APS-C reach? Pixel density, sensor size, shutter-speed, ISO...

bclaff said:

As for shutter speed, let me offer up two variations on an old chestnut.

The old chestnut is shutter speed = 1 / focal length

Variation 1 is adapted solely for viewing the full frame at some reasonable distance.
Variation 1 is shutter speed = (H / 24mm) * (1 / focal length)
where H is the sensor height in mm. (24mm for 35mm film and FX)

Variation 2 is adapted for pixel peeping at 1:1
Variation 2 is shutter speed = (pitch / 30microns) * ( 1 / focal length)
where pitch is in microns

As with all hand-holding rules of thumbs, Your Mileage May Vary (YMMV)
(As we all have different sized thumbs, don't we!)

Click to expand...

It seems that the greater magnifcation factor (between sensor size and displayed size) of "crop" image-sensors would flip the numerator and denominator in the factors that I have bolded above ?

.

It is asserted (here) that Ansel Adams [perhaps peering at negatives through loupes(?)] came up with a " 1 / ( 5 * Focal Length) " criteria appearing to be associated with the use of a Full Frame sized fiilm camera with a 50mm focal length lens:

A test I conducted some years ago, photographing leafless trees against a sky, indicated that, using a normal, (50mm) lens on a hand-held camera-The slowest shutter speed that ensured maximum sharpness was 1/250th sec. I found that even with firm body support image sharpness was noticeably degraded at 1/125 second, a speed that many photographers consider safe for hand-holding a camera with normal lens .

- Ansel Adams, Page 116, "The Camera", Little, Brown & Company, 1980

.

Guy Parsons wrote:

... I do remember reading some Kodak paper in the '60s or '70s where they investigated many images and came to the conclusion that the old rule might easily be multiplied by 5x to eliminate shake.

From the veritable hubbub of a thread at: http://www.dpreview.com/forums/post/40228368

Before starting that thread I did some extensive internet searching for any hint or thread of a source existing where it came to the (so-called) "rule of thumb" I found absolutely no "original source" existing whatsoever- only endless repetition. A mantra existing - without a findable original source.

Though my questioning of "sacred cows" (and the nearly one order of magntiude increase that my finalized identity indicated for the " 1 / focal length multiplier " using Micro 4/3 format) elicited precious little in the way of kind will or agreement whatsoever, this is (the corrected) version of the mathematical identity that I came up with at the time (intended for 100% crop pixel-peepers):

.

I translated Adams' chosen "margin" of 5 times the Focal Length using a 50mm lens with 35mm full-frame film (using Leica's 25 Micron COC at film-plane) to the LX3, LX5, and any M34 camera (with a 14mm lens). I used a 1.0 pixel maximum (peak, not RMS) amount of allowable resulting blur.

The "Ansel Adams Standard" form:

Minimum Shutter Speed (S) = (1) / ( (16) ( Arctangent ( ( (P) (C) ) / ( (2) (L) ) ) ) )
.

The so-called "Rule of Thumb Standard" form (5 times more lax than the "Ansel Adams Standard"):

Minimum Shutter Speed (S) = (1) / ( (80) ( Arctangent ( ( (P) (C) ) / ( (2) (L) ) ) ) )

where:

P is the Pixel Pitch [in Meters]; and

C is the Crop Factor; and

L is the (35mm equivalent) Focal Length; and

the argument of the Arctangent function is entered in (and to be interpreted by the computing device in) Radians of arc.

From: http://www.dpreview.com/forums/post/40228721

Note: Though not specifically stated in the above referenced post, Focal Length (L) is in milliMeters.

.

DM

bclaff said:

Right!

I was wondering when someone would point out that angle alone is not enough.

If you're concerned with motion then angular velocity is what you're after.

Click to expand...

Heretic!

Next, you'll be suggesting radians as opposed to clunky old minutes of arc . . ;-)

Detail Man said:

bclaff said:

As for shutter speed, let me offer up two variations on an old chestnut.

The old chestnut is shutter speed = 1 / focal length

Variation 1 is adapted solely for viewing the full frame at some reasonable distance.
Variation 1 is shutter speed = (H / 24mm) * (1 / focal length)
where H is the sensor height in mm. (24mm for 35mm film and FX)

Variation 2 is adapted for pixel peeping at 1:1
Variation 2 is shutter speed = (pitch / 30microns) * ( 1 / focal length)
where pitch is in microns

As with all hand-holding rules of thumbs, Your Mileage May Vary (YMMV)
(As we all have different sized thumbs, don't we!)

Click to expand...

It seems that the greater magnifcation factor (between sensor size and displayed size) of "crop" image-sensors would flip the numerator and denominator in the factors that I have bolded above ?

Click to expand...

No, they should not be flipped. They are correctly positioned relative to the focal length: If H is half as large, your allowable focal length at that shutter speed will also be half as long. So if focal length is in the denominator, H will need to be in the numerator.

But it can be discussed if "shutter speed" is the correct term for the result which is a duration in seconds.

Allan Olesen said:

Detail Man said:

bclaff said:

As for shutter speed, let me offer up two variations on an old chestnut.

The old chestnut is shutter speed = 1 / focal length

Variation 1 is adapted solely for viewing the full frame at some reasonable distance.
Variation 1 is shutter speed = (H / 24mm) * (1 / focal length)
where H is the sensor height in mm. (24mm for 35mm film and FX)

Variation 2 is adapted for pixel peeping at 1:1
Variation 2 is shutter speed = (pitch / 30microns) * ( 1 / focal length)
where pitch is in microns

As with all hand-holding rules of thumbs, Your Mileage May Vary (YMMV)
(As we all have different sized thumbs, don't we!)

