Imatest – fotosaurier

Analog vs Digital – Angénieux M1 50mm f/0.95

This is one of the most iconic historical lenses (post-WWII), which I know.

I have worked about it in the last two years, because it became available to me physically through foto-friend Thomas.

The facts:

First time I mentioned the „M1“ in my basic essay on Pierre Angénieux here . To date this is only available in German – you will find all the historical facts there:

The 8-lens-Doublegauss was designed since 1953 (patent application) for cine in the different cine-formats in the focal lengths 10mm, 12,5mm, 25mm and finally 50mm and was the first mass-produced lens with aperture <1.

Thomas‘ lens is a unikat item: the historical lens-barrel with aperture ring is integrated into a focussing unit with M58 threaded interface, which allows the adaption to Sony E-mount.

**Fig. 1**: Angénieux M1 50mm f/0.95 in focusing-unit – ultra-slim E-Mount-adapter to M58 on the left – source: fotosaurier

All my findings on the optical performance of this lens measured with Sony A7R4 full-format 62 MP-Sensor you will find here . The lens covers 37mm picture-circle, which is falling a bit short for the need of full-format (43mm) – always remembering, that it is designed for cine „Super35“-format (18.7mm x 24.89mm corresponding to picture-circle 31.2mm).

**Fig. 2**: Angénieux M1 50mm f/0.95 at f/0.95 on Sony FF-Sensor of A7R4 – source: fotosaurier

In addition to the full-format measurements, you will find in the link also the results for the Super35-cine-format, taken with the crop-mode at Sony A7R4 (which has 25 MP resolution in this mode), which is the format it is designed for.

As we know in the meantime, that the resolution of historical lenses in the corners (and also the main picture-area) of the high-resolution sensors may be degraded, I looked for a way to analyse the analog picture on film with the IMATEST-software.

I succeeded with this after nearly two years work – HERE I described the way to do this: taking the pictures of the IMATEST-target on analog film (B&W Agfa APX100) – developing under strictly standardized conditions (Rodinal 1+25, 8′) and generation the digital picture with a film-scanner at 5.000 ppi (reflecta RPS 10M) using multi-scan mode and again strictly standardized sharpening-parameters with SilverFast. This results into ca. 32 MP-pictures at 24mm x 36mm (FF).

The link shows the procedure with a 28mm wideangle Olympus-OM-lens 28mm f/2.8 and unveils essential differences in picture quality in most areas open aperture (with exception of the inner center of the picture).

Now I had to solve the problem, to get the Angénieux M1 50mm-lens focused on film. There is definitely no way, to get the lens via adapter on a 24mmx36mm analog camera. Look at the rear side of the lens in the focusing-unit:

**Fig. 3**: Rear view of the Anénieux M1-50mm f/0.95-lens in the focussing-unit with rear M58mm-thread. The rear lens stands at the level of the M58-thread! – source: fotosaurier

2. Resolution-results on analog film.

What ever you would have constructed, to get this lens adapted onto an M39 or LM-camera body, it would not focus to infinity and not even to the 1,8m-distance, which would be necessary to focus on my IMATEST-target. (I did not take in consideration to use an analog-cine-camera for this purpose, which would of course be fitting to this lens, made for cine35-format … but would have required the use of extensive additional film-equipment and processing systems, which I were not acquainted with!)

I finally did it – read chapter 3 further down, if you are interested how!

Here is the lens transformed into a unikat-combination of Canon7-body and the 50mm f/0.95-lens in focussing-unit. It just allows to focus around the point of 1.84 meters distance to get the IMATEST target sharp:

**Fig. 4**: Angénieux M1 50mm f/0.95 lens adapted on a „stripped-down“ Canon7-body, allowing to focus at 1,8m at farthest … or coser – source: fotosaurier

As there are no focussing aids available (and not even an apropriate finder!) from the side of the camera, the lens has to be focussed, using a ground glass in the waist level finder placed onto the film-plane (with open back!) before the film is loaded. (The good-old Exacta WLF worked excellent for this purpose!)

**Fig. 5**: Focussing the lens on the matte-screen of WLF (Exaktawaist level finder) at open camera befor loading the film – source: fotosaurier

Also the exact framing and adjustment of the film-plane parallel to the target had to be done through the waist-level finder at the open back of the camera.

The adjustment of the film-plane parallel to the camera is one of the most important aspects for precise measurements – especially for the fastest lenses. The depth of focus is extremely narrow at f/0,95 so that sharpness may be given in the focussing location but the film-plane runs out of focus, if it is not precisely parallel to the target.

Using a digicam it is easy to adjust the camera: you take a photo of the target, take this picture out of the camera and analyze it in IMATEST. The software measures the angle between horizontal and vertical lines, which are per definition parallel in the target. That means, the angle should be ZERO, if the camera is perfectly alligned, i.e. the lines are also perfectly PARALLEL on the film-plane.

For a good allignment of the film-plane, I use 0.15° as a limit for the angles between paralell lines in the pictures on sensor or film. In this measurement-session with the Angenieux M1 50mm f/0.95 the values, which I achieved were

0.148° for horizontal lines – 0.004° (= perfect!) for vertical lines.

That used to be sufficient normally. I have, however, to date little experience with apertures of f/0.95 … The horizontal mismatch is within my tolerance, but we will see in the detailed analysis, that there is a important effect in resolution over the picture.

After loading the film, you have to handle the whole set-up like a raw egg, always being aware, that the final digitally converted picture will resolve at 5,08 microns (pixel-size) … and there is a lot of manipulation necessary before each shot:

setting the aperture,
advancing the film,
pre-tensioning the self-timer
pushing the release knob

… and praying, not to stumble over the tripod-legs all the time!

This is the full-frame picture on film:

**Fig. 6**: Full-Frame-Picture Angénieux M1 50mm f/0.95 on Film at f/0.95. Compare with Fig. 2 (on sensor) ! – source: fotosaurier

Even on this small picture at f/0.95, you see already, that the details very close to the edges and corner look sharper than on the Sony-sensor-picture (Fig. 2). Only the very central details show higher contrast on the sensor.

As the lens is designed for the cine35-format it doesn’t make sense to analyze the FullFrame film negative. So I cropped the frame to the width of the Super35 film frame-size in my scanner-preview.

In the following picture you see this situation of the scanned part of the picture. The black rectangle inside the area, which is scanned, shows the size of the super35-frame (24.89mm x 18.70mm) which corresponds to a picture-circle of 31.1mm.

**Fig. 7**: Scanned section of the B&B-negative – the black rectangle shows the size of the super35-film-frame with picture-circle 31.1mm. The whole scanned frame, which is analysed has 34.2mm picture-circle – source: fotosasurier

To enable the automatic resolution-analysis by IMATEST, it is necessary to have the black bars top and bottom of the target. Due to that, the picture area for resolution measurement has a picture circle diameter of 34.2 mm, exceeding the super35-frame by 3.1mm (corner-to-corner).

The following picture shows the set-up for the resolution analysis in IMATESTS. The magenta coloured rectangles mark the contrast-edges at which the resolution is measured – in 44 places all over the picture.

**Fig. 8**: IMATEST analysis set-up for resolution at 44 locations (calles „ROI“), four of which are located in the center (marked by the inner circle), four are located in the far corner areas at ca. 85% of the picture circle. Most of the measurements (36) are belonging to the largest picture-area called here „part way“ – source: fotosaurier

In the next picture you see the individual resolution-values in all these ROIs at f/0.95:

Fig. 9: IMATESTs 44 individual resolution-values (MTF30) within the 34,2mm-picture-circle – source: fotosaurier

The MTF30-mean-values are 1,237 LP/PH for center, for „part way“ (the majority of the picture-area!) mean MTF30 is 560 LP/PH, in the corners it is 378 LP/PH.

In this picture we clearly see, that the MTF30-values at the left side are definitely lower than on the right side edge. Obviously the effect of the small horizontal tilt of the film plane. However, we have to live with that for now, as this was the best allignment, which I achieved during this occasion to measure this lens.

However, the effects on the average resolution readings of this miss-alignment are smaller than it seems on first sight. Due to some field-curvature of the picture plane (which is not an exact plane) the right-edge values are increase, because they moved closer to the best focus, wheras the left edge moved away. I we had a perfect alignment, the right-edge-values would go down and the left-edge-values would increase: as a result of this, the average-values shown in Fig. 11 are not so far from the truth.

The best illustration of the resolution-distribution over the picture is the graph, which IMATEST calls „radial MTF-plot“: it shows the MTF30-value over the distance from the center (in percent!):

**Fig. 10**: The radial MTF-plot for Angénieux M1 50mm f/0.95 at oper aperture f/0.95 on film Agfa APX100 – MTF30 in Linepairs per Picture Height – 100% distance to the center correspondents to 34.2mm picture circle – source: fotosaurier

The software analyses the vertical and the horizontal edges in the target separately. Horizontal values are „mostly“ sagittal and the vertical ones „mostly“ meridional. For both we have very large variations in this graph. This is partly due to the horizontal missmatch of the film-plane, which we have seen already in Fig. 9.

Now let us have a look on the corresponding radial MTF-plot for the digitally taken picture with Sony A7R4 with super35-frame-mode:

**Fig. 10a**: The radial MTF-plot for Angénieux M1 50mm f/0.95 at open aperture f/0.95 on Sony A7R4-sensor – MTF30 in Linepairs per Picture Height – 100% distance to the center correspondents to 31.1mm picture circle – source: fotosaurier

Scattering of measuring points is smaller due to very good parallel allignment of sensor to target (horizontal 0.04°, vertical 0.05°) – but clearly seen is the much lower level of resulution in part way and corners (attention: different scales on vertical MTF30-scale!).

Now we have all bits an pieces together, to show the perfomance of this lens on analog film (Agfa APX100) and on high-resolution sensor (A7R4) in comparison.

This is the resolution of the Angénieux M1 50mm f/0.95 lens at all apertures (0.95 …. 22 – a remarkable span !!) measured on analog film (B&W Agfa APX100) :

Fig 11: MTF30-resolution of Angénieux M1 50mm f/0.95 for all apertures **on analog film APX100 – 22 MP-scan**. Crop from FF-picture is corresponding to a **picture circle of 34.2mm**. Nyquist Frequency of the analog film scans is 2,296 LP/PH (blue line) – source: fotosaurier

Now we can compare the results of this lens measured on the sensor of the Sony A7R4 in Super35-mode, which is restricted to a picture circle of 31.1mm, whereas the analog values in Fig. 11 come from a picture circle of 34.2mm:

Fig. 12: MTF30-resolution of Angénieux M1 50mm f/0.95 for all apertures in **Super35-film-format on Sony A7R4** (NyqFreq 2,080 LP/PH – blue line) – corresponding to a **picture circle of 31.1mm** – source: fotosaurier

Please consider: the green line represents the resolution in a small central picture-area, the yellow lines only the outmost coners (>80% of the picture radius).

The grey curve („part way“) represents the average resolution value of the biggest area of the picture, so it is responsable for the general sharpness impression of a picture.

Though the corners of the pictures taken on the analog film are 1.5mm more distant to the center of the picture, the average corner values at f/0.95 are more than 100% higher than on sensor – which is also similar for the part way resolution-values.

The 10% higher Nyquist-Frequency of the analog-scans does not explain the 20% to 100% higher resolution readings on the analog film pictures. Obviously the lens is not harmonizing with the sensor (inclusive filter-stack and micro-lenses). To me this clearly is an artefact of the sensor.

