AGFA Colour Photo Analysis


As I look around the internet (or, world wide web, if one prefers), I am always struck by the absolute lack of understanding in the average modern person of the nature, properties and behaviour of photographic images. There seems to be no understanding-- or even awareness-- of the differences between a chemically derived photograph (of many different types) and a digital one, nor of what processes take place when the former is transformed into the latter. Moreover, one encounters frequently this kind of quasi-religious, even one might say cult-like, dogma that colours in various images appear and behave consistently throughout the pictorial record! Were this medieval thinking not so manifestly obvious across the 'web it would not be possible to convince me that such exists; but it most surely does so.

But, how can this be!? Daily experience shows us that this cannot be true. Take an image of the same object under identical conditions with different devices (a mobile phone, digitial camera, even a chemical camera if it can be found) and compare the results. No, they don't match, do they? How can anyone believe that these images are going to show the same colours, tints and hues? I find this idea to be utterly incredible.

Film emulsions (the chemical ingredients on the film stock being used to gather light and create a negative image) and CCDs (the chip in your digital device which collects light to make a positive image) have varying sensitivities to various types, intensities and frequencies of light. It is a mistake, in fact to assume that CCDs have less such particularities-- testing over the last decade has proven that they are, if anything, more susceptible to unique behaviour than the average chemical stock. As a result, the colours present in any image will vary considerably-- often wildly-- depending upon a gigantic number of related variables. And that's just for digital images.

Travel back in time 60-70-80 years, or more, and contemplate period chemical photography. An image taken on chemical film will depend upon a) the film stock used in the camera; b) the camera itself; c) the settings employed on that camera by the operator, ergo highlighting any skill or incompetence; d) the lighting and atmospheric conditions at the time of the exposure; e) the developmental process used to create the negative image; f) the paper, glass or plastic stock onto which the negative is developed into a positive image; and so on. One could virtually make a list here without end. And in all that avalanche of variables we are meant to believe that colours and tones seen in these pictures will be uniform!? You couldn't make it up....

A small example of what I mean... All of these wartime iamges show exactly the same US Army Air Corps paints, namely QM-41 O.D. and Neutral Grey [that is to say, Spec.3-1 (1941) and ANA-603].





"But", the Great Learned of the Internet say, "they cannot be the same paint-- the colours don't look the same!" No, they don't. How on earth could they!?!? These images were all taken on different film types, by different camera equipment, by different photographers, under different conditions and were developed using completely different methods onto different stock. It is simply not possible under such a barrage of divergent factors that the colours will be identical. Completely impossible. That is even working from the assumption that none of these images have been modified when digitised, something we would not expect as they fly across the internet.

So much for uniformity! But, worse still, these images show us the second part of the problem (at least in the modern era)-- manipulation of a resulting digitised photo.



One may recognise this as a lesser cropped version of exactly the same image above. So, which appearance is correct, assuming either is so? The colour here is different to the first version, something which the internet learned tell us is "impossible". And it's the same photograph...! One or both have been deliberately manipulated to demonstrate a different tone and hue. The exif information for both JPGs reveal that they were modified in Photoshop. An exmination of the original media-- or better the negative-- by an expert observer expierienced with Kodak emulsions might well solve the problem, but that is something we cannot do here.



Here are again two versions of the very same image, again modified deliberately. Is this Liberator painted with 'yellow-ish' QM-41, or with more 'greenish' ANA-613? One must presume that either image was manipulated to become more in agreement with one or other interpretation, thus forcing the evidence to fit the preconception of "what it should be". This type of behaviour is rampant in internet modelling site discussion and activity, and it is the very bedrock of hubris. I suppose one may not say that this kind of childish and anti-scientific behaviour is the product of the modern age, but the internet has exacerbated this particularly bad situation to astronomical proportions.
I cannot understand, personally, how anyone could fail to see the monumental complexity of photographic interpretation, let alone period photographic interpretation, and the very considerable experience and expertise required to engage in this activity which must be developed over years and years of work on the topic. It is as if the average modern person believes that one may become a brain surgeon by reading an internet forum posting on the topic...!

