Sensor Size - Dynamic Range and Depth of Field

[Detail Man]

The trade-offs between spatial-frequency resolution (referenced to the spatial dimensions of the subject existing at the location of the focus-plane, and dependent upon the fixed, or adjusted, focal length of a camera's lens-system) and depth of field are well known (and relatively in-conquerable with today's consumer-grade image-sensor technologies, cameras and lenses based around present technologies).

The newer, larger 4:3 image-sensors enchant the mind in the department of dynamic range, indeed. A fellow who likes to call himself "Detail Man" cannot help but dream ... But the resulting loss of depth of field of field can only be recovered by keeping minimum focal lengths at correspondingly small values in camera designs. I'm no expert or camera design engineer (so perhaps I can learn from your own thoughts on the inter-relationships between these parameters), but it appears (from noting what is available on the market) that certain camera-design considerations likely (also) dictate that larger sensors be integrated into camera designs having larger (and not smaller) minimum focal lengths. Thus, depth of field tends to be diminished by the increase of (either of) two other highly desirable elements - spatial-frequency resolution (detail), and dynamic range (as provided by the larger active-areas of physically larger image-sensors in general).


Sensor Size: My other personal concern about the physically larger image-sensors in general is that when they are very commonly (it seems) combined with the longer minimum focal-lengths that are chosen to accompany them in camera designs, the Hyperfocal Distance, and the Depth of Field suffer profoundly as a result. In comparison, my DMC-LX3 (at the minimum F=2.0) has a Hyperfocal Distance of only 2.54 Meters (when viewed on a computer display having 1080 vertical pixels), and the Hyperfocal Distance decreases to a mere 1.43 Meters when F=3.6. The macro DOF also tends to be much better in the case of cameras with a smaller image-sensor (without having to resort to the addition of external lenses).

Perhaps the only solution to this dilemma (of diminishing returns where depth of field is concerned when spatial-frequency resolution and dynamic range are increased) exists in large and expensive external optics - that well might transform a "compact" camera package into no more than a physical "place-holder" for a larger and heavier external add-on lens ... What do you think about that?

Is the "state of the art" of image sensing technology stuck with these strictures as a result of Nature, or are these but over-come-able technological barriers that can in the future be fully addressed in compact packages without the need for expensive, large, heavy (and thus cumbersome) external add-on lenses?

An interesting argument (that I am not of a knowledge level to either refute or endorse) that smaller "sensor-pitch" (the distance between adjacent individual photo-sensors themselves) that are fabricated into image-sensors does not (in itself, as a strict and lasting technological barrier) necessarily have to lead to decreased (effective) image-sensor signal/noise ratios is put forth by this chap named Daniel Browning, and an interesting discussion thread between (self-styled) "heavyweights" ensues thereafter (as the fur flies, and the egos burgeon in "nerd-town" somewhere undefined within cyber-space):
http://reduser.net/forum/showthread.php?t=30076


Core Elements for Success: To a large extent (as well), the particular lighting conditions of the shot (the adequacy of the illuminance of the subject, and the relative color-temperature of the light in all of the focused areas of the frame), the quality of focus, and the maintenance of camera stabilization (by some means, utilizing either optical or mechanical servo systems alone or in conjunction with "pods", etc.), together with the ability to realize a desired Depth of Field (or lack thereof) in recording an image-file, all exist as pre-requisites for any of the other technological characteristics and controllable factors to effectuate a use-able benefit to the photographer.

More photons (or more sensitivity in recording them with a better signal/noise ratio over the entire visible spectrum) can help to ensure adequate illuminance of the subject(s). Better quality optical lens-systems can help to ensure the accuracy of the delivery of those photons to the image-sensor. But the depth of field available, the quality of focus, and camera-stability are essential elements. Thus, I might prefer the wider DOF of cameras with smaller image sensors than the 4:3 products (with an emphasis on the integrity and usability of the focus, as well as the image stabilization, mechanisms). As a primarily hand-held shooter at higher F-Numbers and lower light-levels (who even often finds dealing with my mono-pod with pan/tilt restricting), effective Image Stabilization in a camera is (to me) quite important because: when recording in "raw" format - my DMC-FZ50 is limited to about ISO=200 and my DMC-LX3 is limited to about ISO=400 (divide by two in the case of JPG).

"Raw" processors can only adjust the (averaged over the entire image-frame) "color-temperature" reference - whereas the "color-contrast" existing between the various image-subjects within the field of focus is dictated by Nature and not by Humans (or their technologies). Despite our "gear", there remains much about the shot that no mechanistic hardware can adapt to and remedy.