With my increasing experience with many historical „fast“ or wideangle lenses I can state today, that this is more or less happening with all historical lenses – not only those for rangefinder-cameras. With RF-lenses it is however much more dramatical. There are a few early historical lenses, which do not show these artefacts on sensor. This type of artefacts seems to vanish more and more with focal lengths above 65mm and openings <f/2.0.

The only values, which are higher with the sensor, are the central resolution readings at f/.95 – f/1.4. I have observed that for all fast lenses to date: the center-resolution open aperture is falling back on film compared to digital sensor. I am pretty sure, that this is due to the thickness of the sensitive film-emulsion, which is mostly pretty much thicker than my scan-resolution of 5.08µ – and at f/0.95 the bundles of light-rays hitting the surface of the film may be larger than the scan-pixels. Light diffusion in the emulsion-layer may also have an influence.

At maximum-stop-down with f/16 or f/22 nearly all historical („analog“) lenses show the effect of a dramatical drop of the resolution on sensor – and it is not the normal diffraction limitation. This latter you see on the graph for the resolution on film, where you see a small (ca. 15%) drop of resolution from f/16 to f/22, which may be just diffraction-caused. In the sensor-based measurements, the drop starts in the center behind f/8 and generally behind f/11 – dropping by up to 50% towards f/22. This is seen with many historical lenses on FF-sensor (at least in center resolution). But there are also exceptions.

I will add my analog/digital measurements comparison with the Canon rangefinder lens 50mm f/0.95 as soon as it will be finished.

3. How the „one-purpose-camera“ was built:

I mentioned initially, that the physical rear end of the Angénieux M1 lens barrel has to be brought closer to the film plane than 28 mm. This allows no space for adaptation on the original thread M39 or LM-bayonet.

I stripped down one of my Canon7-bodys to get as close to the film plane as possible:

Fig. 13: Canon7-Camera-body „stripped down“ – M39-lens-flange and front-screws removed to generate a flat surface, which is the closest possible plane to the film-plane – source: fotosaurier

The M39-lens-flange and front-screws removed to generate a flat surface, which is the closest possible plane to the film-plane. I planned to use the screws as a safety-lock to keep the new adapter in place.

As an adapter-ring I bought a step-ring 77-58:

Fig. 14: The new Adapter-ring in place on the Canon7-body with the cut-away for the lightmeter-window. To the right you see the stripped-off M39-flange – source: fotosaurier

The inner female M58-thread on the step-ring allows to thread down the focussing unit with the M1-lens until it touches the surface of the camera body

The step-ring only interfers with the protruding light-meter window – this forced me to make a cut-away on one side of the ring.

Fig. 15: Close view onto the cut-away sevtion on the 77-58 step-ring. You see the M77-male-thread in the cut section, which is not used here. The adapter-ring here seems to „float“ over the camera front. This is the 0,7mm-gap, which the ultra-strong double-adhesive-tape takes, to glue the large flat backside of the step-ring to the flat front of the camera-body.

The step-ring is finally glued with ultra-strong double-adhesive-tape to the flat front of the camera-body – and I could leave the two safety-scews finally away

You see the new „single-purpose-camera“, which is not at all a „point-and-shoot-thing“, because it has to be focussed and framed with a ground glass placed at the back with open back. It focuses at maximum distance to 1,84 meters, which is just matching to image my IMATEST target – or closer of course.

Fig. 16: Final result … single-purpose-camera for a special lens – just remembers me of my Hasselblad SWC, which I used to own in earlier times …

Happy „fotosaurier“, analog picture taken with Angénieux M1 50mm f/0.95 at f/2.8:

Berlin, March 21, 2024

Sony A7R4 (61 MP) vs Agfa APX100 (B&W-Film) – Analog vs Digital comparison

This article describes, how I made the resolution-power of lenses digitally measurable on analog film and COMPARABLE to the data, which are directly measured on digital sensors – using the same algorithm: IMATEST.

Since a long time I am looking for an experimental set-up, which allows me to understand, how the information content of the exposure on an analog film compares to the digital data from a digital sensor – looking through the same lens. Resolution being the main point of interest for me in this case.

Just to give you a quick impression of my results I show here the resolution charts from IMATEST on B&W-film (Agfa APX100) and on Sony A7R4 (61 MP), using the same Olympus SLR-lens OM 28mm f/2.8 (introduced 1973) – (the method will be explained in detail further down in this article):

**Fig. 1:** Resolution-chart, generated with Olympus OM Zuiko Auto-W 28mm f/2.8 lens on **black and white negative film (Agfa APX100) and filmscanner reflecta RPS 10M** – MTF30-resolution-values from center to corner for all apertures – source: fotosaurier

I do not think, that these are the „real“ limiting MTF30 resolutions values of the lens. These may be definitely higher – especially in the range betweenf f/5.6 and f/16. For me the purpose of the method is, to clarify the behavior of many (legendary!) historical lenses which show very low resolution values especially in the corners and at stop-down values of f/16 or f/22.

Let us take a look at the digital picture, taken with the Sony A7R4:

Ima_GRAPH_OM28f2,8_A7R4 — **Fig. 2:** Resolution-chart, generated with Olympus OM Zuiko Auto-W 28mm f/2.8 lens **on 62 MP-Sensor of Sony A7R4** – MTF30-resolution-values from center to corner for all apertures – source: fotosaurier

Do not let you confuse by the blue lines on different levels, which represent the Nyquist-Frequency in each set-up: the Sony’s sensor has a Nyquist Frequency of 3.168 LP/PH (linepairs per picture hight) – the filmscanner which was used to digitize the analog picture (reflecta RPS 10M) was used at its max. resolution of 5.000 ppi – that corresponds to 2.383 LP/PH as a Nyquist Frequency and delivers ca. 33 MegaPixel pictures.

There is no affordable filmscanner with higher resolution on the market!

This means: the Nyquist Frequency of the Sony Digicam is exactly 25% higher than that of the scanner, which we used as a A/D-converter for the B+W-negatives on the APX100-film.

The highest resolution in the film-based pictures generated with the analog-digital data-processing chain in Fig. 1 is very close to or above the Nyquist Frequency of the scanner – and over the full format area of 24mm x 36mm the resolution in the analog film is gathering very closely under or around this Nyquist Frequency at nearly all apertures, with the exception of open aperture f/2.8 where it is 10-20% lower.

In contrary to that, in the digital pictures taken with the Sony Sensor (Nyquist Frequency: 3.168 LP/PH) the resolutions vary strongly between corners and center and in between (part way) – and for the different apertures.

Let’s look at the center-values of resolution (green curves in Fig 1 + 2): between f/2.8 and f/11 the analog and digital values develop quite constant around the respective Nyquist Frequency, which explains, that the center values on film are 25% lower than on the 62 MP-sensor. But: The drop-off in resolution at f/16 and f/22 on the digital sensor is dramatical and shows that it is a sensor-created artefact.

Looking at the grey curves in Figs 1 + 2: „part way“ between center and corner represents the biggest area of the picture, dominating the perception of the picture! Here the MFT 30 resolution values are higher on film at nearly all apertures in spite of the lower Nyquist Frequency.

The most dramatical difference between analog and digital pictures, however, is – as expected! – in the corners (yellow curves on Figs 1 + 2):

For a better understanding I put the corner-resolution of film and sensor together in one graph:

**Fig. 3: Olympus OM** **28mm f/2.8 corner resolution on Sony A7R4 (yellow curve) and b+w-film APX100** (grey curve) – source: fotosaurier

The corner-resolution on the sensor with 25% higher Nyquist Frequency starts at f/2.8 at 50% of that of the analog film, exceeds the absolute analog value at f/8, peaks at f/11 with 82% of the sensors Nyquist and drops below the analog-value at f/22, whereas the analog-resolution on film reaches 95% of Nyquist at f/5.6 and stays at about 90% until f/22.

What the resolution-graphs here clearly show: also the very low resolutions in the corners (and even part-way!) of the digital sensor (especially open aperture!) are an artefact of the sensor! We know, that most of the effect is caused by the thick filter stack in front of the sensor. With this picture we know, that this happens not only with rangefinder-lenses, where the corners are literally BLURRED on the sensor – but also with SLR-lenses as in this case! With rangefinder-lenses the difference in corner resolution between analog (film) and digital (sensor) may be 6 to 7 … whereas with SLR-lenses I experience values of 2 to 3.

I confirm again: it is the identical lens in both cases! And these results are pretty much representative for many analog lenses! I will supply you with the results of many more lenses soon. There is one (rangefinder-)lens already analysed with the same method (link here).

EXPLAINING the Method in detail:

1. Extending the digital IMATEST lens testing method and software to pictures taken on analog film:

A. Measuring the optical performance on a digital sensor is facing several facts and influences, which are new and specific: pixel size, algorithm, problems of digital signal-processing systems like aliasing, additional optical elements in the optical path like filter stacks and micro-lenses!

The question: is there an essential influence of all these optical systems on the visual result in the picture over the picture-circle (Bildkreis), e.g. because of the varying angles at which the light-rays hit on the sensors between center and the farthest corner of the picture format or due to the additional optical elements introduced into the light-path?

In the case of RANGEFINDER-lenses we know, that there often is a strong influence of this. These lenses are often made for a very short distances between the last lens and the film – especially for wideangle- and standard-lenses. Little was known to me about historical SLR-lenses, which were never planned and calculated for the use with modern digital sensors. The degradation of the picture quality in the corners of rangefinder-wideangle-lenses is so dramatical, that it is clearly seen, that this is an artefact of the sensor, because we see sharp corners on film with the same lens.

Since several years I do quite a few measurements on historical lenses, using a high-resolution digital sensor with 62 Mega-Pixels, resulting in 60,2 MP effectively on Full Format (35mm stills).

Until now I did not know, whether the measurement of my historical SLR-lens is falsified due to artefacts, generated by the digital recording system. The work, described in this article, was done, to clearify this situation.

I just want to know: how does picture quality of historical SLR-lenses on the analog film compare measurably to that delivered by digital sensors?

Digital cameras are really big number-crunching-machines! And with the right software, I can use the numbers to generate a numerical picture of the optical quality of the lens-sensor-combination. IMATEST is such a software and it uses standardised TARGETS to do that. I use the following target:

DSC03033_Macr-Yashica_55f2,8_5,6-foc Kopie — **Fig 5**: **SFRplus** target for Imatest – it’s height is 783 mm between the horizontal black bars, which means, that the reproduction ratio on film is 33:1 – source: fotosaurier/Imatest – original information graphics from IMATEST

Over years I did – like many other amateur-photographers – compare real-world photos of analog vs. digital processing. But I was never satisfied, because this method gave me only subjective impressions – it did not create reproducible figures, to generate a precise description of the results!

I collected intensive experience with IMATEST on more than 150 lenses over meanwhile 5-6 years using the digital pictures generated by digital sensors (4,9 to 102 Megapixels) of seven different DIGICAMS. During this time, my Standard Digicam to compare lenses was (and still is) Sony A7R4 (62 Megapixels) – since it had arrived in the market (2018/19).

