Need technical issue explained in lay terms

[Mike Buckley]

I am reading Todd Gustavson's Camera: A history of Photography from Dageurreotype to Digital. The author states on page 6, "Even today, the daguerreotypes remain the photographic process with the highest resolution."

As a reminder, the daguerreotype process involves exposing a light-sensitive metallic plate and then exposing the image to mercury vapors to enhance it. How does one measure or otherwise determine the resolution of an image captured on metal? What makes the resolution of an image captured on metal different from an image captured on paper, glass or any other light-sensitive material?

I ask because I've read quite a lot about the history of photography and don't remember anyone mentioning that daguerreotypes provide the highest resolution.

Please remember to explain in lay terms so dummies like me can understand everything.

[chauncey]

Mike, from dummy 1 to dummy 2...I might question the validity of that statement for the simple fact that the exposure time of that technique was so long and the printing technique was very involved.
Additionally, consider that the text is something like five years old and I would submit that printing technology has improved in the interim.

[ajohnw]

It could be that the grain size is smaller than later methods from Talbot but it's more likely to be a scale effect - think just how many "pixels" a 10x8in plate camera has. it would be a mind numbing number. This page suggests the same thing as the book but doesn't really explain why

http://www.rochester.edu/news/photos/daguerreotype.html

I'm fairly sure some of my shots would enlarge better than that one.

When an image is enlarged from as sensor it doesn't gain pixels other than by interpolation. Interpolation can't gain detail.

This all probably relates to why plate cameras still seem to be in use even as small as 5x4in - more "pixels"

John
-

[Mike Buckley]

That web page is really interesting, John. Thanks for sharing it!

from dummy 1 to dummy 2

That's fine with me, Chauncey, so long as I am allowed to retain the title of Dummy 1, as I'm sure that I'm the dummer of the two dummies being considered.

the exposure time of that technique was so long and the printing technique was very involved

I'm not sure, of course, but I don't think the resolution of the light-sensitive material has anything to do with exposure time. There is no printing per se, as the daguerreotype is a positive image that cannot be duplicated other other than by capturing an image of the finished product.

[inkista]

Googled up a WIRED article on restoring daguerrotypes, and it states that given how mercury combines with silver, the effective "pixel" size is between 150 and 800nm.

Camera sensor pixels, otoh, are typically measured in µm (1000 nm => 1 µm). For example, the D800's pixel size is 4.88µm

Add in the fact that a daguerrotype can't be enlarged , and I guess that's where the statement about the resolution is coming from.

[Mike Buckley]

Thank you, Kathy Li. Now for me to restate using different words for you or others to confirm whether I've got an accurate understanding:

Assuming the D800 sensor site is about 4880nm and that the average size of the corresponding area on the daguerreotype is about 475nm, that means that more than 10 times as many sensitive "parts" can be found in every square inch of a daguerreotype than in every square inch of a D800 sensor. Thus, the daguerreotype's resolution is more than ten times greater than a D800 sensor.

Did I get that even close to right?

By the way, the term is "daguerreotype," not "daguerrotype." It's the inventor's name, Daguerre, followed by "otype." It took me forever to get that ingrained into my pea-size brain.

[Manfred M]

Mike - The way daguerrotypes were made was the throught the interaction of halogens (elements in the periodic table that include iodine, bromine, chlorine (astatine and fluorine are also halogens, but are not used in photography). These would interact with the silver plating on the material and form silver halide crystals. These silver salts are the same ones used in film photography (including prints); silver iodide, silver bromide and silver chloride.

The production method meant that very small crystals were formed (sizes, as per Kathy's comment). Just as in digital photography, the smaller the sensor pixel, the less sensitive to light, you have the same thing with the silver halides; larger crystals are more sensitive to light than smaller ones. Think back to the film days; slower films had finer grain than the faster ones.

Add to that dageurrotypes were full size positive images, these very tiny and low sensitivity crystals would be used on relatively large plates (rather like comparing a crop frame image to one taken on a medium format camera); you would in theory have extremely high resolution images.

