Surface & Internal Image theory

[Colin D]

Anyone else familiar with a book by W.F. Berg, Exposure Theory & Practice. I'm reading through it and have come across a section on RF where he gets into the technical aspects of its effects on film grain. Part of the discussion is about the surface image and internal image on the film grains, this is a concept I was not aware of, can someone simply explain the surface and internal images, what they are and how they work?

This discussion on RF is addressed in terms of low intensity and high intensity exposures, and how both are handled by fast and slow speed films, as well as b&w, c-41 and e-6 films. Without going deep into the specifics the chemical response of film grain to long and short exposures as described in the book, there is a reaction that creates a surface image and an internal image. If I have grasped this correctly the surface image disappears early in development and the internal image is set in the next phase of chemical reaction.

While the book is quite technical and stretches my thinking, I'm enjoying the challenge of going into uncharted and possibly superfluous areas of photographic understanding, particularly around exposure and development concepts.



Cheers

[Nodda Duma]

Internal and external image refers to the formation of the latent image speck upon exposure either internal to the silver halide grain or on the surface.


https://en.m.wikipedia.org/wiki/Latent_image

[Colin D]

Thanks ND, that makes sense of it. Now I'm interested in testing the different developing methods and conditions and their influence on the latent image being processed. Most of my work is low intensity, this helps understand the chemical reaction in the process of developing.

[Steven Ruttenberg]

This sounds interesting being I am an engineer with a physics background.

[Pere Casals]

Jason pointed here https://en.m.wikipedia.org/wiki/Latent_image information about position o the latent image inside an emulsion...

But let me speak about another thing, how a BW film may have two separate emulsion layers, this impacts in the grain structure we perceive.


A film like TMX is able to resolve 63 lines/mm in 1.6:1 contrast conditions but 200 lines/mm with 1:1000 contrast (datasheet). This is from 31.5 lp/mm to 100 lp/mm, because a line pair (lp/mm) is 2 lines.

TMX has two emulsions inside, in two separated layers. The outer layer facing the lens it's the high speed layer that have larger sensitive crystals, this layer is panchromatic, with that particular spectral response that governs TMX tonality from colors in the scene translated to gray. Of course this layer is formulated with a mix of different crystal sizes and sensitized to obtain the amazing performance that kodak (and Sexton) liked. In this emulsion we have a lot of flat chrystals, as it is of Tabular T-Grain type. IIRC (from Making Kodak Film book) we can obtain flat grains by ajusting pAg concentration during crystal growth.


But that amazing outer emulsion is not perfect, it can be burned in the highlights at some amount of exposure. So to extend the recording of detail at higher exposures, and the linearity of the sensitometric curve in the highlights they place another different emulsion under the main emulsion, the low speed emulsion.

That low speed emulsion has a quite slow ISO, so it builts additional density with increasing exposure when the (main) outer emulsion is saturated and not recording more detail in the highlights, the capability to record highlight detail it is even increased because the outer emulsion casts a shadow on the inner one, working a bit like if it was a ND filter for the inner emulsion.

That inner slow emulsion is (IIRC) orthochrmatic (red insensitive) and of classic cubic grain, with very small grains. Those small grains, when exposed, are able to produce a high resolving power, I guess that TMX produces 200 lines/mm resolving performance at 1:1000 contrast because in those conditions is when we see the inner emulsion in action, 1:1000 is 10 stops dynamic range.

The capabilities of film when recording highlight detail are amazing, it's hard to belive the amount of extreme highlight detail we can recover from it. This comes from the formulation (grain size diversity) of emulsions and from layered emulsions.

Digital devices have been trying to get those capabilities. The antique Foveon sensor had pixels of different sizes, and Arri Alexa movie cameras have DGA, dual gain amplifier technology, the same analog readings of the pixels are feed into two separated amplifiers working at different ISO...

The highlight detail is critical in Hollywood, because glares in the faces of actors are very important to depict face volumes and skin textures, you always pay too much for a charming actress if at the end you are to depict her face volumes as it was the face of a president Reagan on a dollar coin. Note that digital camera manufacturers always speak about high ISO and never about highlight detail, when lack of light is not a problem in Hollywood, they have truckloads of lightning equipment. This 2018 the challenge in a good movie still it is highlights, not the shadows.