IMATEST (Studio) software delivers MTF-based resolution data – as it can do that separately in three RGB-channels, it also delivers lateral CA-data. Using the Target structure of Fig. 5, the software selects 46 local areas, and runs the MTF-measurement automatically for all these 46 areas. The following picture demonstrates the automatic areas, which are typically selected – but you could choose others as well:

ROI-chart (standard) — **Fig. 6**: The **46 magenta rectangles** (called „**ROI**„) frame the Edges in the target, at which the 46 MTF-measurements are made – source: fotosaurier/Imatest

These are the curves, which are generated from each digital picture (black&white):

Zusammenstellung_IMATEST_A7R4_OM28d2,8_2,8 — **Fig**. 7: Summary of the IMATEST-results for the **OM28mm f/2.8** at open aperture f/2.8 **on Sony 62 MP-sensor** (A7R4) – explanation see text beneath – source: fotosaurier

The upper left curve shows the edge-profile at center of the target (ROI no. 1, which is the left (vertical) edge of the center square in Fig. 6). From this graph the edge-rise between 10% and 90% is taken from the x-coordinate in pixels. The lower left curve is the MTF-curve (contrast over spatial frequency) for the same location. From this graph the MTF30 value (Frequency at 30% contrast) is taken: follow the horizontal line at 0,3 MTF-value to its section with the curve and take the frequency on the abscissa. The right curve shows the MTF30-values of ALL 46 ROIs plotted over the distance from the center in the 35mm-fframe.

I have resumed the IMATEST test-method in more detail in this article here in my blog!

B. Digital measurement of resolution on analog film

Now I decided to make the following experiment:

Take a photograph of the IMATEST-target on analog film;
digitize the picture with a film-scanner;
analyse the resulting digital picture with IMATEST.

For the tests, which I describe here, I used the following hardware:

28mmf2,8-on-OM4Ti_DSCF1655_blog — **Fig. 8**: Analog SLR Olympus OM-4 Ti with Zuiko Auto-W 28mm f/2.8, loaded with „fresh“ Agfa APX100

Camera for the shooting on analog-film: Olympus OM-4Ti
Lens: Olympus Zuiko Auto-W 28mm f/2.8 (Ser.no. 232073)
Film: B&W negative film AgfaPhoto APX100, iso125, developed in Rodinal 1+25 (8′)
Scanner: reflecta RPS 10M film scanner

The OM-4Ti (about 25 years old) and the lens (nearly 50 years old) work still perfect. I let the OM-4Ti automatically generate the exposure time: from 0.4 seconds to 1/250 seconds. The densitiy of the negatives was pretty constant on the film-strip! I use a sturdy tripod, which is made for use with long astronomical telescopes.

With this method I hope to use the full analyzing-power of IMATEST-software on a picture-frame, which is generated through the lens WITHOUT the typical artefacts, which digital sensors MAY generate in the optical path of a historical lens.

ON THE FILM we have now the IMATEST target-pattern, which allows to make a fast and powerfull analysis of optical data over the full picture frame – also very close to the edges and into the corners. This pattern is superimposed by the typical grain-structure of the light sensitive layer – and potential light-diffusion-effects within the film thickness. Both (analog) effects LIMIT the resolution, which can be achieved on FILM.

My first and major interest was always focused on the observation of the enormous difference between the center-resolution (see Fig. 7), which is digitally measured on A7R4 with ca. 3,000 LP/PH or higher) and corner-resolutions of <200 to 600 LP/PH on the sensor .

The question is: are the low values on edges and in coners of the frame, measured with the digital sensors, an artefact, caused by the different light-path? We know definitely about these effects with rangefinder-lenses, which have a very short back-distance between last lens and film, causing big trouble on sensors of mirrorless cameras. This is today well known, to be mainly caused by the thick filter-stacks in front of the sensors (creating field-curvature and cromatic aberrations with analog lenses).

It has been shown, that this can partly be „cured“ – or at least reduced – by reducing or deleting the filter-stack, and/or putting a positive lens (so-called „PCX-filter“) in front of the lens-sensor-combination.

The 35mm-negative-film:

I made my first attempts to photograph the IMATEST-target on film with

b&w-film Agfa APX100, iso 100

which is still available as „fresh“ product. For this first step I decided to stay with b&w-film, because I can process it myself under controlled conditions. With colour negative film I would have an external influence, which I could not control! Just for resolution this means no restriction in the information, because CA-errors also blurr the B&W-picture!

I did the devellopment of the b&w-film myself with Rodinal.

The A/D-converting:

The negatives were digitized through my film-scanner reflecta RPS 10M,which offers a maximum linear resolution of 10,000 pixel per inch (PPI).

To me, this step seemed to be very important: to avoid new artefacts from the digitizing algorithm. So I chose a spatial frequency, which is higher than the expected limiting spatial frequency of the film: I set the scanner at 5,000 ppi. On pixel-level this corresponds to an imaging-sensor of ca. 33.7 MP (for 24mm x 36mm).

From my earlier estimations I had found, that a normal recording film for general imaging purposes should correspond to a digital FullFormat-sensor with 20-12 MP.

The picture height, which the scanner digitally delivers (24mm minus a bit of crop to frame the target safely), was 4,676 pixels and so the „Nyquist Frequency“ of the scanner set-up corresponds to 2,338 LP/PH – corresponding to an effective sensor-size of 32,7 Mpxls.

Fig. 7 shows the b&w-picture, which was generated with the scanner:

AGFA100_OM28f2,8_2,8_H4536 — **Fig. 7:** Scanner-output from the b&w negative-film Agfa APX100 from Olympus OM 28mm f/2.8 at full aperture f/2.8. **Picture-hight** of this original scan is 4.676 Pxls. You see, that the light-fall-off of this lens into the corners is very moderate … and the linear distortion with exactly 1% acceptable as well! – source: fotosaurier

Let’s have a closer look into the structure of this image – in Fig. 7a you get an impression of the grain structure of the films emulsion at about 200% enlargement of the 33 MP-image:

Enlargement-Film-200% — **Fig. 7a**: Overview of the grain-structure at ca. 200% enlagement of original scan in Fig. 7. The pixel-size here is 5,3 µm – the grains of the film are bigger than the pixels – source: fotosaurier

Following picture is the MTF-curve of the analog image „as scanned“ (in the center of frame):

Fig 8: MTF-curve in. Center (ROI no.1) of OM-Zuiko 28mm f/2.8 at f/2.8 – source: fotosaurier

The „noise“ in the curve is caused by the film-grain, which is about the same size as pixels.

Film_3024-pixel-height_at-800% — **Fig.** 9: Here we look at about 1,000% into **the pixel-structure of the scanned image**. At the edges of the dark rectangle (where the resolution is analysed!) the grain-diameter is about the same size. Only some local „grain-clusters“ are considerably bigger – source: fotosaurier

Previous trials had shown, that with a film with this grain-structure, this digital image-size would give adequate results for MTF and resolution.

In the case of a digital sensor of a digicam I avoided generally to use RAW-data, which would have urged me to use my own very personal „development-parameters“ in Lightroom or other software to generate the final picture. I use OOC-JPEG-Data at „Standard“-settings, due to generate conditions (all important parameters set to „zero“), which are transparent and reproducible for everybody with the same camera-model! That means: it would also have been possible to create pictures with much higher resolution results in Imatest, e.g. by setting higher sharpening-parameters or the „clear“-mode.

Now with a film-scanner I had to go myself through a very intensive process of defining the „development-parameters“ in Silverfast. Starting with the setting to 5.000 ppi for the basic scan-resolution. With 10.000 ppi, which is offered with this model, you will get no REAL increase in EFFECTIVE resolution.

However, using the „Multiple Scan Mode“, you extend the accessible resolutions above the „Nyquist Frequency“, which would be 2.383 LP/PH, corresponding to a Picture size of 32,7 MP.

My target was, to reach about the same level of resolution in the center of the scanned images on analog film as with the Sony A7R4 images, which means in the range of 3.168 LP/PH, which is the Nyquist Frequency of the Sony Sensor.

This corresponds with a resolution of 260 Lines/mm.

I came close to this with the following settings:

**Fig. 10**: Scan-parameters in Silverfast 8 on film-scanner RPS 10M – source: fotosaurier

See the complete results here:

**Fig. 11**: Analog on film resolution results of Olympus OM 28mm f/2.8 SLR-lens with b+w-film APX100, scanned with RPS 10M film-scanner – source: fotosaurier

The interpretation of this in comparison with the measurement-results on the 62 MP-sensor of the Sony A7R4 (Fif. 2) has been given in the first section of the Article.

Finally I asked myself, whether a PCX-filter (lens) could improve the resolution-artefacts which are found on the sensor? But I found no real positive effect.

**Fig. 12**: Resolution of OM 28mm f/2.8 lens with PCX-3m lens on the Sony A7R4-sensor: no improvement at all – source fotosaurier

**Fig. 13**: Soon I will enter a new article, showing the performance of this wideangle-lens on seven different cameras – **link not yet available** … stay tuned!

Herbert Börger, Berlin, November 2023

My Crazy Lenses – Topcor R 30cm f/2.8 and its Modern State-of-the-Art Counterparts – „Supertele-Lenses“

Travel on my time-machine
The known Facts – Topcor 30cm f/2.8
Topcor 30cm f/2.8 – Optical Performance
The Reference: Canon EF 300mm f/2.8 IS USM
Three more 300mm f/2.8-teles

DSCF2458_Alle300f2,8-5_blog — **Fig 1**: From left to right – Topcor R 30cm f/2.8, Arsat Yashma 300mm f/2.8, Tamron SP LD (IF) 300mm f/2.8, Minolta AF Apo-Tele 300mm f/2.8, Canon EF 300mm f/2.8 L IS USM

Attention: I part from my „crazy lenses“ due to my age: it takes too much time to take care for my optical baybies! If you are interested in the Topcor R 30cm f/2.8, please leave a price-proposal in the comment-section (this is seen only by me) or by mail to webmaster@fotosaurier.de

1. On my time-machine:

I own the Topcor R 30cm f/2.8, which I am looking at here, since a few years – but I have not used it too often. It is very heavy, long and dark, giving the impression of a tank-breaking weapon: you definitely will get trouble at any security check nowadays … and in the best case you will earn compassion instead of admiration! Too bad, because it is an ingenious piece of optical engineering.

Information about Topcor lenses today are rare and not always reliable. I will restrict myself to reliable information and I will try to verify legends … or destroy them.

So I entered my time machine and travelled back into the year 1958. I was 13 years old at my arrival there – and at the Topcon (Tokyo Kogaku) factory I met a team of innovative engineers, who were fanatically burning for the QUALITY of their products – and really proud of it! The year before (1957) they had introduced a new SLR-camera (Topcon R), which was designed in Bauhaus-style, i.e. with clear and modern lines – and they were ready to ignit a firework of innovations around the SLR-concept within the next few years (from first-in-industry TTL-exposure-metering to first electric winder).

And they had introduced a line of lenses for this SLR-system-camera, among which the Topcor 30cm f/2.8 peaked out. Another „first-in-industry“-innovation.

I looked around in the photo-stores and could not find any Canon- or Nikon-SLRs there: the dealers told me, that both companies were just bringing out SLRs. It seemed, that the Topcon-people had considered the German SLRs, which were already on the market, as their competition. Personally at that time I was already a SLR-user (of my father’s Contaflex – which means, that from time to time my father was still allowed to use it himself).

Everybody, who is acqainted with the rules of the market, would have expected, that shortly after an innovation like the Topcor R 30cm f/2.8, the major competitors would bring out a similar product.

But that did not happen – so I returned in my time-machine. Finally I found out, that it took the new japanese competitors more than a decade! And there was no comparable Lens in Europe, as far as I could see. 13 years later Nikon presented a prototype, to be tested during the Olympic Winter Games of Sapporo in 1972.

The real next step was taken by Canon with a 300mm f/2.8 Lens for their new FD-System, using a lens made of FLUORITE in 1973 (some say 75)! This was finally 16 years after the arrival of the Topcor-lens … and just in that year, when Topcon stopped the production of their supertele-lens.

