Digitizing a B&W Negative While Preserving Its Subtlest Textures

Alain Oguse - Mars 2025

Technical Excerpt - Page 1
Towards a Prototype

It would be a contradiction in terms to think that
fine details in the negative would be maintained
while dust specs are be washed away.

JENS JØRGEN JENSEN - Engineer CEO – Durst US - March 2002

Table of Contents

1. A Disappointing First Attempt
2. Understanding the problematics
3. The Role of Diffuse Light
4. A Prototype and Its Optical Principle
5. Theoretical and "Pragmatic" Schematic of the Prototype
6. Necessary Equipment
7. Settings

Towards Page 2 - Results and Improvements

Foreword

This article is the result of my attempts to obtain, from non-tabular 24x36 B&W negatives https://en.wikipedia.org/wiki/Film_grain, "prints" intended for an exhibition, in A3+ format, with quality equivalent to the best that can be obtained from an end-to-end silver process: details, values, textures. The choices, preferences, and conclusions expressed here should only be considered in relation to this specific technical goal.

A Disappointing First Attempt

It was in 2015 that I first heard about Piezography https://piezography.com/. I was immediately attracted by the idea of overcoming the poor durability of OEM inks in inkjet printers. For me, it was a time when, with age, I was stepping back from my professional activities. Thanks to my new availability, this discovery rekindled my desire to return to photography, a field I had left more than 30 years ago.

Then I discovered Paul Roark's website https://www.paulroark.com/BW-Info/, which offers formulas for diluting your own carbon inks. I was enthusiastic about his "open source" approach https://en.wikipedia.org/wiki/Open_source. So, I proposed to my friend François Huchet, who was preparing an exhibition of his B&W photographs taken throughout his career, to digitize his negatives and make A3+ prints with carbon inks on cotton paper. The digitization itself—which ultimately seemed to me to be just a macro reproduction task—should not pose any problems. I was in familiar territory From 1970 to 1976, I made each year 10,000 reproductions of paintings, manuscripts, and engravings for auctioneer Maurice Rheims until he became an academician. https://en.wikipedia.org/wiki/Maurice_Rheims, or so I thought...

On the internet, one opinion prevailed: good scanners were too expensive and time-consuming to use. The often recommended alternative was a DSLR on a light table; users praised the results. Indeed, I quickly obtained encouraging images this way.

Yet, something was not right with the results.

I was disappointed by the washed-out, blurred textures that became waxy, like overly retouched skin in beauty photos. And especially that strange mottling that pretended to look like silver grain in some skies when I increased the contrast or clarity. It had little to do with the fine grain I knew. I suspected a halo effect without really understanding what caused it.

And I suddenly remembered the 1970s In October 1968, I had the chance to be accepted by Jean-Pierre Sudre to participate in the first training course he decided to organize to address the cruel shortcomings in teaching at the main photography school in Paris at the time. He became aware of this issue during the events of May and June 1968. Twelve young people thus learned the profession of photographer under the guidance of his wife Claudine and himself in their B&W laboratory, which was one of the best in France at the time. He then offered me to become his assistant to help him in his teaching mission., when we sometimes replaced the opal bulbs of Durst 138S enlargers with a slightly frosted bulb to obtain a sharper grain. Similarly, could the opalescent appearance of the light table used for digitizing negatives be reduced? I might have found the culprit: was diffuse light ruining the grain?

Understanding the Issues

In a longer forthcoming version of this article, subtitled "My Adventure with Point Light and Some Ensuing Elucubrations," I recount how, after much research, I began to understand something even though I know (almost) nothing about optics.

A Soviet Photographer Opens a Path

One day, I found the webpage of a Russian photographer: Mikhail Solunin Unfortunately, his domain name no longer belongs to him. But on the web archiving site his publication is still there. This archiving site is a marvel.. I could find it because he had the good idea to publish an English version of his point light digitization tests on a condenser enlarger⊕ Mikhail used a Krokus 4SL enlarger.. And I savored this astonishing and delightful fact that photographers from the Soviet era—who sought to overcome the lack of paper grades they needed by using tiny motorcycle bulb filaments—ended up bringing me the beginnings of a solution. Hats off!

The results he published at the time were convincing (see below), although one must remember the near-impossibility of evaluating the actual print effect of this type of image on a screen. They allow one to get an idea of the relative value effects of the treatments.

Determinant Publications for Me

The confirmation of the possible guilt of diffuse light and the advantages of point light allowed me to better orient my research and thus discover two essential but well-hidden technical sources.