The thought of comprehensively ignorant people arguing passionately matters which they cannot understand would make for a fine comedy sketch (or a political event), but it makes for a depressing reality. Right, enough of the Grumpy Old Man bit, let us move on.


The Starting Image (from the WWW)

We have been deluged with complaints that this author's interpretation of VVS green paints (AII Green here)-- all of which have derived from the inspection of archaeological evidence, not photographic evidence, I hasten to add-- cannot be correct, as they "don't look the same in this picture". Really? Bearing all of the foregoing discussion in mind, is anyone still confused by this fact?

I have chosen this particular image to analyse not because it was especially fine, or demonstrates professional technique, but rather because it is extremely common. Colour period images of VVS aircraft are virtually always found on AGFA film; at least so far as is known to date (we can only dream of a Kodachrome colour image of Yaks, for example). The majority of these images were shot by rank amateurs using popular cameras. In the world of photography, common usually means bad, and this photo is a classic example of that body of work.



We must first bear in mind that this is not the original image. This is a digitital facsimile of a copy (or copies) of the original slide, which is itself a developed copy of the actual negative. Thus, we are at least four steps-- each with its own problems, difficulties and modifications-- away from reality at this point, and we've not even begun our investigation. It should be clear to any observer that this image does not present what is often called a true-to-life appearance. That is to say, what the scene would look like to the human eye if present. [Absolute Colour and Perceived Colour are enormous concepts which are outside the scope of this article. We will assume here that true-to-life means a perception of the actual colours present to a human eye which functions without wavelength sensitivity/insensitivity, which is surprisingly common in fact.]

Clues supporting this observation are everywhere. Firstly, anyone can see that the colour and appearance of the sky is odd. A real sky does not look like blue-wash aluminium dope. The colour of the German uniforms (feldgrau) is well known, and what we see here does not match that appearance. Moreover, anyone familiar with the colour of grass across the huge areas of the Ukraine, Belorussia and eastern Russia will recognise that this appearance is incorrect, being too dark, which means that the colour of the green paint in view is also (the main bone of contention when discussing this photo). We have plentiful and outstanding surviving examples of other paints in view in this image, too. AII (A2) Red is well documented, and its appearance certain. The orangey appearance of these national star markings does not match it. Parts of the aircraft with different surface reflectivity, and wearing the same paint, have a distinct tonal appearance (compare the dural sheet fillets to the resin-cloth covered fuselage, for example). And so, and so... this is the correct basis from which to begin our analysis of the image, and try to understand how and why the colours in it are not true-to-life.

The image was taken using AGFA Neucolor stock. There are several versions of AGFA Neucolor, as well, including C, B2 and G types. I do not know which sub-varient was used here. In fact, I must state for the record that I am not an expert on AGFA film emulsions. Should anyone experienced with these films observe this article, their contribution to, and critique of, the findings herein would be hugely valuable.

However, as a series product of AGFA manufacture and development, we can understand that these related films share a number of properties. The company itself discusses such behaviour in extraordinary scientific detail, examples of which documentation may be found for instance in
Farbfilmtechnik: Eine Einführung für Filmschaffende (Schmidt, Richard; Kochs, Adolf 1943). Voluminous documention exists elsewhere, in addition. These AGFA papers discuss such useful data as monochrome tonality when employing this film, tonal reaction to colour filter use, as well as physical calculations and chemistry concerning light sensitivity, and so on. Some proper science, as one might say.

.....

Looking not only to this documentary material, but also examining many AGFA Neucolor images, we can recognise several features of this species of film. Firstly, there is the classic AGFA over-sensitivity to blue light. It's a tell-tale giveaway when spotting this type of film. Secondly, red colours always tend to appear orangey in tone, and less vivid than reality. Thirdly, all green shades appear darker than reality, perticularly greens with a lot of yellow in the colour. Fourthly, the surface reflectivity of an object usually changes its resulting tone in the image. An area with a matte finish and gloss finish showing the same colour will not look to be identical. These four main characteristics are common across all Neucolor types. As well, the appearance of the over-all image on this film is usually quite 'soft', one may say, with a distinctive pastel type appearance.