2. The known facts:

This Topcor R 30cm f/2.8 monster-tele-lens with 300mm focal length was presented to the world in 1958 („Topcon Club“ says 1957!) – one year before Canon or Nikon started to produce any SLR – and 13-16 years before any other lens- or camera-maker presented such a fast 300mm tele-lens. Not only at the 1964 Olympic Games in Tokyo but all the time until 1972 it was without any competition. As a consequence, there even was produced quite a number of lenses with Nikon mounts! Next to Topcon, Canon brought out its Canon FD 300mm f/2.8 S.S.C. Fluorite lens in 1973 – setting the level for professional superlele-lenses for the next decades and until today. Just a few years later Topcon went completly out of the business with SLR-cameras and lenses. Sad, but even the extensive book „Topcon Story“ by Marco Antonetto and Claudio Russo (1) does not answer the question „why?“. Today Topcon is a market-leader in geodesic instruments.

Stephen Gandy (3) estimates – cameraquest.com – that 700-800 lenses have been produced in total during 18 years of production.

Topcor-R-300f2,8_DSCF2335_blog — Fig. 2a: R.Topcor 1:2.8 f=300mm on Topcon SuperD- source: fotosaurier

Titel_DSCF2320 — Fig. 2b: R.Topcor 1:2.8 f=300mm – source: fotosaurier

R328cut — Fig. 2c: Lens scheme of Topcor 1:2.8 30cm – source: http://www.topgabacho.jp/Topconclub/lens3.htm

The lens is made of six single lenses in four groups – of which lens no. 6 (group 4) is the filter (diameter 39mm), which is, of course, part of the optical design! This filter is an early (maybe the first) example of a filter which is positioned in a slot in the rear part of the lens-body. In the book „Topcon Story“ (page 128) there is an error in the spreadsheet listing of the data of the R.Topcor-lenses: the data in the last line are the data of the „300mm 2.8“ and not of the f/5.6-lens. Here the no. of elements is „five“, which is correct, when you don’t count the filter as an active optical member …

The lens has a preset diaphragm and has a built-in sunshade (telescoping in two stages!). It is 383 mm long (from camera-flange to front-edge of the pulled-back sunshade – total length with shade pulled out is 477 mm) and weighs 3.1 kgs (without front and rear caps). Measured at my sample (ser. no. 34.1359). The initial sales-price was $ 1.125,–. (In the literature you will find: 415/412 mm length and 3.3 kgs weight).

It may be interesting to mention here, that right away from the introduction of the first Topcon-SLR, an extremely ambitious lens-program was planned – however, realized only partly. The Topcor R 13,5 cm f/2.0 (6 lenses) had also preset diaphragm and it was discontinued with the Topcon RE camera system – so it is said to be extremely rare. It has a yellowish color cast (due to rare-earth-glass?), not a big problem with todays digital cameras …

However, a 50mm f/0.7 lens, which is mentioned in „Topcon Club“ only, was never made for the SLR-camera market – maybe, this was one of the very early oscilloscope-registration-lenses, which are also known from Germany and GB even at WWII-times.

And a 1000mm f/7 catadioptric lens was only experimentally made in 1958.

„Topcon Club“ (2) writes about this:

„The interchangeable lenses which appeared with the appearance of TOPCON R are various kinds of the Auto Topcor of 35mm/100mm, and R TOPCOR (a preset diaphragm) of 90mm/135mm/200mm/300mm among these – although the bright thing and the dark thing were prepared about 135mm and 300mm – it should mention especially – it is the „high-speed lens“ of 135mm f2 and 300mm f2.8. 50mm f0.7 – such a bright lens was already completed during wartime by the Tokyo optics. Do you believe it ? Although possibly this grade was an easy thing, even so, the 300mm f2.8 lens will be an astonishment thing in 1957. I talked in detail on „the page of TOPCOR“ about this lens. We have to wait for marketing of the product of NIKON which is the next 300mm f2.8 lens at any rate till 1977. However, TOPCON did not build the super telephoto lens 500mm /800mm those days. Furthermore, the Refrector Topcor 1000mm f7 is appearing in the catalog in ’59. However, this was not launched regretfully.“

Later – from 1969 on – a RE Topcor 500mm f/5.6 telephoto-lens was even produced with automatic diaphragm and meter coupling!

Can such a fast long telephoto lens like this early 300mm f/2.8-design without Fluorite- or ED-lenses be any good – on the scale of professional photography? There are hints, that rare-earth glasses were used to make these lenses (also for the other famous 13,5cm f/2.0, also supplied since 1958). But I do not know details about this.

I will answer the question about the optical quality here – also comparing this lens with a modern top-notch tele-lenses like Canon EF 300mm f/2.8 L IS USM, which I personally classify as today’s state-of-the-art reference, supported by photo-friend Thomas, who borrowed his Canon lens to me.

Finally I will take a glance on a state-of-the-art modern astronomical refractor, which normally does perform at diffraction-limited resolution on stars!

Topcor 30cm f/2.8 – The Optical Performance on analog film (year 1969):

Stephen Gandy (3) wrote in his blog:

WIDE OPEN its resolution was 56lines/mm center and 34lines/mm at the edges. By f/8 it was 80 lines/mm center and 65 at the edges. Many normal lenses don’t achieve this sharpness — much less 300/2.8 leviathans ! Camera 35 summed it up by saying „INCREDIBLY FANTASTIC.“ I would have to agree. „

(In the original text in Stephen’s blog, the reported resolution values are noted as „56mm“ or „34mm“. I have taken the freedom, to correct this to what it should read: lines per mm, „lines/mm“!)

The resolution values, which I use in my digital IMATEST measurements, typically are given in „Line-pairs per picture-height“ = „LP/PH“. Picture-height being 24mm with 24×36-format, you have to divide the „lines/mm“-values by two to get to „line-pairs“ – and then multiply with 24 to achieve LP/PH.

The highest given value of 80 lines/mm corresponds to 960 LP/PH stopped down to f/8 in the center or 760 LP/PH at f/8 at the edge – the lowest value 34 lines/mm with open diaphragm at the edge corresponds to 408 LP/PH.

What does that mean?

In 1969 the test results for resolution were measured on film – „Modern Photography“ used Plus-X Pan with standardized development – and the reading of the „just resolved“ line-pattern was done with a standardized enlarging glass … I personally used the method myself at that time, too, and it is quite reproducible as long as the same person does the reading … It is somewhat sensitive to the vision-capabilities of the reading person! And of course the grain of the analog film material (negative b&w film!) is the limiting factor for the resolution-reading on film for really high resolutions.

Today’s modern 24 MP-sensors deliver resolutions of 2,000-2,400 LP/PH using MTF30 (30% contrast) as the parameter for reading out the resolution values from the MTF-curve. My Sony A7R4-Camera (62 MP), which I use for my measurements, has a Nyquist frequency of 3.168 LP/PH and delivers up to 3.800 LP/PH-readings with the best known lenses.

The following spreadsheet gives an overview on the physical data of the Topcor-lens and the other lens-monsters, analysed here:

300f2,8_physData — **Fig. 3**: Physical Data of the five 300mm f/3.8-Lenses – source: measured by fotosaurier

3. Topcor 30cm f/2.8 – Optical Performance

My IMATEST-Results of the optical properties of the Topcor R 30cm f/2.8 lens:

To exclude potential vibration-initiated degradation of resolution in my test-shots at these long focal-lengths I used my heavy (>10 kgs) and sturdy astronomical telescope-mount:

DSCF2537_Topcor_AufAstroMontierung_blog — **Fig. 4**: My massive astronomical lens mount – here with SonyA7R4 attached to Topcor 30cm f/2.8 – source: fotosaurier

DSCF2535_OnTargetTop — **Fig. 5**: The set-up keeps the lens and camera steady even at 0,4 seconds. – source: fotosaurier

Following you see the results of my IMATEST-measurements:

Topcor_R-300f2,8_Spreadsheet-23 — **Fig. 6:** Optical measurment-results for **Topcor R 300mm f/2.8** adapted to Sony A7R4 with 62 MP – resolution values given in LP/PH – source: fotosaurier

Topcor_R-300f2,8_Graph-23 — **Fig. 7**: Resolution measurement-results for **Topcor R 300mm f/2.8** as graph – source: fotosaurier

The lens is unique at that time regarding to „speed“ – an extremely ambitious piece of optical engineering. Remind, that the distortion is practically zero and the CA-area in the center 0,8-1,4 pixel – 1 pixel at Sony A7R4 is 3,8 microns on the sensor!

What is center, what is part way and what is corner? In the following graphs from IMATEST you see: „Part-Way“ is the large part of the picture extending close to the narrow side (left/right). „Corner“ is the narrow area outside the second dotted circle on the picture below.

DSC07122_Topcor300f2,8_8,0_Multi-ROI_2023-02-04_01-00-54 — **Fig. 8**: „**Cente**r“ resolution is calculated as mean from the values inside the inner circle (in my setting always two values), „**part way**“ is the mean of all values between the inner and outer circle, „**corner**“ is the mean of all values positioned outside the outer circle – source: fotosaurier

DSC07110_Topcor300f2,8_8,0_Lens_MTF_2022-11-29_22-55-46 — **Fig. 9:** Topcor R 30cm f/2.8 resolution plotted over radius of picture circle – source: fotosaurier

So, let’s compare the measurements to the value, that were given in analog times on film:

The comparison in the spreadsheet Fig. 10 shows: The lens „out-resolves“ normal analog films by far! Stopped down it reaches the limits of the analog medium even at the edges of the frame!

Analog-digital-resolution — **Fig.10:** „Camera35’s“ resolution measurements for Topcor R 30cm f(2.8 of 1969 on film compared with digital IMATEST values (at 30% MTF = „MTF30) with Sony A7R4 – source: fotosaurier

I found no real technical explanation, how Topcon-engineers managed to generate this phantastic lens at that time without ED/LD/AD/Fluorite-glass. There is a second tele-lens – the 13,5cm f/2.0, also introduced 1958, with first-in-industry potential – and finally the Topcor 2,5cm f/3.5 super-wide, which surprises with best-in-class resolution values (see my blog-article on historical 24/25mm-lenses!).

If somebody knows the secret: please, tell us!

Look at a sample picture taken with the Topcor at the end of this article at 65% enlargement size (see Fig. 24).

Now, let’s have a glance on some other historical Superteles:

Alle_300er_2,8_DSCF2573 — **Fig. 11**: From left to right: Topcor R 300mm f/2.8, Canon EF 300mm f/2.8 IS USM, Minolta AF Apo-Tele 400mm f/2.8, Tamron SP (60B) 300mm f/2.8 LD (IF), Arsat Yashma-4H MC 300mm f/2.8

4. The Reference: Canon EF 300mm f/2.8 IS USM

Canon-EF_300f2,8_DSCF2451_blog — **Fig. 12**: **Canon EF 300mm f/2.8 IS USM – source: fotosaurier**

Canon EF 300mm f/2.8 IS USM is rated as the reference of this class of lenses. In this case it is not the latest „Mk II“-version of it, which came out 2011 – but the first version of 1999, which is tested here. It represents nevertheless already the top-class of the super-teles (as all its predecessors since 1973!)