Durst Resists

Indeed, well-hidden behind a strange title, SOMETHING ABOUT BILL CLINTON AND LIGHT HEADS!, this publication by Jens Jørgen Jensen, engineer CEO of Durst USA, published by Durst USA, Hillsboro, on March 29, 2002. Since the disappearance of the Durst USA website in 2011, it could still be found thanks to the web.archive.org archiving site https://web.archive.org/web/20110627091537/http://durst-pro-usa.com:80/pdf/COLIMATED%20LIGHT%20VS%20DIFFUSED%20LIGHT.pdf.

This entire 60-page text is a goldmine for anyone wishing to address these issues. I invite the reader to delve into it. Here is a short quote that sets the scene well:

“Soft-light is less of everything, less work; less spotting and less expense (compared to condenser-light). It is also less light and less sharpness, both real sharpness – detail - and perceived sharpness equal to local contrast. In some instances it is also less tonality [...]
At one point diffused light is “more” – it is more flexible and easier to use than a Condenser system.”

A Reference Academic Study

This study, Chromatic Callier Effect and its Repercussions on the Digitization of Early Film Colors, was published in the Journal of Imaging Science and Technology by Giorgio Trumpy and Barbara Flueckiger of the Department of Film Studies at the University of Zurich. It was later made available online Chromatic Callier Effect and its Repercussions on the Digitization of Early Film Colors (PDF) in 2019.

Some sentences comfort me in the idea of daring to question diffuse light, even if in this case the main question is color:

“ As long as film scanners use diffuse illumination, even the most rigorously executed color management practices [...] cannot fix the color discrepancies [...].”

A Roadmap Emerges

1. Understand the optical principles of this process
2. Build a prototype based on an enlarger
3. Find a point light source
4. Eliminate any cause of light diffusion
5. Test, correct, test, correct, test...

The prevalence of scratches and dust is the common criticism of using a (quasi) point light source. In the context of my tests, this argument no longer holds:

• Because today, graphic software allows you to erase scratches and dust more easily, effectively, and quickly than the traditional method of retouching with a brush and India ink,
• Because the methods I present here are extremely time-consuming. It is therefore logical to apply them only to images to which you attach enough importance to devote yourself to them for a long time, and preferably with pleasure. In such a context, a little more or less time to erase dust hardly matters. I have sometimes spent long hours restoring an image whose negative had been "plowed" by the advertising agency to which it had been entrusted. The final result was a real satisfaction.

The Role of Diffuse Light

To ensure I understood correctly, I needed to resort to my own drawing Each point of the negative receives light from all points of the diffuser. . Thus, I see how in point light each point of the negative receives light from a single direction, whereas in extended (or diffuse) light, each point of the negative receives light from all points of the diffuser. I realize that indeed each point of the diffuser behaves like a source of divergent rays. It is precisely this multiplicity of divergent sources that causes a penumbra around the image of an object if it is interposed. And one can guess that this phenomenon is proportional to the relative surface of the diffuser and its diffusion factor.

Moreover, something I did not expect also appears in this drawing, likewise caused by the diffuser: the surface of the cast shadow is reduced and is partly replaced by penumbra that extends amply. I found nothing on the internet that delves into the consequences of this phenomenon, except in the Durst publication.

In summary, I retained that a diffuse light source causes a double alteration of the edges by adding a penumbra zone and reducing the size of the cast shadow. This double alteration of edge precision finally explained to me how the rendering of silver grain becomes confused.

Let's recall something that often causes confusion. In B&W negative film, the silver grain is black. Consequently, in the positive on the print, the grain is white. In the highly enlarged positive image below, the dark spots are not the silver grains, which hide sneakily in what appears as a white interstitial network.

A Prototype and Its Optical Principle

To limit the risks, I looked for the same enlarger model⊕ The Krokus Color 69S as the one Mikhail used, but it is rare in France. The most common used model is the Krokus 3 Color. To be avoided absolutely, as its condensers often contain air bubbles that remain visible on the produced image in collimated light. My choice fell on another model, the Krokus Color 69S, which I was able to buy second-hand for around a hundred euros. Its design and manufacture are more recent. The quality, although still mediocre, is sufficient to make a prototype.

Thus, I was able to start drawing the optical principle ⊕ Optical Principle of a Condenser System of my prototype.

In its aforementioned publication, Durst offered a mini-guide (below) summarizing the issues of building a condenser system. The first four points now seem perfectly clear to me. However, the last two points (E and F) highlight problems that I had glimpsed but had dismissed as simple handling issues. Because they did not fit with the little drawings I had gathered on the internet. In short, I still had gaps.