In Search of the Miraculous: That I was able to achieve some beautiful results with a $200 USD Panasonic 6 Mpixel DMC-LZ5 manufactured in 2006 is a testament to the dominant importance of the above listed factors. Those results arose primarily out of the perseverance of taking many redundant shots of the intended subjects at different times and under different conditions, varying the shooting angles and the exposures slightly, and searching for those precious few "gems" to arise from the "rough". Interestingly, though the quality of my hardware, software and the extent of my technical understanding have increased in recent years, my technical expectations and standards also increase. The average yield of "gems" remains roughly constant (a few percent) ...

There seems (to me) to be something magical about the process of mining for "gems", and something mystical and mysterious (even to us) about what moves our heart through our minds' eyes. The simplest of images and perspectives transmitted through the humblest of dynamic ranges can move the soul - where (at the same time) all of the machinery on earth cannot (in itself) substitute for, or artificially manifest, the intangible elements that may unexpectedly constitute beauty to the soul. Perhaps (in a way) photography may be more about eliminating the grotesquely complicated distractions that exist all around us in the world and in our lives in the world - in the imagination's attempt to rediscover the beauty contained in remembrances of eternity within our "dream time".

"Man is not weak, knowledge is more than equivalent to force. The master of mechanics laughs at strength."
- Samuel Johnson

"Reason respects the differences, and imagination the similitudes of things."
- Shelley

 

[thelps]

Ahh, the philisophy of compromises. Lens/camera design to me is like trying to balance about 20 to 30 scales all at once, then put into a package the customer will buy. :? If you have ever tried to make or design something you will have some insight.

It was only a day or so ago, I read someone suggest or rather ask, "why is the digital camera sensor flat? - it doesnt need to be" The flat sensor is probably easier to make but I suspect its a throwback from the design of 35mm film which was flat to go through the gate of a movie camera. So question is, if a lens designer is not constrained to focusing on a flat CCD fields but can project onto a curved CCD or dare I say 'foveon' would this free his lens design somewhat? Or does it introduce a new whole slew of lens abberation problems?

As to sensor size I am led to think of the human eye, maybe the best compromise between resolution and size is the size of our own eyes. Funny.. I think the retina is curved isn't it?

 

[Detail Man]

Thelps,

I edited my post above in order to qualify more precisely what I meant to state about "spatial-frequency resolution" - as being referenced to the dimensions of the photographed subject (as opposed to referenced to the frame of the captured image itself). Using the phrase "spatial-frequency" is probably more clearly applied to the latter reference (of the image-frame photographed), and I probably could have stated the relationship more clearly had I chosen less technical terms in the first place. In short, longer lens-system focal lengths (such as those resulting from increasing the Zoom Factor of an adjustable telephoto lens-system, in addition to a fixed-focal length situation) reduce the Depth of Field of that lens-system. The effect is quite dramatic, and proportional to the square of the ratio of the increase in focal length.

Thus, we can better resolve details that exist at the focal-plane with longer focal lengths (in an amount directly proportional to the ratio of the increase in focal-length), but must pay a dear price for that - as the "width" (between near-focus and far-focus distances) of the "focus-field" reduces by an amount equal to the square of the amount of the ratio of increase in focal-length.

Richard Clark has an interesting web-page about human vision is at:
http://www.clarkvision.com/articles/hum ... index.html

Some (approximate) figures than emerge:

Mpixel equivalence (for a 120 degree field of view) = (approximately) 576 Mpixels;
Dynamic Range (contrast detection) > 10,000 (over 13 "stops");
ISO Sensitivity (dark-adjusted) = (approximately) 800;
Focal Length = (approximately) 22 mm;
F-Number = (approximately) 3.2;
"Shutter-Time" (in low light) = up to 15 Seconds.

I have read elsewhere that (in very low light conditions), the human eye can detect as few as a small handful of photons arriving per second. Given how much better the low-light vision is of many other animals relative to our own, it is truly remarkable how the Eye (one of the most ancient of all organs in an evolutionary sense) can detect so few photons per unit time.

The critters who live only at depths of 1,000 Feet or more in the oceans (many of which are bio-luminescent themselves) must (it would seem) be able to sense single photons (as they don't arrive very often down there). After all, they have had around 1 Billion years to practice (the estimated evolutionary time-span of one the Sea's first creatures, the anemones). Did you know that (to this day), some anemones when they die are (literally) "reborn", then becoming certain species of jellyfish? Now, that's versatility ...

 

[jpersson]

Sorry for an extremely late reply. What you are discussing is more apparent within traditional photography but the same rules applies in the digital world. Different formats produces different dynamic range, different depth of field and even different sharpness (a larger format is more forgiving to bad optics). We are basically back to thelps' compromises. To get a picture with a medium format feel in the digital world isn't cheap *sob* but it's getting there.

The eye VS photography is a different beast all together. Your eye (and brain) is a living organism, working constantly to adjust the light locally on whatever you are focusing on. You will never be able to imitate the way you see in a photograph and you will never be able to capture the dynamic range in real life. This is where you as a photographer enters the stage -- to tell your story by controlling the light.