-stop is simply a ratio between focal length and entry pupil. The capture format has nothing to do with it. At f/1.4, the projected light intensity is always the same.

From there on out, you go off on the same tangent a thousand thousand kooks walking on the slimy sea of the Internet have gone before -- that is, coming up with abstruse calculations that end up demonstrating that "all else being equal, bigger pixels are better." Well duh.

So give it a rest already, willya?

Petteri
--
Me on photography: [ http://www.prime-junta.tk/ ]
Me on politics: [ http://p-on-p.blogspot.com/ ]

MArtin64 said:

If this is "yet another one of these discussions" - could you
please point me to either a discussion thread or a web site that
totally explains the issue?

I am still convinced that using a 1.6 crop camera makes me lose a
factor of 1.6 on the lens speed considering DOF and ISO. This is
particularly bad for wide angle lenses, but also for tele - a 85
1.2 L, worth 2000$, becomes an equivalent of a 136/1.9 L - you can
get the 135/2 L at half the price, producing the same picture on a
full frame camera.

Click to expand...

OK, let me pose a simple question here.

I use my 70-200mm f2.8 L IS for this example. Lens set to 100mm. For a given scene (a wall, evenly lit), right now here in the livimg room, I have the camera set to ISO 100, the lens is wide open at f2.8, and my D20 meters the scene and tells me I will need a shutter speed of 1/250.

By your argument it would take a shutter of say 1/175 or so to get me the same exposure using my 1.3 crop factor 1D set too ISO 100? And maybe 1/150 or even slower for my full frame film EOS-3 assuming 100 ISO film?

Then, howcome my 1D meters the scene as needing the same 1/250 shutter speed as my 20D and D60? And my EOS-3 calls for the same shutter speed on full frame film? While I can not immediately shoot the film and check it, I can and did shoot the 1D and the D20 images. They both appear correctly exposed. Both using ISO 100, f2.8 and 1/250 and the same lens set to the same focal length, 100mm.

Oh, and by the way, when I meter the scene with my hand held light meter it calls for f2.8 and 1/250 also, for an ISO of 100. My meter has no idea what I am going to use, APS-C 1.6 crop DSLR, 1.3 crop DSLR, or full frame film.

How does this fit into your thought?

T!

--

but it is primarily an astronomy term.

In astronomy, effective focal length refers to the focal length of the objective lens (if the telescope is a refractor) multiplied by any barlow factor or divided by any telereducer factor. So if you have a scope with a focal length of 1000mm and use a 2x barlow, you have a system with an EFL of 2000mm.

I imagine you could properly refer to a 500mm lens using a 2x converter as a lens with an EFL of 1000mm.

Regards, Joe Jones

Johanfoto said:

There is no such thing as an "effective
focal length", not even in theory. What you refer to is the
difference in angle of view for the same focal length when used on
different formats. That has been around a long as photography has
been around.
--
Johan

Click to expand...

--
Joe Jones
Redland, Florida

For nature and astro photography see http://www.pbase.com/jayseejay

Token said:

I use my 70-200mm f2.8 L IS for this example. Lens set to 100mm.
For a given scene (a wall, evenly lit), right now here in the
livimg room, I have the camera set to ISO 100, the lens is wide
open at f2.8, and my D20 meters the scene and tells me I will need
a shutter speed of 1/250.

By your argument it would take a shutter of say 1/175 or so to get
me the same exposure using my 1.3 crop factor 1D set too ISO 100?
And maybe 1/150 or even slower for my full frame film EOS-3
assuming 100 ISO film?

Then, howcome my 1D meters the scene as needing the same 1/250
shutter speed as my 20D and D60? And my EOS-3 calls for the same
shutter speed on full frame film? While I can not immediately
shoot the film and check it, I can and did shoot the 1D and the D20
images. They both appear correctly exposed. Both using ISO 100,
f2.8 and 1/250 and the same lens set to the same focal length,
100mm.

Oh, and by the way, when I meter the scene with my hand held light
meter it calls for f2.8 and 1/250 also, for an ISO of 100. My
meter has no idea what I am going to use, APS-C 1.6 crop DSLR, 1.3
crop DSLR, or full frame film.

How does this fit into your thought?

Click to expand...

Pretty good, to be honest. Isn't that the way ISO is defined? ISO translates into sensitivity/gain of the sensor, and that doesn't necessarily have to be the same on those cameras.

BTW, you have taken 2 different pictures - to take the same picture with the 1D you would have had to use a focal length of 123 mm. And do you get the same DOF if you use that focal length?

Regards,
Martin

Martin:

I see how you have come up with flawed logic.

You first calculated the aperture diameters for the kit lens as follows
Maximum Aperture Diameter at wide end is not:
18mm / 3.5 = 5.14mm

Maximum Aperture Diameter at telephoto end is not:
55mm / 5.6 = 9.82mm

Then you applied the 1.6x "Effective Focal Length" to calculate the "Effective f/stop", as follows:

Effective maximum f/stop at wide end: 29mm / 5.14 = 5.642
Effective maximum f/stop at telephoto end: 88mm / 9.82 = 8.961

You proceeded to round up the f/stop ratios to convince yourself that the 18-55mm kit lens has the following "Effective" designations:

EF-S 29-88mm f/5.6-9

From there, you proceeded to convince yourself that your argument on "Effective Speed" of the lens and "Effective Depth of Field" was correct.

I am so sorry for you that you are so wrong.

Just curious, have you handled the 10-22mm lens and the 16-35mm lens, and compare pictures taken from them?

Have you compared a Canon 50mm f/1.8 prime and 85mm f/1.8 prime? Did you compare the pictures taken with the shorter prime, at "effective" 80mm focal length and "effective" f/2.8 aperture value, with picture from a real 85mm f/2.8?

Just curious...

--
Bill
----------
Vancouver Snow Days: http://www.pbase.com/bill_ting/snowdays
People Shots: http://www.pbase.com/bill_ting/people
Miscellany: http://www.pbase.com/bill_ting/miscellany