This is a factor contributing to the present situation, film is testimonial in the industry but this 2018 still some selected movies were still shot in film, for example Mission Impossible: Fallout or StarWars VIII. Anyway digital cameras are always improving...

So this is the importance on the inner emulsion and/or grain formulations, IMHO.


Regarding aethetic grain structure (mostly important in smaller formats), each level of density in a film has a particular coarseness, some films have more grain in the mids and some more in the shadows.

TX/TXP has more grain in the shadows than in the mids, while HP5 has more grain in the mids than in the shadows. From that, depending on how we expose a film, we'll have more grain in an area or in the other.

https://c1.staticflickr.com/9/8561/28286548926_1400c4c21a_h.jpg

https://www.flickr.com/photos/125592977@N05/28286548926/

[interneg]

That inner slow emulsion is (IIRC) orthochrmatic (red insensitive) and of classic cubic grain, with very small grains. Those small grains, when exposed, are able to produce a high resolving power, I guess that TMX produces 200 lines/mm resolving performance at 1:1000 contrast because in those conditions is when we see the inner emulsion in action, 1:1000 is 10 stops dynamic range.

...

some films have more grain in the mids and some more in the shadows.

TX/TXP has more grain in the shadows than in the mids, while HP5 has more grain in the mids than in the shadows. From that, depending on how we expose a film, we'll have more grain in an area or in the other.


I think you are confusing the structure of TMX & TMY-II slightly here. And neither have Ortho sensitive only layers - it would look extremely weird at the crossover between the emulsions - I think you are confusing absorber dyes with sensitising dyes. Ilford use similar technology in the Delta films in terms of flat crystals & cubic crystals in different parts of the emulsion structures. Essentially you're using different kinds of monodisperse crystals to try & get the benefits of polydisperse crystals without the downsides. TMX may be blended emulsions rather than multilayer, though I'm not sure. It was available as a glass plate product & it's never been entirely clear to me whether there was ever much work done on multilayer coating on plates - especially given that the majority of the glass plate market in the later 20th Century was BW emulsions for specific scientific purposes. It is worth noting that Agfa's documentation shows that APX25 & APX100 had single emulsion layers & APX400 had two. Given the relatively limited ways to skin the cat that is emulsion making, it's likely that their competition were/ are not dissimilar in approach.

Furthermore, TMX is unusually accurate in terms of colour representation, such that it is recommended as a separation film.

Finally, regarding 'more grain in the shadows or mids', that has more to do with printing approaches, exposures, processing time & how expired your film is (sorry, but 12-years out of date TX is not a valid comparator) than perceived inherent aspects of the film. It isn't difficult to create prints that would fool someone's received wisdom of what TX & HP5 'are', but if you understand the actual characteristic curves & how they behave relative to process time etc, telling them apart becomes relatively easy.

[Steven Ruttenberg]

Nice discussion, but my head hurts now ��

[Pere Casals]

TMX may be blended emulsions rather than multilayer


quoting: "T-Max has a combination at the top of the emulsion are two layers of 'tabular' grains that are larger at the top record and smaller in the lower one they are uniform in size (monodisperse).
Beneath that is a layer of 3D grain (Kodak speak for cubic type) that is finer and slower probably polydisperse."

For simplicity I didn't mention that the T outer layer is in fact composed of two layers...






Furthermore, TMX is unusually accurate in terms of colour representation, such that it is recommended as a separation film.


Well, the spectral sensitivity curve does not show a flat spectral sensitivity... anyway TMX is suitable for color separation with the right illumination and calibration.

182033



Finally, regarding 'more grain in the shadows or mids', that has more to do with printing approaches, exposures, processing time & how expired your film is (sorry, but 12-years out of date TX is not a valid comparator) than perceived inherent aspects of the film. It isn't difficult to create prints that would fool someone's received wisdom of what TX & HP5 'are', but if you understand the actual characteristic curves & how they behave relative to process time etc, telling them apart becomes relatively easy.

Please read comments in the flickr link, that sample does not show actual film images but how a film grain simulation tool uses different (scientifically measured) curves for density vs grain_size. Just to show the shape of that curve.