Here are the IMATEST results of its optical properties:

Canon-EF_300f2,8-L-IS-USM_AF_Spreadsheet — **Fig.13**: Optical properties of **Canon EF 300mm f/2.8 IS USM** from my IMATEST-measurements, with autofocus – source: fotosaurier

And here the Graphs of resolutions center, part way and corner:

Canon-EF_300f2,8-L-IS-USM_AF_Graph — **Fig. 14**: IMATEST-Resolution (LP/PH) of **Canon EF 300mm f/2.8 IS USM –** center – part way and corners – source: fotosaurier

Not may comments necessary to this – the figures and graphs should speak for itself … Just to mention: the distortion at the Topcor-lens is even lower than that of the Canon – but both are neglectable for a supertele!

Canons leadership in this class of professional supertele-lenses was generated by the policy, not to drop a product into the market, which was „just possible“ at present, but to persue a consequent plan for the future: to solve the „secondary spectrum“-problem of long tele-lenses, which means to use extreme „anormal dispersion“ lens-materials, which do the job without optical compromising.

So in 1975 – 2 years after Nikons first presentation of its first 300mm f/2.8 ED-lens (which was not very convincing and had to be replaced four years later by the ED-IF-version) – Canon introduced their FD 300mm f/2.8 Fluorite-Supertele, in which they used a front-lens made of fluorite-monocrystal material (no glass!) and a UD-glass-lens. This lens was already praised close to perfect (absence of chromatic aberrrations). Canon accepted for this a compromise, which made the lens longer and heavier: to protect the soft and sensitive fluorite-crystal-material in the front lens, there was a fixed additional plane protection element of glass in front!

Finally new fluorite-glass-formulations became available, which allowed to drop the sensitive crystal-lens. Over the introduction of Autofocus (EOS – 1987) and still more glass-elements, Canon finally introduced the legenday lens EF 300mm f/2.8L IS USM in 1999 with very fast AF and image-stabiliser, which is tested here.

Enjoy the results!

5. Finally – three more 300mm f/2.8-teles:

Minolta AF APO-Tele 300mm f/2.8 (1985)
Tamron SP LD (IF) 300mm f/2.8 (60H) (1984)
ARSAT MC Yashma-4H 300mm f/2.8 (1990?)

For these three lenses I also have to thank foto-friend Thomas, who borrowed them to me!

5a. Minolta AF APO-Tele 300mm f/2.8 (1985)

Minolta-Apo_300f2,8_DSCF2460_blog — **Fig. 15**: **Minolta AF APO-Tele 300mm f/2.8 – source: fotosaurier**

This lens had a mechanical defect: the diaphragm could not be closed below f/5,6. However: in these lenses principally mainly the open aperture is really significant – why should you carry around such a weight, to make pictures with f/11?

Minolta-AF-Apo-Spreadsheet — **Fig. 16:** Optical properties of **Minolta AF 300mm f/2.8 Apo – source: fotosaurier**

Minolta-AF-Apo-Graph — **Fig. 17:** IMATEST-Resolution (LP/PH) of **Minolta AF 300mm f/2.8 Apo –** center – part way and corners – source: fotosaurier

This Minolta lens comes closer to the Canon-legend than any of the others – but with quite some distance in resolution in the corners open aperture.

Excelent lens!

5b. Tamron SP LD (IF) 300mm f/2.8 (60B) (1984-1992):

Tamron-SP_300f2,8_DSCF2475_blog — **Fig. 18**: **Tamron SP LD (IF) 300mm f/2.8 (60B) – source – fotosaurier**

This is the shortest and lightest lens of the quintuple, which arrived even one year before the Minolta – containing two low-dispersion (LD) lenses – with manual focusing:

Tamron-SP_300f2,8_Spreadsheet — Fig. 19: Optical properties of **Tamron SP 300mm f/2.8 LD (IF) 60B – source: fotosaurier**

Tamron-SP_300f2,8_Graph — **Fig. 20**: Imatest resolution graphs of **Tamron SP 300mm f/2.8 LD (IF) 60B – source: fotosaurier**

Tamron – third party winner: Great Lens!

5c. ARSAT MC Yashma-4H (1990?):

Yashma_300f2,8_DSCF2478_blog — **Fig 21**: **ARSAT MC Yashma-4H – source: fotosaurier**

I do not know much about this lens. Funny about it is to me, that in most cases, when it is offered as a used lens, it is given the addendum „sovjet lens„! In 1990, when it was delivered first (I saw other sources with the date 2007 …) the Sovjet Union no longer existed – which means that, ARSAT being located in KIEW, the lens has UKRAINIAN roots.

As far as I know, it was generally produced in Nikon-mount.

ARSAT_Yashma_300f2,8_Spreadsheet — **Fig. 22**: Optical performance of **Arsat MC Yashma-4H 300mm f/2.8** – source: fotosaurier

ARSAT_Yashma_300f2,8_Graph — **Fig. 23**: Resolution graphs of **ARSAT MC Yashma 300mm f/2.8** – source: fotosaurier

Open aperture and stopped down the lens is convincing in the center – about 10-15% below the other superteles – but with still very good CA in the center.

From f/4.0 it is also very good in the large part of the frame – just 10% below the Topcor.

In the corners it is on par with the Topcor open aperture – but it does not improve so much while stopping down. For analog film use it was also a good lens – with exception of the softer corners with typical CA-values of non-apochromatic lenses … and a much higher distortion than all the other superteles.

What about apochromatic correction in supertele-lenses?

Lenses of 300mm f/2.8 need apochromatic correction to be really sharp. The chromatic aberrations („secondary spectrum“) are the major restictions in sharpnes for these long focal lengths all over the frame! All these lenses, tested in this report, have apochromatic correction – in varying degrees of perfection! In the ARSAT Yashma the apo-correction is only partly successful.

Herbert Börger

fotosaurier, Berlin 13.02.2023

Literature:

1- „Topcon Story – Topcon Enigma“ by Marco Antonetto and Claudio Russo, by Nassa Watch Gallery, Collectors Camera Publishing, CH 6907 Lugano, Switzerland – 1997

2- Web site „http://www.topgabacho.jp/Topconclub/FPslr1.htm“

This, the first super fast long telephoto lens produced for any camera system world wide, came to the market in 1957. This was a large and heavy lens, with a 130mm maximum diameter, a length of 412 mm and a weight of 3.3 kg. The optical design was one of 6 elements in 4 groups. The selling price, at the time, was 135,000 Yen making it the most expensive lens on the market. Special filters slide into a slot at the rear of the lens barrel and this lens was probably the first to use this method. Unlike the 135mm f2 R Topcor, this lens was listed in catalogues into the later half of the 1970s. Because of it’s large aperture it was chosen as the official lens of record for the Tokyo Olympics. An odd thing concerning this lens is that many of those remaining have been modified for the Nikon mount, while those with the original Topcon mount are very scarce. The early lens case was made of leather but later on Topcon began supplying a hard case with the TOPCON emblem promontory displayed. The R Topcor 300mm f2.8 lens still compares favorable, with regards to regards to sharpness and contrast, to modern lenses with fluorite elements. Today this lens is almost forgotten but was highly praised in former times.

3- Web site of Steven Gandy: „https://www.cameraquest.com/top30028.htm“

**Fig. 24**: Mathäuskirche in Hambühl, seen **from 1,2 km distance** with Topcor 300f2,8 (taken at f/5,6 with Sony A7R2 at iso800) – narrow vertical crop of nearly full frame, which you see here at about 65% enlargement – „**ooc“** – no post-treatment of the picture) – source: fotosaurier

Two crazy lenses of the 1950s – Angénieux 50mm f/0.95 and Carl Zeiss Jena Biotar 50mm f/1.4 for 35mm Cine-Format – plus Canon Lens 50mm f/0.95 from end of 60s

A few weeks ago I was blessed, having an Angénieux 50mm f/0,95-lens and a „Biotar 50mm f/1.4″, at the same time in the same place !

An Angénieux 50mm f/0,95-lens in perfect optical quality and with aperture-mechanism and rehoused into a perfect Sony-E-body, focusing to infinity and ready for measurement in my optical IMATEST-Lab…. this is really a „unicorn“!

**Fig. 1:** Ultra-rare 50mm f/0.95-lens for Cine 35 movie-format – this lens-series (10mm, 25mm and 50mm) founded Pierre Angénieux‘ high reputation in cinematic optics! – source: fotosaurier

The „Biotar 50mm f/1.4″, in great overall condition, which I even did no know about, before I saw it for the first time.

Biotar58f1,4-front_DSCF1765 — **Fig. 2:** One of the best high-speed-lenses ever made in Jena – Biotar 50mm f/1.4 of 1955/56 for Pentaxflex AK-16 cine-camera system – professional performance for professional use! – source: fotosaurier

Photo-friend and co-nerd Thomas handed out both ultra-rare lenses to me for closer optical inspection. I am a happy man!

Fig. 3: Two very rare lenses at the same time in the same place … in my IMATEST-Lab! Sheer happiness! – Source: fotosaurier

Angénieux 50mm f/0,95 (Type M1):

Thomas has proven, that it is possible to re-house the Angénieux-lens for general photographic use with infinity focus:

Fig. 4: The early super-fast Angénieux 50mm f/0.95 lens 0f 1954/55 here in a „Unikat„-version – the basic lens is directly fitted to E-Mount for Sony – source: fotosaurier

Starting in 1953 Pierre Angénieux brought out a series of lenses with f/0.95. In 1953 it was firstly the 25mm f/0.95 (which became the most famous Angénieux lens due to the use in NASA-spaceflights to the moon!) made for cine 16mm format and the 10mm f/0.95 for 8mm-cine.

A few months later he pushed out also a version for 35mm-cine: the 50mm f/0.95 – probably this was in in 1954 – originally in C-Mount. Hartmut Thiele dates this to 1955. It is important to understand, that this is not a lens made for still-photogray amateur use – but Pierre Angénieux showed here all his knowledge dedicated for professional cine-use. He went to the limits of everything, which was possible with glass-types and design- and production-methods at that time!

If you need more information on Pierre Angénieux, please look up my Blog article here!

Following my measurements on the IMATEST-target the picture-circle, that this lens covers is 37mm – so it is falling a bit short from the 43mm needed for covering the still-photo-35mm-full-format (24 x 36 mm).

DSC05014_Ang_50f0,95_0,95-foc1,4_Bildkreis — **Fig. 5**: Picture of IMATEST-Target through **Angénieux 50mm f/0.95 at f/0.95** in the 24 x 36 mm full-frame of the Sony A7R4 – Source: fotosaurier

This test-set-up generates the following resolution-measurement results:

Fig. 6: Resolution at center/part way/corner of Angénieux 50mm f/0.95 on Sony A7R4 (60,2 MP-sensor – 9.504 x 6.336 pixels!) at standard distance full-frame (24×36) – Source: fotosaurier

In spite of the heavy darkening in the corners, the system does still generate results, but these readings are not very reproducible … these corner-readings are located clearly outside the picture-circle for this lens!

So I made a second set-up with the camera set a little bit further away from the target, so that the individual measuring areas move somewhat towards the center of the picture and do not suffer too much from the dark areas out of the picture circle of the lens.

Fig. 7: Angénieux 50mm f/0.95 moved a bit backwards from the target – measurement-areas (marked magenta rectangles) moved somewhat further towards the picture center – avoiding overlap with the dark corners – this picture is at f/8, showing a sharper limit to the dark corner-areas! – source: fotosaurier

Now the furthest measurement locations are at 82% of the full-frame picture radius, clearly inside the bright circle which this lens covers at 86% of full-frame radius!