Here we have another digital analogy; the lenses of the day had relatively poor resolution as compared to modern lenses, so all that resolution was somewhat irrelevent. These images were lens limited, rather than photosensitive medium limited.

The other issue is that the book came out in 2009, and may have been correct at the time it was written, but with the 4.88µm D800 sensor and 3.9µm D7100 sensor, it may have been a correct statement when the book came out, but cannot be considered to be so today.

[Mike Buckley]

Mandred,

I understand everything you explained except that I don't understand the math, or at least I don't think I do. I would aprreciate it if you could run the math for me proving a daguerreotype's resolution compared to whatever sensor you decide to include in your comparison.

[Manfred M]

Mike - I was asleep at the switch when I drafted my previous reply. I did a graphic to show things and the statement in the book is even less convincing when one looks at things graphically..





I used the numbers from Kathy's post and showed (more of less) what I would expect the halide distribution to look like (likely a normal distribution). The modern sensor sits right at the very left of the image;; its resolution blows the dageurrotype right out of the water; and remember we are looking at a B&W image so in reality a single pixel is a valid comparison, rather than the usual 4 elements of a Bayer sensor.

[lukaswerth]

My two pennies:
1) daguerreotypes were/are (yes, there are still practitioners, google Jerry Spagnioli) a one-only process: you cannot copy a daguerreotype. A silver-plated copper plate is exposed in the camera, developed, and this forms a picture which looks like a mirror with a memory. Given this situation, the question of grain sizes is pretty academic, however: The public in the 1840ties was encouraged by Daguerre and others to inspect the detail of the pictures with a magnifying glass, and of course, people, only used to paintings, drawings, engravings, were mesmerized. The other photographic process, invented independently at the same time, the calotype, could be copied, but relied on paper negatives and was not capable of producing details anywhere near the daguerreotype. That was one big reason for the daguerreotype's early success.
2) However, just developing silver nitrate is capable of producing a pretty grainless image (that is, with very small grains), as is shown by the wet collodion process. This used glass panes as a substrate, and such images, though they were not at the time, can be enlarged, and the grains is far smaller than any film grain. (Of course, wet collodion is also far less sensitive than any film.)
3) Mike, I suspect Gustavson wanted to make a point relativising a common assumption that everything in photography is abut technical progress. I myself find it worthwhile to remember that a salted paper print (the modern version of a calotype) is capable of producing a tonality unsurpassed in its richness by any other process.

Lukas

[lukaswerth]

Manfred,
very interesting. You posted this while I as writing my reply. That would mean that daguerreotype grains are in fact rather large - as might be expected when using mercury vapors as a development agent.
But as I mentioned, the public at the time was still impressed by the details of the images (and just consider the quality of the lenses of that time!)

Lukas

[revi]

@Manfred:
I'm afraid you're a factor 1000 off for the daguerreotype:
The halide size distribution centers on about 500 nm, and pixel pitch is about 5 µm (=5000 nm) or 10× the halide crystal size.

But in any case, the discussion is a bit academic from my point of view:
My eyes can resolve details of about 0.1 mm (100 µm) at 30 cm distance. Any more detail in the image is nice, but not at all relevant. So that means that most of the detail in a daguerreotype is invisible to me.

Otoh, a sensor image must be enlarged, easily 10× (2.4 cm → 12cm) . So that 5µm sensor pitch becomes a 50µm pixel pitch... That doesn't leave all that much leeway.

And as a side effect: a lens that would be more than adequate for a daguerreotype, could well be insufficient for a modern sensor.

[Mike Buckley]

Such an enlightening discussion. Thank you to everyone.

Now if someone could please either confirm that my math in my third post is accurate or explain the math that corrects mine. It seems to me that the daguerreotype has about ten times the resolution of a digital sensor, depending on the sensor.