The challenges involved [...] is:
A. To focus the lamp or lamp filament right in the center (Nodal point) of the enlarging lens.
B. To place the negative exactly where the light rays are produced with the highest degree of
Collimation.
C. To position the lens at a point where it renders a sharp print simultaneously with demands A and B being fulfilled.
D. To  incorporate condenser  lenses  with  a  diameter  large  enough  to  cover  the  negatives  in question  while  at  the  same  time  maintaining  the  size  of  the  system  small  enough  to  be manageable.
E. To enable different degrees of enlargement for each size of negative.
F. To create a bundle of light rays with the highest possible degree of Collimation.

Theoretical and "Pragmatic" Schematic of the Prototype

Necessary Equipment

An LED Bulb and Its Support

Initial Version: Recovering the LED from a Flashlight

I simply disassembled a small flashlight to recover the 4W-6V bulb with a 2mm diameter diode (almost point-like). Powered by a 4-battery R6 box, it provides more than enough light output for this type of use.

Fixing the bulb is easy to DIY. However, on the Krokus, the height adjustment system is too approximate for our use. I still haven't found the right solution. On the other hand, it turns out that the height adjustment does not require great precision The continuation of my tests would later change my mind about this... . It should only vary according to the negative format. The most critical is the alignment of the bulb, perfectly on the optical axis, to the nearest millimeter. Once adjusted, it should be almost fixed. It would be most convenient to have two independent adjustments, one for height and the other for alignment See at the end of this article the test giving excellent results that I carried out later..

Second Version: An LED for Color

For those who wish to test digitizing color photographs (negatives and slides), it would be advisable to test these LEDs used in museums or for photo/video shooting. There are products whose spectrum is much more homogeneous than that of basic militant LEDs. They are distinguished by their Color Rendering Index, or CRI:

“The Color Rendering Index, or CRI, indicates the ability of an artificial light source to reproduce the color nuances of a surface. Ranging from 0 to 100, it is established in relation to the colors rendered with a reference source of the same color temperature Wikipedia https://en.wikipedia.org/wiki/Color_rendering_index.”

I finally found one of these LEDs ⊕ LED Bridgelux 3000K CRI 97/98 - Actual size 3 × 2.5 mm - https://www.tme.eu/en/ on a site where the minimum quantity is not 4000 pieces but 51. Here are its main characteristics: BXEN-30S-11M-3C, SMD, 2835, CRI 97/98, 56lm, 3000K, 150mA, 3V, 3 × 2.5 mm. I quickly tested... The B&W results are similar to previous tests. However, the color results are much better. But I won't dwell on it: color is not my area of expertise.

Third Version: Would a Green LED Be More Effective?

After a good hundred "prints" in A3+ judged excellent—by my entourage or by me;-)—I still couldn't content myself with the results. I wondered if a green LED would allow us to get rid of the chromatic aberrations of my condensers.

Then I found this https://www.tme.eu/en/details/in-s63wtg/diodes-led-smd-couleur/inolux/ :

An LED smaller than a pinhead emitting in the middle of the visible light spectrum with a narrow angle... how can one resist? This LED was invented for me!

It's time to recall these two schematics from Wikipedia and Durst. The first schematic shows why monochromatic light is free from chromatic aberrations, and the second highlights how using the most central part of the condensers—which is indeed the case with my prototype in 135—reduces the consequences of spherical aberrations.

I must admit that soldering such a tiny LED is not easy. But with the help of some tutorials and by accepting to waste 1 or 2 (whose cost is negligible), one can manage it.

⊕ Sliding table and support for the LED, which will then be painted black https://fr.aliexpress.com/item/1005006085578165.htmlAs soon as this tiny green LED was installed, I was once again confronted with the difficulties of adjusting its position, as already mentioned above, due to the fact that the height and alignment on the optical axis are adjusted simultaneously from the single sliding ball joint provided by Krokus.

Looking for a way to solve this problem, I finally found this sliding table. Granted, it's "using a sledgehammer to crack a nut," but I couldn't find anything simpler in the same category. And the price is not excessive.

I added a small bracket made of white aluminum (left photo) which will be painted black and on which is precisely fixed a small PVC plate (right photo) on which the LED is glued.

Thus equipped, the adjustments become easy.

Modification of the Film Holder

It is better to do without the glass of the film holder. One less dust trap. An order on a specialized site allowed me to replace it with two laser-cut metal sheets that firmly hold the negative. To be painted black, of course! With the negative well sandwiched, the cylindrical drip tray formed by any 135 negative is minimized.