Click to expand...

It seems that the greater magnifcation factor (between sensor size and displayed size) of "crop" image-sensors would flip the numerator and denominator in the factors that I have bolded above ?

Click to expand...

No, they should not be flipped. They are correctly positioned relative to the focal length: If H is half as large, your allowable focal length at that shutter speed will also be half as long. So if focal length is in the denominator, H will need to be in the numerator.

But it can be discussed if "shutter speed" is the correct term for the result which is a duration in seconds.

Click to expand...

As "Shutter Speed" has units of 1/Seconds (as opposed to units of Seconds), I (understandably) "keyed-in" on the arrangement of the variables within first factor of the identities accordingly.

Had I looked closer at the " 1 / focal length " factors - and noticed that they were (also) "flipped", I would have realized that the issue was (instead) one of the accurate identification of relevant units.

Allan Olesen said:

But it can be discussed if "shutter speed" is the correct term for the result which is a duration in seconds.

Click to expand...

Good point. It should read exposure time. eg. 1/300 is 1/300s

Horshack said:

D Cox said:

Hand held shots at 1:1 would normally be made using flash for live subjects, or a tripod (or copy stand) and long exposures for still subjects.

So the advantage of having small pixels holds in practice.

Click to expand...

Not for wildlife or sports, which are the scenarios often discussed for reach debates.

Click to expand...

The OP is specifically talking about macro.

xpatUSA said:

mosswings said:

The way I've been told to think about this is to relate everything to an angular density. That way you can take into account changes in resolution, sensor size, etc. fairly directly. This takes into account the crop factor magnification of the lens, but not its resolving power. It's also a relative measure, but allows you to quickly compare. <. . .>

Click to expand...

+1 one to that. I greatly prefer angular measure, a measure that is unloved by so many photographers who prefer tables, cheat sheets, on-line calculators, and parameters that are always 'dumbed down' for the 35mm film shooter (are there any left? Just kidding).

Oops, that turned into a rant, sorry. Might as well mention that I personally detest the word "reach", too :-|

Click to expand...

A lens is a device for converting angles into linear coordinates, assuming the sensor is flat, which it usually is.

Even with a spherical sensor like the retina, angular positions (bearings) are converted into coordinates. Two points that are in the same direction (such as two stars) come out at the same position on the sensor even when their XYZ rectilinear coordinates are wildly different.

So the problem is about converting angular into linear. Or seconds into pixels.

mosswings said:

Incidentally, I've long noted on this site that I found that I needed to kick up my shutterspeed by about 1/2 stop in going from a D90 to a D7100 to keep the same pixel-level sharpness. The change in angular resolution supports that finding. FX and FX cameras should see the same crop factor related behavior.

Click to expand...

Or in other words, a better sensor demands better technique. Otherwise the improvement may not be realized in practice.

bclaff said:

FWIW, when I use an APS-C (DX) for reach I typically use a VR (OS) lens which to a large extent mitigates the shutter speed issue.

Not directly to your point; but I simply find DX the sweet spot for camera and lens weight/size that I'm willing to deal with to get my reach. CX is not enough reach and FX is too heavy.

Click to expand...

I think anything with 4 micron or smaller pixels is OK, but the extra angle of view of a 24 Megapixel APS-C sensor over a 16 Mp M4/3 sensor makes aiming easier.

There are no FF sensors with pixels this small as yet. Maybe next year.

D Cox said:

mosswings said:

Incidentally, I've long noted on this site that I found that I needed to kick up my shutterspeed by about 1/2 stop in going from a D90 to a D7100 to keep the same pixel-level sharpness. The change in angular resolution supports that finding. FX and FX cameras should see the same crop factor related behavior.

Click to expand...

Or in other words, a better sensor demands better technique. Otherwise the improvement may not be realized in practice.

Click to expand...

I would say this is true if you are trying to find a minimum shutter speed (in seconds) that guarantees maximum sharpness. The question is, what is the limit in blur displacement (angular or in pixels) where it is no longer detectable (registered) by the sensor? This is the point where better technique matters.

If you are just talking about acceptable sharpness (where blur is present for both high and low density sensors) then both will render the same proportionate loss in sharpness, so better technique for the denser sensor is not necessarily required to maintain its advantage.

D Cox said:

Or in other words, a better sensor demands better technique. Otherwise the improvement may not be realized in practice.

Click to expand...

Higher pixel density does not demand better technique, or even better optics; it allows you to appreciate them more, but it benefits any optics or technique, at least to a small degree.

John Sheehy said:

D Cox said:

Or in other words, a better sensor demands better technique. Otherwise the improvement may not be realized in practice.

Click to expand...

Higher pixel density does not demand better technique, or even better optics; it allows you to appreciate them more, but it benefits any optics or technique, at least to a small degree.

--
John
http://www.pbase.com/image/55384958.jpg

Click to expand...

John, can you expound on how a higher-pixel density helps IQ for exposures where the shutter speed was too slow (blur)? I've seen you reference the benefit before but haven't seen you describe the specifics. Doesn't the higher pixel density just oversample the blur, resulting in IQ equivalent to the same exposure taken with a lower-density sensor when normalized for sensor area?