For real samples you can see this TX image, you may note that on the car the darker areas have more grain, see similar HP5 shots and you'll see that peak graininess is more in the mids, anyway this depends on exposure.

https://c2.staticflickr.com/8/7486/28114869295_bd8576812b_k.jpg




And neither have Ortho sensitive only layers - it would look extremely weird at the crossover between the emulsions


No... because the possible ortho layer is of extreme low ISO and works only with extreme highlights, showing textures of glares and illumination sources, that are usually neutral in color. I'm not completely sure that the cubic (very low ISO) deep layer is ortho, IIRC it is. A factor to not sensitizing it panchro would be a simpler process, and not increasing emulsion speed, as that emulsion is responsible for the linearity in the extreme highligts, where TMX/Y is extremly linear compared to other films that have more or less a shoulder.

[Nodda Duma]

Back to internal vs external, I think it’s Tadaaki Tani who goes into the specific details of the differences and impact on emulsion sensitivity between the two in his book “Photographic Sensitivity: Theory and Mechanisms”.

In short, emulsions with external or just-under-the surface developing centers are practically more sensitive since the developer can get to the latent silver atoms easier (either directly at the surface, or slightly dissolving the halide crystal if just under).

[Bruce Watson]

Anyone else familiar with a book by W.F. Berg, Exposure Theory & Practice.
...
While the book is quite technical and stretches my thinking, I'm enjoying the challenge of going into uncharted and possibly superfluous areas of photographic understanding, particularly around exposure and development concepts.

If you like that one, you should read (that would take awhile as both volumes are absolutely huge) Grant Haist's Modern Photographic Processing. Detail you won't likely find anywhere else, all backed up by his career as a lead research chemist for Kodak. The man knew a bit about photo-chemistry.

Hard to find a copy to own; your best bet may be a university research library. At least the university closest to me has a copy (https://catalog.lib.ncsu.edu/?Nty=1&N=0&Ntt=grant+haist&Ntk=Author).

[Pere Casals]

emulsions with external or just-under-the surface developing centers are practically more sensitive since the developer can get to the latent silver atoms easier

Jason, I don't understand that this has much effect... developer reaches quickly the emulsion inside, modern emulsions include wetting agents to help it. Also in color films we have all 3 layers (or 4 some cases) de layers, developing at the same time...

It is about developing centers or about image centers ?

[koraks]

Pere, note the difference between emulsion and individual grains. Wetting agents won't do much in terms of penetration of grains.

[Nodda Duma]

Yep. Also, I should have added that the effective sensitivity difference becomes apparent with dissolving vs. non-dissolving developers.

The link I posted above to the wikipedia page on latent image says it better and has references that provides further detail.

[Dan Fromm]

Papi, n lines give n-1 line pairs.

||| Three lines. First pair: || Second pair: ||

[Pere Casals]

Papi, n lines give n-1 line pairs.

||| Three lines. First pair: || Second pair: ||

Dan, here it says: "be aware that numerically L/mm is double LP/mm"

https://luminous-landscape.com/mtf/

Is this wrong ?

[Nodda Duma]

Papi, n lines give n-1 line pairs.

||| Three lines. First pair: || Second pair: ||


Actually the first illustration shows 6 lines and thus 3 line pairs. Each white line and each black line counts as 1. Per convention, the pattern starts with the white line to the left and ends in a black line to the right (er...note I’m looking at this thread as white text on a black background).

Further detail: When used in optics, the value(s) or plot with units line pairs / mm is referring to spatial frequency, not an actual count of the pairs of lines in a pattern. A 3 bar pattern a half mm across isn’t 3 line pairs / mm, but is actually considered to be 6 line pairs / mm. A subtle but distinct difference. It is a count of how many pairs (consisting of a white bar and a black bar) of lines can fit in a mm, not the actual number of pairs.


To get even more complex, a repeating bar pattern actually consists of multiple spatial frequencies (accurately called cyc/mm, but often erroneously interchanged with lp/mm). .. just like the Fourier Transform of a square wave audio signal has multiple harmonic frequencies.

It’s important to note the distinction for generating contrast and modulation transfer functions, which are simply plots of the best contrast at which an imaging or lens system can image, respectively, bar patterns or spatial frequencies (sinusoidal patterns) of different widths.

[Dan Fromm]

Dan, here it says: "be aware that numerically L/mm is double LP/mm"

https://luminous-landscape.com/mtf/

Is this wrong ?

Here's the problem. Consider a single black bar: | Where's the adjacent white bar and how wide is it?