Here is a famous photo (on Neucolor film) from a series of German propaganda images taken in Paris directly after the occupation during the summer of 1940. The cafe La Vielleuse still exists, by the way, as does all of its period advertising material, and the vivid dark red colour of these itmes may be confirmed (as opposed to the orange appearance in the image). The washed out looking sky colour is evident, as is a pastel appearance. The professional skill used to create this image (camera settings, equipment, development) is manifestly obvious.



Here is a Neucolor shot which was allegedly taken on B2 stock somewhere in the Ukraine. The unusually dark appearance of the grass and leaves is evident, as is the odd sky colour behaviour. The image is less pastel, one may observe interestingly, but the original exposure settings look to have been quite amateurish (which we can surmise will affect the appearance).




Photoshop Auto-Colour

Digital image processing software depicts and analyses colour in a manner which seems a bit counter intuitive-- in monochrome. What!? Did we read that correctly? Yes. A digital colour is-- very crudely speaking-- the result of a pixel's intensity and super-position mapped to a colour space. Ergo, software like Photoshop have tools to analyse and compare these factors, and in such a way to estimate a "corrected" intensity (appearance). This is the Auto Colour feature, and it can be found in most such image processing software. Auto-colour tools are not magical, they do not always suggest a better appearance, and they sometimes make a hash of things. But, they do make choices and changes for a specific reason-- as defined by the coding and complex algorithms which underlie it-- and therefore they can be superb aids for photographic colour analysis.

Here is the resulting appearance after the use of Photoshop's Auto Colour feature. Some aspects of the image do look more true-to-life, but when comparing this appearance to our known colours the image is still far from convincing.



Here is the same image as processed by Micrografx's Auto-Colour feature. We see quite a bit of 'tearing' (pixelation) along the fuselage which is likely due to the very poor exposure settings used when taking this photograph. The appearance of the grass colour is notably improved, but the red colour is not, and neither the feldgrau of the soldiers' uniforms.



Neither image, of course, reveals a true-to-life appearance. What they do suggest are modifications which would be required to obtain such an appearance. By contrasting the two processed images we gain more information about the reference picture.

Photographic Filters

Software such as Photoshop have built into them some tools for applying common photographic lens filters. Back in the days of this image, coloured lens filters were ubiquitous in photography, and indeed they are common enough now. Mainly these were used in conjunction with monochrome photographs, but not exclusively. Be that as it may, one can use these self-same tools to create custom photo filters which intend to compensate (so as to move towards a true-to-life appearance) for the known behaviours of various film types. Filters to try to rectify the notorious appearance of Orthochromatic photos are very common, for example. Filters of this kind have been created for AGFA Neucolor.

Now, allow me to clarify for our modern readership that filter tools are not fool-proof. They do not hand us the solution to colour investigations on a silver plate! These tools are used to assist with a proper analysis of photography, and no more than that. As well, these AGFA filters were not developed by myself-- I have not a fraction of the expertise required with Neucolor emulsions to even attempt such a filter. I cannot, therefore, be an expert judge regarding their accuracy nor function; I will simply try to observe if a better true-to-life appearance emerges.

Here is the starting image with an AGFA Neucolor Slide Stock Filter applied. My first impression is that the resulting green colours are greatly improved. The various mid-tones do not seem to be altogether better, and the sky remains washed-out (albeit, tonally better). The AII Red colour is still orangey.



Therefore, the next step is to apply a common red light filter which shifts all of the hues in the image while attempting to match the known red shade. Here is the AGFA filter corrected image with a red light correction filter added.



The appearance looks much improved. The star marking on the fuselage looks much more like the true-to-life colour, and so does the grass. Of very notable improvement is the feldgrau colour, which now looks pretty good really. The sky is still bleached, as are the parts showing AII Blue (cowling part and ramp demarcation), as per the usual AGFA behaviour.