The result is seen in the following picture:

Fig. 8: Resolution with refocussed Angénieux lens 50mm f/0.95. The corner-resolution-values are still located outside the Cine35-picture-frame!!! The „peak“ at f/4 in the corner reading is real – no error – never seen anything like this with any other lens! – source: fotosaurier.

In Chapter 4 at the end of the article I will ad thwe measuremts at cine-format for all three lenses (Super 35: 18,66mm x 24,89mm). This will give more realistic resolution-readings. The Super35 crop-mode on the A7R4 is 6.240 x 4.160 pixels.

2. Carl Zeiss Jena Biotar 50mm f/1.4:

About the same time, DDR-based Carl Zeiss Jena created a high-speed lens for its own Pentaxflex AK-16 cine-camera system in Pentaflex-16 mount.

It seemed logical to follow the already successfull BIOTAR-formula and it came out around 1955 or 1956 the Biotar 50mm f/1.4:

Biotar58f1,4-2_DSCF1757 — **Fig. 9:** **Carl Zeiss Jena Biotar 50mm f/1.4** for Cine-Format, arriving 1955/56 – Source: fotosaurier

Looked at with the sensor of the Sony A7R4, the picture-circle is a bit larger than with the Angénieux … there are only minimal dark corners!

Bildkreis_DSC05072_Biotar-50f1,4_1,4-just-foc — **Fig. 10**: Full-frame picture of IMATEST-target through Biotar 50mm f/1.4 at f/1.4 – Source: fotosaurier

Of course, we have here the same situation, that the corner-measurements are quite a bit outside the cine-picture frame of typically 16mm x 22mm:

Fig. 11: Biotar 50mm f/1.4 in the same frame as Angénieux seen in Fig. 7 – source: fotosaurier

I will also with this lens repeat the measurement, restricting the resolution-target to the cine-picture frame – see section 4 at the end of the article.

The results show for both lenses, that the resolution in the center is extremely high – even wide-open! Both lenses are extraordinary lenses of their time – the mid-1950s!!!

Unique: „first-in-industry“ point of view for the Angénieux 50mm f/0.95 in its extreme speed, without sacrifycing to the center resolution!

3. Canon Lens 50mm f/0.95 for rangefinder (Canon7) cameras with LTM 39mm – of 1969

As we are just talking about early historical high-speed lenses, the step to the famous CANON 50mm f/0.95 (for rangefinder) is logical. It is a step of 15 years in time – and this time the lens is really dedicated to 35mm still-photo full-format 24mm x 36mm!

Noch'nPaar_DSCF1775 — **Fig. 12**: Angénieux 50mm f/0,95 of 1954, left, and Canon 50mm f/0.95 of 1969 / the normal still-photo-version here – Source: fotosaurier

Here is my comparable resolution-measurement with two samples of (s.n.18924 and s.n.22372) of the Canon 50mm f/0.95 on Sony A7R4 for this lens at full 24×36-format:

Crf_50f0,95_Graph — **Fig. 13**: Resolution-Graph of **Canon 50mm f/0.95** on Sony A7R4 (60,2 MP) – Source: fotosaurier

Fig. 13a: Resolution-Graph of Canon 50mm f/0.95 on Sony A7R4 (60,2 MP) – Source: fotosaurier

To allow for the necessary rangefinder-coupling besides the huge rear lens, this lens is „cut free“ at the edge for this purpose.

Crf59f0.95_DSCF1687 — **Fig. 14**: Cut-away at the 50f/0.95 Canon’s rear lens, to allow for the rangefinder-coupling! – source: fotosaurier

However, the 50mm f/0.95 lens was also released in a version for video cameras, with an additional engravure „TV“ on the nameplate: consequently these lenses were delivered with C-mount. As these lenses do not need the rangefinder-coupling, the rear lens is not cut at the edge here:

Thank you, foto-friend Thomas, for the 50mm f/0.95 TV-lens rear section photo.

4. Finally: Resolution-Data of these Lenses, measured for the Cine Super35-format, which the Angénieux and CZJ Biotar Lenses are originally dedicated to – on all three lenses:

a) Angénieux 50mm f/0.95 – at Super35-format:

Fig. 16: Angénieux 50mm f/0.95 at Super35-frame on Sony A7R4 – absolutely phantastic for this „first-in-speed“ – source: fotosaurier

b) Biotar 50mm f/1.4 – at Super35-format:

Fig. 17: CZJ Biotar 50mm f/1.4 at Super35-frame on Sony A7R4 – absolutely phantastic at one stop slower – source: fotosaurier

c) Canon 50mm f/0.95 – at Super35-format on Sony A7R4:

Canon lens f=50 mm f:0.95_A7R4_Super35_Graph — **Fig. 18**: **Canon Rangefinder 50mm f/0.95** – primarily dedicated to still-photo 24×36 but also delivered as a TV-version – just a bit better than the Angenieux, but 15 years later! – source: fotosaurier

All three lenses have very low chromatic aberrations in the center but the Canon peaks out in maximum CA, Biotar and Canon are close to zero in distortion, while the Angenieux has around -1% distortion, which is still excellent for such an early, extreme lens!

5. Appendix:

Here you see all properties of the three lenses in detail – for 24×36 (full frame) and Super 35 (cine-format).

5-a1. Angenieux M1 50mm f/0.95 – FullFormat 24×36.

5-a2. Angenieux M1 50mm f/0.95 – Super35.

5-b1. Carl Zeiss Jena Biotar 50mm f/1.4 – FullFormat 24×36.

5-b2. Carl Zeiss Jena Biotar 50mm f/1.4 – Cine35.

5-c1. Canon Rangefinder 50mm f/0.95 – FullFormat 24×36.

5-c2. Canon Rangefinder 50mm f/0.95 – Cine35.

Herbert Börger

Berlin, 24.12.2022

Long Telephoto-Lenses and Temperature

Would you expect, that the optical performance of your photographic lenses can be seriously influenced by the operating temperature? Have you ever realized lack of sharpness in extreme environmental temperature conditions?

The simple answer is, of course, that within the specifications for use, given by the makers, there should be no such concern. But it is not that simple.

For amateur astronomers with their mostly very long telescope-focal-length optics (mirror or lens) this fact is very common:

before using the instrument in the clear and mostly cold winter-nights, you have to put the telescope early enough outside (shielded against due) to bring it into a thermal equilibrium with the ambient air at the time you start your observations. The reason: during essential temperature-changes of the optical components (mirrors, lenses) and their mounting devices, their surface-shapes and adjustment change and destroy the extremly precise optical alignment – until the thermal equilibrium is restored. The refractor-lenses may be mounted to allow for some thermal differences, but large mirrors have to be mounted and adjusted extremely precise, so that the cooling-down of the mount, that holds the mirror, may even generate mechanical tension on the mirror – and that generates optical distortions! So we should remind: the absolute temperatures are not the problem – but the thermal transition stages from warm to cold or opposite way!

This fact is also an important design aspect for telescopes: the preferred structure is „as open as possible“ to allow the air to circulate and to generate a good heat-exchange with the internal telescope structure to speed up this process. While the air gets colder during the night, the instrument’s optics can follow close enough to keep the temperature difference low.

There is an impressive document in the archives of the Mt. Wilson Observatory (near L.A., USA) describing the „first-light“-moment of the new 2,5 meter mirror telescope (Hooker-Telescope) on November 1, 1917 – use this link to the adventurous story! („First light“ is the moment, when somebody looks through the finished instrument for the first time.) Here the first-light moment at Mt. Wilson is described near the end of the long text in this link and shows, what a three hour cool-down time made to the optical properties of the 2.5 meter mirror, (which was made by George Willis Ritchey – and allowed for the detection of the expansion of the Universe by Edwin Hubble shortly after taking this telescope into service.).

Picture 1: 2,5 m (100 inch) Hooker-telescope on Mt. Wilson: just struts hold the mirrors to ease the circulation of air for for a fast achievement of temperature equilibrium – source: Ken Spencer, CC BY-SA 3.0 <https://creativecommons.org/licenses/by-sa/3.0>, via Wikimedia Commons

Many instruments in astronomy are closed assemblies, using a corrector-plate (Schmidt-system) or meniscus-lens (Maksutov-System) in the entry of the tube and the mirror at the rear-end (catadioptric telescope – see also my specific blog-article here.) The big disadvantage of these closed systems is the „inertia“ in cooling down due to the closed volume in the telescope tube. Therefore often slits around correctors and mirrors are placed, which allow for sufficient circulation of air through the tube – and even active ventilation is used to shorten the period to reach equilibrium. In some big modern telescopes, the mirror may even be actively temperature-controlled.

Picture 2: „Closed“-tube optical system Maksutov-Cassegrain-Teleskop – source: Wikipedia – Author: Halfblue – http://creativecommons.org/licenses/by-sa/3.0/.

Long telephoto-lenses for normal photography can not be open systems, because the lens-barrels definitely have to be tightly sealed to avoid the invasion of dust, humidity or corrosive gases.

This means, that you have to plan and prepare carefully to bring your equipment to ambient temperatueres in time to avoid these thermal problems. For photographic equipment this would equally refer to the situation, when you come from climate-controlled environment (e.g. hotels) into wery hot (and humid) areas. There is an additional problem, that in bringing cold equipment into hot-humid environment, there might be condensation of humidity on the lenses/mirrors.

This problem is even more delicate with catadioptric lenses (mirror/lens-systems often called just „mirror-lenses“ – in German „Spiegel-Objektive“). In these the surface-shape of the mirrors and the adjustment from mirror to mirror is extremely sensitive for the optical performance of the lens-systems.

I have to-date not realized this with focal lengths of up to 350 mm (though it might be also there to a certain dergree) – but this is definitely an important aspect for focal lengths between 500 mm and 1,000 mm or longer.

From which focal length on these problems may occur, will mainly depend of the type of optical system – and of course the resolution of your cameras sensor!

Here I want to show you this effect with an example of a catadioptric lens of 800 mm focal length: the Vivitar Series 1 Solid Catadioptric 800mm f/11, used on the Sony A7Rm4 (60,3 MP, 35mm format – 3.77 µm pixel-pitch).

Picture 3: Vivitar Series 1 Solid Catadioptric 800mm f/11 – source: fotosaurier

It was the first day this year with just sligtly above zero outside temperature (+2 degree Celsius) and very clear air. At ca. 1:15 p.m.I set out the 800mm f/11 lens on the tripod on the balcony and tried to focus on my favorite landscape test target: a roof-top at about 40 m distance.

The advantage of this target is, that it has large AND fine details, low contrast AND high contrast areas and – most important – a sufficient depth, so that I can detect focusing errors very well!

Picture 4: Overview picture – complete field of view of the „roof-top“ landscape target in ca. 40 m distance taken with Sony A7Rm4 and Vivitar Series 1 Solid Cat 800mm f/11 – this is the „sharp“ picture after the cool-down period of the lens – source: fotosaurier

It was nearly impossible to meet the positive focus position – so I did the best guess and made the photo – and here is the 100%-crop around the focus-position, which is the first steel spring at the right side of the roof edge:

Picture 5: The 67% detail of the focus-area (clamp and spiral-spring!) made 15 minutes after setting the lens outside. Best guess of focus, however, you will find no sharper point in front or behind – the distance scale on the lens says 50 meters in this non-equilibrium temperature situation – source: fotosaurier

At this point of time the lens internally is still on room temperature of about 21 degrees … starting to cool down for about 15 minutes, which it took me to set everything up and focus carefully – but desperately, becaus no really sharp focus was seen in high viewing-magnification.