One way of measuring resolution uses target patterns of evenly spaced bars (alternately dark and light, all the same width) at a range of spacings (closer spacing, narrower bars). With the sort of test, resolution is the smallest spacing at which the bars are separated. For practical purposes, n bars/mm is usually close enough to n-1 bars/mm. But lines/mm is not twice line pairs/mm.

[Nodda Duma]

Here's the problem. Consider a single black bar: | Where's the adjacent white bar and how wide is it?

One way of measuring resolution uses target patterns of evenly spaced bars (alternately dark and light, all the same width) at a range of spacings (closer spacing, narrower bars). With the sort of test, resolution is the smallest spacing at which the bars are separated. For practical purposes, n bars/mm is usually close enough to n-1 bars/mm. But lines/mm is not twice line pairs/mm.

Sorry Dan, it's rare that you are incorrect but in this case you have it wrong. See my post above. By convention, a single black bar is considered 1/2 of a pair of lines or combined with the background to be considered one full cycle. Either way, the spatial frequency value is the same: If the single black bar is 1mm wide (1 line / mm), than the corresponding first harmonic is 0.5 cyc/mm, and the bar pattern is 0.5 line pairs / mm.

Think of it in terms of the pixel magnitudes across the pattern: A plot of a row of the values will be one cycle of a square wave.

[Pere Casals]

Well, what happens is that there is some confusion there, technically "a resolution of 10 lines per millimeter means 5 dark lines alternating with 5 light lines, or 5 line pairs per millimeter (5 LP/mm)" https://en.wikipedia.org/wiki/Image_resolution

But there is confusing literature around... depending on epoch. Some say that lpmm is line pairs per mm, some say lpmm is lines per mm because line pairs per mm is lppmm... and this is a source of confusion...

Today in photography we usually find lp/mm or cycles/mm.

I guess that kodak did not want to say 31 lp/mm as this is a lot less than what modern LF lenses can resolve, but it has to be explained that this is a great number because it is measured with 1.6:1 contrast, when lenses (I'd say) are measured with 1:100 contrast in the target, that's what a printed chart may have. So Kodak said 63 lines/mm in the datasheet.

In the ADOX CMS 20 datasheet they are proud of the 800 lp/mm the film has. Well, it's a microfilm recording medium, not a pictorial film... developed for pictorial usage...

they say:

Resolution:
Resolution in high contrast 1000 : 1 = 800 Lp/mm (line pairs/mm).
(Do not confuse with lines/mm)
http://www.adox.de/Technical_Informations/CMS20_ADOTECHII_instructions.pdf

800 it's a nicer number than 31.5 :)

[interneg]

quoting: "T-Max has a combination at the top of the emulsion are two layers of 'tabular' grains that are larger at the top record and smaller in the lower one they are uniform in size (monodisperse).
Beneath that is a layer of 3D grain (Kodak speak for cubic type) that is finer and slower probably polydisperse."

For simplicity I didn't mention that the T outer layer is in fact composed of two layers...




Yes, but that's TMY-II that you are describing there. 20 years newer in design than TMX or TMZ. The grain growth & sensitising techniques used in TMY-II are shared with cinema & still colour neg films I believe, though I also think many of those techniques significantly post-dates the arrival of TMX in the mid-late 80's. Compare the MTF curves of TMX & TMY-II between 5 & 30 cycles/mm & it's quite obvious how the use of absorber dyes & grain growth technologies etc have evolved. Furthermore, I also understand that if you can (ie you're making a sub 200 speed BW film) you don't multilayer coat the discreet emulsions for a variety of reasons, not least because of the problems of interlayer artefacts/ reflections and sharpness, instead you can blend two or more emulsions & end up with potentially better performance. Worth noting too that colour films may have three discreet layers for R, G & B, but 5 or more emulsions in those 3 layers, for a total of 15 or more in the finished product. Same principle applies to BW - would not surprise me if TMY-II had 5 emulsions in its structure.



Well, the spectral sensitivity curve does not show a flat spectral sensitivity... anyway TMX is suitable for color separation with the right illumination and calibration.


It is specifically reccomended as a separation film in several Kodak technical documents relating to Dye Transfer - and it is explicitly stated that this is because of its favourable response for accurate separation making. Indeed, TMY (not II, the first one) is specifically reccomended against for not having the right spectral response.