Now, can we improve on the appearance of this version of the image? We know that AGFA Neucolor is blue insensitive, and the poor over-exposure of the photo has exacerbated this problem. If we therefore apply a mild blue light filter to the image-- seeking to correct this unintended shift in hue-- what will get? The following image is the result of this filter application.




Does this version of the image demonstrate a more authentic, true-to-life colour appearance? It's a very fascinating question. To my own eye, I would propose that the grass' green colours are better, but those on the airframe are a bit desaturated. The German uniform colours seem to be pretty good, and where the sky actually appears (rather than a big glare), it looks to be improved. The under-surface colour is an unrecoverable wash-out. The appearance of AII Red is closer. That is of course my opinion; it impossible to prove any colour hue with such an analysis.

AII Red
AII Green (typical)
AII Blue
Feldgrau





As an extra exercise, it might be interesting to apply the same filters to an AGFA photo of excellent quality. Here we have an image taken on Neucolor slide stock of exceedingly professional competence; the exposure and focal composition are first class. One would assume that the camera in use was, as well, and that the image was deveoped in a quality lab. This is a digital facsimile, of course, so all of the previous caveats are required, but on balance this looks to be a fairly un-retouched image.



My initial impression is mainly that this is very good image! The red colour is orangey in the AGFA tradition, but much better than the previous photo. The sky looks especially authentic for this type of film, and this makes me wonder if the photographer might have used a light blue lens filter? Or considerable skill in setting the aperture? Impossible to know. The various green colours are dark, a la AGFA film, and the feldgrau area mid-tones are indistinct. The propensity for areas of different light reflection to shift tone is very much reduced-- albeit present-- and again I wonder about the very high photography skill level evident in this shot.

In the next image the
AGFA Neucolor Slide Stock Filter has been applied, precisely as above. I believe here that the grass' colour is hugely improved, as is the airframe's colour. The red areas look better, but still the feldgrau is not distinct, just as occurred on the previous image with the same filter. It seems clear that this characteristic of these AGFA films-- to shift red hues towards yellow-- is critical in the appearance of the final image. A thorough reading of AGFA's technical material will likely explain why this is so.



And finally, the image now with the red light filter applied (as above). Some strange effects can be seen in the image where the Pikpast watermark appears, so these have been highlighted for clarity. Beyond that small problem, I think this image is fairly authentic with respect to true-to-life colour appearance. At the least, it isn't bad.



Conclusions

I hope that this exercise demonstrates how and why period images cannot show us true-to-life colouration. The idea is a naive a pipe-dream: virtually impossible to do with modern images, so one can just image what the chances are on 70 year old chemical photos! No. The Internet Learned must adjust their perceptions on this topic: colours which show an irregular hue/tone appearance across historical photography are normal. That is exactly what one should expect.

This fact explains why photographic colour analysis is what it is. There is an almighty misunderstanding by would-be internet "experts" regarding the use and nature of such analytical work:

Firstly, that photo colour analysis is used to determine the appearance of an unknown colour on a period image. This is completely incorrect. Colour analysis is used to recognise a known colour or lacquer on a period image (the former being impossible). Actual colour is determined by the examination of physical evidence. Only. It may be the case, under very advantageous conditions, that an expert with very considerable experience with a given film emulsion may be able to roughly estimate a colour-- or better to say a range of plausible colours-- from photographic evidence alone. No more than that; and even then this is absolutely the best case scenario, rarely seen in historical reality.

Second, that photographic filters are used to achieve a true-to-life appearance of colours in a period image. Completely incorrect (as we have just seen). Photo filters are used to identify the film and process used to create a period image so as to assist with colour recognition. They may also be used to extract hidden detail from a given image, or to analyse other similar features (surface reflectivity, material, etc).

The misconception of these properties and uses of photographic analysis are so common that I suggest they are standard in modelling forums and internet sites. Worse still, it seems to be impossible to convince the new generation that hard work is required to achieve anything, presumably because they have never heard of such a thing and therefore it must be irrelevant. An "expert" in today's world is someone who can search on Google effectively and obtain popularity cudos ("likes"); these stupid old ideas about science, learning and experience are obviously not required....


---------------------------