I had focused using the maximum viewfinder enlagement in the Sony camera and was sure: this is not a really sharp picture. But I could not find a better focus. Picture 5 is a 67% crop of the image taken. And as the subject has some depth: no – there is no better focus to be seen on this picture in front or behind the plane of the spring.

I left the lens with camera in this position for three hours and refocused the lens: now I experienced a quite snappy focus – and you can see the same crop-area here:

Picture 6: The 67% detail of the focus-area (refocused!) after additional 3 hours of the lens outside – source: fotosaurier

The gain in sharpness is damatical – and it exists over the whole field of view, not only in the plane of focus! Also out-of-focus areas show higher contrast now.

However, it connot be ignored, that this catadioptric lens in this picture does by far not use the potential 3,168 Line-Pairs per Picture Height Nyquist frequency of the cameras sensor. My estimate is, that we have here an MTF30 of about 1,100-1,200 LP/PH. So either the three hours of cool-down time were not yet sufficient – or the lens may be not better than this.

(The 1,200 LP/PH MTF30-resolution would correspond to 100 Lines/mm in older „analog“ data. Very good CATs in the 1970s had center-resolutions (measured on film) between 50 and 60 Lines/mm. This relation makes sense, as the difference (factor 0.6 lower for film!) may be owed to the effect of grain and the thickness of the emulsion.)

The „Solid Cat“ 800mm f/11 is a massiv piece of optics – the lens barrel is nearly completely filled with glass, as you see in the lens-scheme:

Picture 7: Lens-scheme of the Vivitar Series1 Solid Cat – source: Perkin Elmer Patent application

It is an absolutly unusual mass of glass – so I would not exclude, that the cooling time should even be longer to reach the thermal equilibrium. My plan is, to make a sequence of photos taken in shorter intervals and over a longer time – as soon as the outside temperatures go down again.

I am not so happy with the fact, that I had to use landscape-scene-shots to demonstrate the performance of the lens, however, for 800mm focal length my IMATEST testing-arena is too short. Maybe I will make a parallel IMATEST-trial then with a 500mm CAT.

So, please, consider this as a first teaser for the topic which has shown clearly, that photographic lens performance may seriously suffer during the time, a lens is undergoing strong temperature-change and before equilibrium is reached.

I promise to come back with a more elaborate research-plan soon.

Herbert Börger

Berlin, December 4th, 2020

Aphorism of the day: Scientific research is most successfull, when it brings up more new questions than it has answered. (fotosaurier)

Die Rand-/Ecken-Auflösung historischer SLR-Objektive (Test-Targets)

Beim „Neustart“ der Foto-Objektiv-Produktion direkt nach dem 2. Weltkrieg lag die Rand-/Ecken-Auflösung typischer Objektive für das Kleinbildformat im Bereich von 300 … 400 … 500…600 Linienpaaren je Bildhöhe von 24 mm (entsprechend ca. 25 … 32 … 40 … 50 Linien/mm), während diese Objektive in der Bildmitte (auch bei Offenblende) über 3.000 LP/PH liefern können. („Bildhöhe“ engl. „picture height“ – daher LP/PH in der IMATEST-Software!) Bei den damals neuen Retrofokus-Weitwinkelobjektiven konnten bei offener Blende die Auflösungswerte in den Ecken auch bei 200 LP/PH oder darunter liegen (entspr. 17 Linien/mm).

Das sind nüchterne Zahlen – der Fotograf „denkt“ aber in Bildstrukturen! Ihn interessiert, was er SIEHT.

Was bedeutet dieser Auflösungsabfall von der Bildmitte zu Rand/Ecke für die praktische Fotografie?

Zunächst möchte ich dieser Frage an reproduzuierbar verfügbaren ebenen Bildstrukturen in einem Testbild für Auflösungsmessungen nachgehen, in dem man außer dem allgemeinen Schärfeeindruck auch Erscheinungen wie (Rest-)Astigmatismus und Farbfehler beurteilen kann.

40 L/mm am Rand galten bei Fotoobjektiven der 1950/60er Jahre bereits als „sehr gut“. In den 50er Jahren erreichten Objektive nach den Stand der Technik am Rand ganz selten Werte über 50 … 60 Linien/mm nach den damaligen Tests auf üblichen, feinkörnigem und normal bildgebenden Filmemulsionen, wie sie auch vom Normal-Fotografen verwendet wurden. In der Bildmitte gemessen erreichte die „analoge“ Kombination Objektiv/Film selten Werte oberhalb 90 L/mm. Auf Spezial-Platten mit hoch-auflösenden Emulsionen – ausgewertet unter dem Mikroskop – konnte man aber auch damals durchaus bis zu 500 Linien/mm messen, was „digital“ 6.000 LP/BH entsprechen würde.

Der Bild-Sensor in der hier verwendeten Sony A7Rm4 erreicht 3.184 LP/PH (60,2 MP).

Schon in den ersten 25 Jahren des 20.Jh. konnte mit den ausgereiften Anastigmaten in der Bildmitte („axial“) praktisch „beliebig hohe“ Auflösungen erreicht werden und es standen dafür auch geeignete Glassorten zur Verfügung. Man betrachte die mit IMATEST ermittelte Auflösungskurve (über dem Bildradius aufgetragen) des 1923er Ernostar 100mm f2.0 bei nahezu voller Öffnung (f2.8) an der 60MP-Sony-Kamera:

Ernostar100f2_2,8_Vgl — **Bild 1: Kantenprofil, MTF-Kurve in der Bildmitte und Auflösung (LP/BH) über Bildfeld des Ernostar 100 f2.0 bei Blende 2.8**

Es ist ein 4-Linser mit vier einzel stehenden Linsen – ohne Vergütung! Dafür erscheint Kantenprofil und MTF-Kurve sehr gut. Aber die Auflösungskurve über dem Abstand von der Bildmitte (100% auf der Abszisse entsprechen einem Bildkreis von 21,5mm Radius!) zeigt einen beängstigenden „Absturz“ von über 2.600 LP/BH auf ca. 300 LP/PH an Rand/Ecken!

Hier die Situation dreißig Jahre später – dazwischen liegt der 2. Weltkrieg:

Ang90f2.8_Vgl — Bild 2: Angénieux 90mm f2.5 von 1951 – Auflösung Rand/Ecken liegt bei 400/600 LP/PH – **bei f2,5 – immerhin leicht verbessert**

Die deutlich größere Verbesserung gegenüber dem Ernostar zeigt sich erst abgeblendet:

Ernostar100f2+Ang90f2,5_f11_Lens_MTF — **Bild 3: Ernostar 100f2.0 (links) und Angénieux 90f2.5 (rechts), jeweils abgeblendet auf Blende 11 (optimale Blende)**

zwar hat sich das Ernostar noch einmal auf olympische 3.000 LP/PH in der Mitte gesteigert (was 93% der Nyquist-Frequenz der verwendeten Kamera entspricht!) aber am Rand bleibt es bei 700-800 LP/PH (allerdings: immerhin verdoppelt).

Das Angénieux 90mm f2,5 erreicht nun aber über die gesamte Bildfläche gemittelt 2.789 LP/PH.

Machen wir noch einmal einen Sprung 30 Jahre weiter in das Jahr 1987. Die Entwicklung neuer, leistungsfähiger Glastypen hat nun weltweit neue Voraussetzungen geschaffen und war die Voraussetzung für das folgende typische Ergebnis am Beispiel einer anderen Optik-Legende:

Apo-Macro-Elmarit100f2,8_f2,8_Vgl — Bild 4: Leitz Apo-Macro-Elmarit 100mm f2.8 volle Öffnung Blende 2.8 – die extrem nach unten streuenden Messpunkte im rechten Bild stammen von der linken-unteren Ecke des Bildes, in der die Auflösung lokal dramatisch abfällt – die Ursache kenne ich nicht (ein Leitz Apo sollte eigentlich keinen so großen Zentrierfehler haben…).

Dank der neuen Gläser ist das Apo-Macro-Elmarit nun „offenblendentauglich“ – obwohl Kantenprofil und MTF-Kurve in der Bildmitte sehr ähnlich den Kurven des über 60 Jahre älteren Ernostar 100mm f2,0 sind! Abgeblendet, bei optimaler Blende (5,6) ist der Mittelwert der Auflösung über das gesamte Bildfeld des Apo-Macro-Elmarit (2.907) dann gerade mal 120 LP/PH höher als der Wert des „ollen“ Angénieux – und die Maximal-Auflösung des Apo-Macro-Elmarit in der Bildmitte ist abgeblendet nicht höher als beim Ernostar ….

Noch eine für seine Entstehungszeit sehr bemerkenswerte Eigenschaft des Angénieux 90mm f2.5 sticht hervor – der sehr niedrige Farb-Fehler (CA):

Angén90f2,5_f11+Apo-Macro-Elmarit100f2,8_Radial_Vgl Kopie — **Bild 5: Achtung: unterschiedliche Nullpunktlage und Maßstäbe in den Ordinaten!**

Auf sehr geringen Niveau ähnlich Apo-Macro-Elmarit bei blau, dreifach so groß bei rot! Aber immer noch ein Drittel vom Contarex-Sonnar 85mm – zehn Jahre später. Einen Kompromiss musste Angénieux aber seinerzeit offensichtlich eingehen, um das zu erreichen: eine relativ hohe Verzeichnung von -1,2% gegenüber +0,4 beim Ernostar und +0,17 beim Apo-Macro-Elmarit.

Man kann also sagen:

der Fortschritt in der optischen Technologie lieferte für die Foto-Objektive überwiegend verbesserte Randauflösung bei Offenblende bei gleichzeitig verbesserter Farbkorrektur, Verzeichnung und erhöhtem Kontrast und verbesserter Streulichtresistenz bei niedrigen Frequenzen – letzteres nicht zuletzt durch die dramatisch verbesserte Beschichtungs-Technologie.

In diesem Link finden Sie Vergleiche des Angénieux 90mm mit weiteren Objektiven über den gesamten Zeitraum 1923 – 2015.

Ich schließe aus meinen vielen Messungen an historischen Objektiven aller Epochen, dass man ab Anfang der 1970er Jahre, den extremen Randabfall der Objektive bei Offenblende schrittweise reduzieren konnte – bereits 1977 gibt es ein Beispiel eines quasi „Ideal-Objektivs“ im Bereich Kurztele (Porträt): das VivitarSerie1 90mm f2,5 Macro! (Mit Einschränkung bei der Streulichtfestigkeit…)

Bei wesentlich größeren Bildwinkeln war das natürlich wesentlich schwieriger und gelang bei Weitwinkelobjektiven entsprechend später mit immer höher- und niedriger-brechenden Gläsern – und im Extremfall (großer Bildwinkel und hohe Lichtstärke) zuletzt erst mit dem Einsatz asphärischer Linsen.

Was bedeuten aber nun die niedrigen Rand-Ecken-Auflösungen bei den frühen historischen Optiken in den Bildstrukturen?

Fangen wir mit einer reproduzierbar beleuchteten, ebenen Objekt-Situation an, in der wir auch diese Auflösungswerte messen: dem detailreichen Test-Chart, das wir abfotografieren. Die Beschreibung der Testmethode finden Sie in diesem Link.

Das ist das Test-Bild, hier durch das Angénieux 90mm f2.5 bei voller Öffnung fotografiert.

#TestChart_Angén90f2,5_f2,5 — **Bild 6: Imatest-Test-Chart SFRplus, fotografiert im Kleinbild-Format 3:2**

Der Abstand zwischen den oberen und unteren schwarzen Balken ist 783 mm im Original.