An Objective

The objective I have is the fabulous Apo Rodagon 1:4 90mm enlarging lens, which resurfaced from the time when I was doing silver prints. It's perfect for this type of work on a full-frame sensor—thus at a 1:1 ratio: little distortion and an elongated distance between elements facilitating manipulations.

The Apo is mounted in reverse. I was always taught to do this whenever the optical enlargement becomes significant, to use the lenses in angular conditions as close as possible to those for which they were designed. True or false? I don't know, but it seems plausible to me.

The only (but significant) flaw of this old Apo Rodagon is that it does not have multicoating. A new compromise that I must accept for a controlled budget. Another reason to test in green light?

Often, on an enlarging lens, the numbers indicating the diaphragm are illuminated by an internal optical device that diverts a small portion of the light from the enlarger's bulb. This device creates a light leak. In a classic reproduction bench-type configuration, to avoid an additional source of flare, care must be taken to protect the ring from external light to plug this leak.

A Macro Bellows or Extension Tube

I use a Nikon PB-5 macro bellows. Second-hand, it's affordable. The most common criticism of bellows in macro is bulk. This is not a problem for our application. And above all, this choice allows us to extend our prototype to digitizing 4.5×6, 6×6, and even 6×9 formats. You'll see: as soon as your friends learn about your new skills, drawers full of more or less old and often interesting medium-format negatives will resurface.

Various rings can be found to interface between the camera mount (camera itself or extension tube, bellows...) and your macro lens, or even to the thread diameter of your enlarging lens. Mediocre quality of this ring is sufficient given the vertical position of the lens, which does not stress the mount much.

Anti-Flare Shield

⊕ Anti-flare shield added to the condenser block My condenser block is mediocre. All the more reason not to add unnecessary diffusion. So, I masked it by adding a custom-cut black shield, taking care not to encroach on the useful light flux that must pass through the negative.

A Digital Camera and Its Support

I use a Nikon D810 equipped with a 36.3MP full-frame sensor that gives me the results I expected. I had done preliminary tests with a 12MP APS-C sensor that was not sufficient On this subject, I agree for the most part with this article: https://www.cmp-color.fr/scan2020.html. It only lacks the issue of the type of light source, which, in most of the cases mentioned, is diffuse; something the manufacturers obviously do not boast about..

A "mirrorless" camera would undoubtedly be perfectly suitable. I would also be very curious to see what a Leica Monochrom would produce.

Initial Version

⊕ The lens board of the enlarger will be lowered for shooting, thus avoiding the need to darken the room completely As shown in the adjacent photo, a stable support is needed that allows the camera-bellows-lens assembly to slide laterally and smoothly. Ours is made of a fairly heavy wooden block and a right-angle folded metal sheet. The adjustments must ensure perfect verticality and alignment on the optical axis.

Second Version

Since today's cameras are equipped with connection means (USB, HDMI, WiFi), I thought it would be practical to control the camera in "Live view" on a computer from a graphic application supporting tethered connection. Under Windows, there are many solutions for this, including open-source software: digiCamControl (https://digicamcontrol.com/). This proved to be correct for framing and enlargement ratio. However, for focusing, it does not allow precise enough adjustment with DigicamControl. Perhaps proprietary solutions are more efficient?

⊕ Second, more convenient version of the camera support, which unfortunately does not have a rotating screenSo, I tried a second approach by modifying the DSLR support to slide a mirror under the camera screen and benefit directly from the Nikon D810's Live view. Here is a video on its use https://www.youtube.com/watch?v=rojgUQgvBnw. A rotating screen, found on many recent DSLRs, would be even more convenient, as it wouldn't require the body to be raised.

I explain below that this turned out to be a significant improvement, but still insufficient...

Settings

For ease of use, I recommend proceeding in the following order.

Framing, Enlargement Ratio, and Focus

Place a negative in the film holder (24×36 – 6×6 – 6×9), then adjust the assembly so that the negative image fills almost the entire sensor and adjust the sharpness. At a ratio close to 1:1, these adjustments become delicate and seem to be reversed. Thus, it is more convenient to adjust the sharpness by varying the distance and the enlargement ratio by varying the extension. At first, it's destabilizing, but you get used to it.

Distance from the LED to the Condenser

Initial Method (Inspired by My "Finds" on the Internet)

Next, the distance from the bulb to the condenser must be adjusted. Theoretically, the height of the bulb should be adjusted to focus its sharpest possible image exactly at the level of the lens diaphragm. But one cannot stick to the ideal theoretical position, and it must be adapted to the situation according to:

• The enlargement ratio and the focal lengths of the condensers and the lens,
• The quality of the condensers and their spherical and chromatic aberrations.