Please read comments in the flickr link, that sample does not show actual film images but how a film grain simulation tool uses different (scientifically measured) curves for density vs grain_size. Just to show the shape of that curve.

For real samples you can see this TX image, you may note that on the car the darker areas have more grain, see similar HP5 shots and you'll see that peak graininess is more in the mids, anyway this depends on exposure.

I don't need irrelevant 'real' samples, I've made plenty of prints from pretty much all of the current 400 speed films in the market & the immediate identifiers are spectral response & characteristic curve. Not grain. Indeed, grain 'character' is pretty easily manipulated at various stages of the process. Most of those 'film simulation' plugins seem to be astonishingly poor at actually representing the real behaviour of various films. Tri-X & HP5+ exposed in similar lighting conditions, processed to the same CI & contact printed (taking grain completely out of the way) will tell you far more about their relative characteristics & identifiers than spending time playing around with plugins.



No... because the possible ortho layer is of extreme low ISO and works only with extreme highlights, showing textures of glares and illumination sources, that are usually neutral in color. I'm not completely sure that the cubic (very low ISO) deep layer is ortho, IIRC it is. A factor to not sensitizing it panchro would be a simpler process, and not increasing emulsion speed, as that emulsion is responsible for the linearity in the extreme highligts, where TMX/Y is extremly linear compared to other films that have more or less a shoulder.
I doubt it's orthochromatic in the historic sense, but it's likely tuned to a specific sensitivity which will not adversely affect the ability to hold a range of specular lighting colours or cause strange tonal shifts at the point the emulsions cross over.

[Dan Fromm]

Jason, thanks for the correction and for encouraging me to think in terms of spatial frequency.

[Pere Casals]

No, your layer order is wrong. Emulsion at the bottom is the fastest for obvious reasons.

???

what obvious reasons ?

the emulsion at the bottom is the fastest in color films, this is because the upper layers are casting a shadow, and all layers have to deliver the same efective ISO response, so the shadows on the bottom layer are compensated with a higher sensitivity.

but in BW it's the counter, the fast layer is facing the lens to get top practical speed, while the low speed layer, at the bottom, benefits from the casted shadow to get a extremly low effective ISO to have linearity (TMX/Y/Z) in the extreme highlights.



I don't need irrelevant 'real' samples,


No problem...

[interneg]

???

what obvious reasons ?

the emulsion at the bottom is the fastest in color films, this is because the upper layers are casting a shadow, and all layers have to deliver the same efective ISO response, so the shadows on the bottom layer are compensated with a higher sensitivity.

but in BW it's the counter, the fast layer is facing the lens to get top practical speed, while the low speed layer, at the bottom, benefits from the casted shadow to get a extremly low effective ISO to have linearity (TMX/Y/Z) in the extreme highlights.



Brain not awake, yes, but very important to note that inhomogeneous colour sensitivity between emulsions would cause obvious tonal shifts at the upper end of the scale in different lighting conditions. Bit like using MG paper unfiltered as a negative.

[Pere Casals]

Brain not awake, yes, but very important to note that inhomogeneous colour sensitivity between emulsions would cause obvious tonal shifts at the upper end of the scale in different lighting conditions. Bit like using MG paper unfiltered as a negative.

interneg, this is different...

we have to consider what does the cubic bottom layer: it deals with extreme highlights. I guess that we are talking about of how film behaves at Z-VIII to Z-XI, perhaps 100 times more light than in Z-V meter recommended exposure. In those extreme highlight areas the thing it's not about tonality or spectral sensitivity, we are speaking about glares and shining lights.

[Pere Casals]

Jason, thanks for the correction and for encouraging me to think in terms of spatial frequency.

I give thanks to you because you tried to correct me what it could be a missunderstanding. This time it was not the case, but I've learned a lot from the corrections you made to me in other threads.

Also you were not completely wrong because there is literature around making equivalent pairs an single lines, at the end it's about nomenclature...

[Nodda Duma]

My pleasure, Dan. I appreciate that you are receptive to the correction, not many are willing to do so. I won’t talk about how long it took me to get that straight in my head when I started off my career.


Pere, the optics and imaging world — not the world of the internet but the people that actually have to make sure stuff works — uses the nomenclature as I describe above.

Photographers, on the other hand, I’ve found pretty much define it as whatever they damn well please. :)