The halide size distribution centers on about 500 nm, and pixel pitch is about 5 µm (=5000 nm) or 10× the halide crystal size.

That explains why I don't understand how even the most modern digital sensor approaches much less surpasses the resolution of a daugerreotype. Hopefully someone can explain this to me.

[Manfred M]

@Manfred:
I'm afraid you're a factor 1000 off for the daguerreotype:
The halide size distribution centers on about 500 nm, and pixel pitch is about 5 µm (=5000 nm) or 10× the halide crystal size.

But in any case, the discussion is a bit academic from my point of view:
My eyes can resolve details of about 0.1 mm (100 µm) at 30 cm distance. Any more detail in the image is nice, but not at all relevant. So that means that most of the detail in a daguerreotype is invisible to me.

Otoh, a sensor image must be enlarged, easily 10× (2.4 cm → 12cm) . So that 5µm sensor pitch becomes a 50µm pixel pitch... That doesn't leave all that much leeway.

And as a side effect: a lens that would be more than adequate for a daguerreotype, could well be insufficient for a modern sensor.

My fault for trying to think at midnight with a head cold. Yes; went the wrong way in my math (screwing up μm vs nm); current sensors are around 5000 nm, so roughly an order of magnitude larger than the dageurrotypes. Current sensors are well to the right on the graph.

[steve welle]

Such an enlightening discussion. Thank you to everyone.

Now if someone could please either confirm that my math in my third post is accurate or explain the math that corrects mine. It seems to me that the daguerreotype has about ten times the resolution of a digital sensor, depending on the sensor.



That explains why I don't understand how even the most modern digital sensor approaches much less surpasses the resolution of a daugerreotype. Hopefully someone can explain this to me.

Your math is fine. The daguerreotype could give you a higher resolution picture if you could put a modern lens on the camera and all other things were equal except the photon detection method.

[Mike Buckley]

Thank you for the confirmation, Steve.

Sorry to learn, Manfred, that you still have not fully recuperated. Hang in there!

[ajohnw]

A couple of links relating to plain ordinary film which has a bigger gamut as well - at least in terms of how many colours it can display.

http://www.kenrockwell.com/tech/film-resolution.htm

http://www.kenrockwell.com/tech/mpmyth.htm

Also this one

http://www.luminous-landscape.com/es..._vs_film.shtml

One thing that fascinates me in this area is one aspect of the size of the sensor / film. Given that which ever of them can resolve so many lines per mm the larger format versions can always out resolve a smaller one as the image scale is larger but how many people argue that their full frame camera can fundamentally achieve higher resolution than cropped ones. Not that life is that simple in practice. Then some would have use believe that if the pixel count of that sensor is increased sufficiently it will match a medium format camera.

John
-

[Manfred M]

Let's try it again... This is what the chart should look like with dageurrotype showing roughly an order of magnitude better resolution versus the highest resolution sensors on modern cameras.





The median halide size, following a normal distribution should be just under 500nm (or 0.5μm) versus the highest resolution sensors found in commercially available cameras that come in at just under 4μm - 5μm; i.e. the dageurrotype shows an order of magnitude (10x) advantage in resolution.

That being said; when looking at any system; we cannot look at a single component and use it to judge end-to-end performance. If this were the case, we would still be shooting the older technologies (and in the case of dageurrotype, busily killing ourselves with mercury vapours). The lenses, cameras and support equipment altogether result in what we get in a final image. I remember examining a dageurrotype image through a magnifying glass and remember remarking about how the edge sharpness was rather soft, even under relatively low magnification.

[ajohnw]

That' correct for black and white Manfred as far as pixels are concerned. Colour needs a 4 pixel square and it's rather difficult to establish what the resolution really is. It could be argued that these 4 pixels equal 1 line in a line pair. A mere 50 lp/mm with 5um pixels.

John
-

[Mike Buckley]

Thank you, Manfred! I'm counting on you to update that graph every time each new Nikon sensor with improved resolution is released.