Depending on the quality of the condensers, one will have to settle for the least blurry result possible. And if the installation uses a significant part of the condenser surface, it may even be necessary to shift the position of the lamp, as Durst advised, to mitigate the effects of spherical aberrations.

Here is how it is usually recommended to proceed We will see below that I discovered a more subtle solution;-)

1. Mark with precision, on the enlarger base, the position of the camera-bellows-lens assembly base. This is to be able to return to it easily. Then move it laterally to provide free access to the light cone exiting the condensers.
2. ⊕ Adjusting the height of the lamp. Its image on the cardboard is not as focused as desired :-( At the height of the diaphragm, in the light cone thus made accessible, place a white cardboard perpendicular to the optical axis to act as a screen. In the past, some lenses had a mark indicating the location of the diaphragm. This was very convenient. Otherwise, one can estimate that it is located at the level of the diaphragm ring itself. This approximation is sufficient for this non-critical measurement.
3. Adjust the height of the bulb to obtain on the cardboard an image with the highest possible degree of collimation. This will be far from resembling a sharp image (see adjacent image); this is the consequence of the compromises mentioned above (enlargement ratio, various aberrations...)
4. Carefully replace the camera-bellows-lens assembly in its marks on the enlarger base.

Optical Axis and Parallelism: And Everything Becomes Simple

If you have experience with macro reproduction work (stamps, autographs, coins), you know that perfect parallelism between the document plane and the sensor plane is essential to avoid distortions and, of course, to ensure sharpness across the entire image. To digitize a negative, it is the same, and it is particularly critical because a slight distortion that may go unnoticed at a 1:1 ratio becomes unacceptable when enlarged 10 or 12 times on a print.

It is often said that adjusting the position of the bulb in point light is difficult. This way of presenting things can be misleading! In reality, it reveals a very clumsy approach. The right method is to focus a priori only on two things: parallelism and perfect alignment on the optical axis of the entire system: bulb, condensers, negative, lens, sensor. It is strict, it is mechanical, it is simple (not always!), it is magical; all you need is a little patience, a spirit level, a plumb bob, a square, a laser level...

Initial Method

When these adjustments are correct, the flow of light on the printing surface is necessarily perfect. That's how it is. And there is an effective and convenient way to ensure proper adjustments. The two images below, taken with a smartphone slipped under the enlarger above the lens and barely touching the light beam, show the reflections of the light passing through the lens. Thanks to this, the precision of the centering is visualized. The slightest misalignment is revealed by illumination of the diaphragm edge, especially when the aperture is reduced.

In these images above Note, by the way, how collimated light highlights dust on poorly dusted lenses;-), the degree of collimation does not seem very sharp. And if it is impossible to obtain better, is it because the condensers are too mediocre?

And suddenly, I discover that a smartphone gives us a definitive way to adjust the degree of collimation (i.e., the height of the lamp) with unbeatable precision. Just close the diaphragm completely. It is then on the diaphragm itself that the image of the lamp can be projected. Impossible to be more precise.

Of course, if part of the light is blocked by the diaphragm, the light beam will be uneven.

⊕ Here, at F8 it overflows! The degree of collimation is not sufficientOnce these manipulations are completed, remember to set it back to F8 or F11 before shooting. This is an indication given by Durst, unfortunately without further explanation. I wondered about this for a long time. Because the light beam being narrower than the diaphragm, the latter can no longer have an effect on sharpness. Except that... I wonder if it doesn't ultimately have a non-negligible anti-flare role by stopping the light reflected on the multiple surfaces of the lenses.

At this stage, a pleasant surprise awaits us: if the entire installation is well adjusted, with this 4W LED bulb only, the exposure time oscillates between 1/500 and 1/1000 s at 200 ISO. This is not a reason to neglect, on an SLR, to check that the mirror will be raised well away at the shooting time to avoid vibrations.

Method 2 – After Installing the Green LED and Its New Support

When installing the green LED, improvements to the position adjustment system, together with a reduction in color aberrations, enable finer focusing. This makes it easier to distinguish the right setting.

Refining this adjustment according to the already described method, but with a now much more convenient device, an idea came to me. To verify its relevance, I slipped my smartphone above the lens again, but this time before and after removing the negative:

But of course, why didn't I think of it sooner? The negative itself causes some diffusion. This became well visible, no doubt due to the elimination of chromatic aberrations thanks to the green LED.

This last problem being resolved in a reasonably convenient way, I declare that my prototype, although still artisanal, is now officially operational;-)