Die Analyse-Software von IMATEST verwendet übrigens nicht die kleinen Rosetten, die in die dunklen Quadrate eingebettet sind, sondern die Seitenkante der Quadrate, die um 5.71° VERDREHT sind. Mehr erfahren Sie in dem oben aufgeführten Link.

Das Übersichts-Bild soll Ihnen ein Gefühl davon vermitteln, wie fein die Rosetten-Details sind, wenn man ein Bild im normalen Betrachtungsabstand ansieht.

Hier das Detail eines Quadrates mit Rosette in einer Größe, die der Betrachtung des mit der 60MP-Kamera aufgenommenen Bildes bei „100%-Betrachtungsmaßstab“ entsprechen würde (d.h. 1 Bildschirmpixel entspricht 1 Kamerapixel) – wenn Sie das Quadrat auf Ihrem Bildschirm mit ca. 22cm Kantenlänge sehen.

Dies ist das Quadrat genau im Zentrum:

#TargetCenter_Angén90f2,5_f2,5 — Bild 7: Zentrales Target-Quadrat, 100%-Ansicht (966 x 966 Pixel) Angenieux 90mm f2.5 bei Blende 2.5 – laut Analyse beträgt die Auflösung des Objektivs hier 2.500 – 2.700 LP/PH (sagittal/meridional) – 100%-Ansicht bei 60 MP!

Folgend nun der entsprechende Ausschnitt in der oberen-rechten Ecke (wegen der sichtbaren Verzeichnung von -1,2% sind die Qadrate in der Mitte und in der Ecke nicht genau gleich groß!):

#TargerCornerUR_Angén90f2,5_f2,5 — **Bild 8: Target Nr.3 (obere rechte Ecke),, 100%-Ansicht (966 x 966 Pixel) Angenieux 90mm f2.5 bei Blende 2.5 – laut Analyse beträgt die Auflösung des Objektivs hier im Mittel 560 LP/PH**

Die Vignettierung (im Mittel über alle Ecken 2 f-stops) hat hier natürlich noch einen bedeutenden Einfluss auf das visuelle Betrachtungsergebnis! Es fällt allerdings sofort auf, dass trotz der hohen Vergrößerung fast keine Farbsäume zu sehen sind – allenfalls ein sehr kleiner roter Schimmer, wie vom CA-Diagramm zu erwarten ist.

Das folgende Bild zeigt dasselbe Detail, auf das ich nun die Vignettierungs-Korrektur von ca. zwei Blendenwerten angewendet habe, wie man Sie mit Photoshop oder als kamerainterne Korrekturmaßnahme durführen könnte:

#TargerCornerUR_corr_Angén90f2,5_f2,5 — Bild 9: Target Nr.3 (obere rechte Ecke), 100%-Ansicht (966 x 966 Pixel) Angenieux 90mm f2.5 bei Blende 2.5 – Vignettierung kompensiert. **Meridional ca. 400, sagittal ca. 600 LP/PH**

Hier erkennt man drei Dinge:

Die 560 LP/PH-Auflösung liefern tatsächlich noch klare Bildstrukturen – wenn auch „weicher“
Die Farbreinheit der Abbildung bestätigt sich – allerdings erkennt man einen leichten generellen Gelbstich hier in der Bildecke
Man erkennt sogar den Unterschied zwischen ca. 400 LP (meridional) und ca. 600 LP (sagittal) in den Rosetten-Details: die Ringe sind in der Bild-Diagonale von links oben nach rechts unten erkennbar „kantenschärfer“!

Die Struktur ist „weicher“ wiedergegeben – aber dennoch deutlich und mit gutem Kontrast sichtbar.

Beachten Sie bei diesen Bildern bitte: es handelt sich um die 100%-Darstellung des 60 MP-Bildes!

Anmerkung: In Imatest-Diagrammen wird der angelsächsischen Nomenklatur folgend „meridional“ meist als „tangential“ bezeichnet (tangential = meridional) diese Kuven sind durchgehend gezeichnet, die sagittale Auflösungskurve gestrichelt. In MTF-Diagrammen der Fa. Zeiss ist die Zuordnung umgekehrt: gestrichelt meridional und durchgezogen für sagittal

Kritischer ist diese Situation bei Weitwinkel-Objektiven, bei denen Farblängsfehler und Astigmatismus an Rändern und Ecken eine deutlich größere Rolle (wegen der viel größeren off-axis-Winkeln) spielen.

Wir betrachten das folgend an von 24/25mm-Retrofokus-Objektiven „der ersten Stunde“ (1957/71):

Angénieux wahrte seinen zeitlichen Vorsprung konsequent und brachte seine „Retrofocus“-Weitwinkel-Brennweiten in schneller Folge auf den Markt: 35mm f2.5 in 1950 (6-Linser) vorgestellt und in kleinen Mengen geliefert (ab 1953 Großserie!), 28mm f3.5 (6-Linser) ebenfalls ab 1953, 24mm f3.5 (8-Linser) ab 1957. (Besonderheit: danach wurde von Angénieux niemals wieder eine Neuberechnung dieser Foto-Optiken herausgebracht sondern diese Optiken bis 1971 unverändert geliefert und das Segment der Festbrennweiten dann völlig eingestellt.

Bei diesen frühen Weitwinkel-Objektiven ist bei Offenblende die Auflösung noch deutlich niedriger als bei dem 90er Objektiv. Bei dem Angénieux Retrofocus 24mm f3.5 liegt die Auflösung in den Ecken bei 310-354 LP/PH (sagittal) und ca. 600 LP/PH (meridional) bei den Einzelwerten – der Ecken-Mittelwert beträgt 328 LP/PH:

Angén24f3,5_Offen_sagittal — Bild 10: Angénieux 24mm f3.5 bei Offenblende – Auflösung über Bildfeld der sagittalen Strahlenbündel

Sehen wir uns das Target Nr.5 in der rechten unteren Ecke an (sagittal mit 345 LP/PH gemessen – meridional mit 560 LP/PH):

#Target RU_Angén24f3,5_f3,5 — **Bild 11: Angénieux 24mm f3.5 bei Offenblende f3.5** – Target Nr. 5 – rechte untere Ecke (Vignettierung kompensiert) – sagittal 345 LP/PH – meridional 560 LP/PH

Trotz der deutlichen Rest-Fehler ist die Struktur noch deutlich erkennbar, wenn auch richtungsabhängig. Der sagittale Wert entspricht 29 L/mm. Die visuelle Auswirkung des Farbfehlers ist – trotz des hohen CA von 8 Pixel! – auf die Farbsäume begrenzt.

Das Nachbar-Target (Nr. 21) links davon hat 500 LP/PH sagittal und 502 LP/BH meridional – also frei von Astigmatismus, aber mit CA von ca. 4,5 Pixeln:

#Target21_corr_Angén24f3,5_f3,5 — **Bild 12: Angénieux 24mm f3.5 bei Offenblende f3.5** – Target Nr. 21 – links von der rechten unteren Ecke (Vignettierung kompensiert) – sagittal 500 LP/PH – meridional 502 LP/PH

Folgend sehen wir das entsprechende Auflösungs-Diagramm des Zeiss Jena Flektogon 25mm f4.0 (1959):

Flektogon25f4,0_f4,0_Multi-ROI — Bild 13: Flektogon 25mm f4.0 bei Offenblende – Auflösung über Bildfeld der sagittalen Strahlenbündel

Angesichts des in den Ecken „noch“ bei 301 LP/PH liegenden Mittelwertes (gilt für sagittale und meridionale Strahlen) liegen hier die sagittalen Einzelwerte Rand/Ecken bei erschreckend niedrigen 104 – 222 LP/PH.

Sehen wir uns den Linken Rand (Mitte) mit sagittal 222 LP/PH / meridional 610 LP/PH an (Target-Nr.10):

#Target LRmitte10_corr_Flektogon25f4,0_f4,0 — **Bild 14: Flektogon 25mm f4.0 bei Offenblende f4.0** – Target Nr. 10 – linker Rand, Mitte (Vignettierung kompensiert) – sagittal 222 LP/PH – meridional 610 LP/PH

Hier ist die Struktur schon sehr weich aber deutlich zu erkennen – kräftiger Rest-Astigmatismus, aber sehr geringer Farbfehler. Es ist schwer zu sagen, wie diese Situation analog auf Film gemessen worden wäre: 222 LP/PH entsprächen 18,5 Linien/mm… das wäre wohl nicht mehr als gut bewertet worden.

Nur wenige mm weiter nach außen am Target 17 (rechter Rand ein Taget nach unten) liegt die Auflösung bei sagittal 160 LP/PH und meridional bei 591 LP/PH:

#TargetNr17_corr_Flektogon25f4,0_f4,0 — Bild 15: **Flektogon 25mm f4.0 bei Offenblende f4.0** – Target Nr. 17 – rechter Rand, eins unter Mitte (Vignettierung kompensiert) – **sagittal 160 LP/PH** – meridional 591 LP/PH

Hier bricht im sagittalen Sektor der Struktur der Kontrast endgültig ein – fast schon verschwommen und man erkennt, dass noch weiter rechts am äußersten Rand (es fehlen noch 4mm bis zum Rand) der Kontrast noch einmal dramatisch absinken wird.

In der Ecke oben rechts (Target Nr. 3) mit 104 LP/PH sagittal, 338 LP/PH meridional:

#Target3-UR_corr_Flektogon25f4,0_f4,0 — Bild 16: **Flektogon 25mm f4.0 bei Offenblende f4.0** – Target Nr. 3 – Ecke oben rechts (Vignettierung kompensiert) – **sagittal 104 LP/PH** – meridional 338 LP/PH

Man kann die Struktur nur noch erahnen – die extrem niedrige sagittale Auflösung und der hohe Rest-Astigmatismus lösen die Bildstruktur auf – obwohl die Chromatische Aberration mit ca. 1,6 Pixel nur ein Fünftel der CA bei dem Angénieux 24mm in der Ecke ist.

Betrachten wir im direkten Vergleich das entsprechende Objektiv von Zeiss-West, das 3 Jahre später heraus kam und eine Blende lichtstärker ist – Distagon 25mm f2.8 (für die Contarex 1961):

CtrxDistagon25f2,8_f2,8_Offen_sagittal — Bild 17:

Auch hier liegen die sagittalen Werte am Rand bei Offenblende f2.8 unter 200 LP/PH.

Ich zeige folgend die beiden Targets Nr.10 (linker Rand, mitte) und Nr.5 (rechte untere Ecke):

#TargetNr10_corr_CtrxDistagon25f2,8_f2,8 — **Bild 18: Zeiss Distagon 25mm f2.8 bei Offenblende Target 10 (linker Rand mitte) – Vignettierung korrigiert**

Hier beginnt bei sagittal 195 LP/PH die Bilddefinition sich durch eine Kombination eines starken Rest-Astigmatismus (meridionaler Wert: 917 LP/PH) und des Farbfehlers aufzulösen – der Kontrast ist schwach.

#TargetNr5_corr_CtrxDistagon25f2,8_f2,8 — **Bild 19: Zeiss Distagon 25mm f2.8 bei Offenblende Target 5 (rechte untere Ecke) – Vignettierung korrigiert**

In der Ecke sagittal 185 LP/PH mit starkem Rest-Astigmatismus findet sich nur noch in einem sehr schmalen meridionalen Sektor eine klar definierte Struktur (mit 379 LP/PH) mit niedrigem Kontrast.

In dieser Gruppe der FRÜHEN Retrofocus-Objektive mit 24 oder 25 mm Brennweite (Angénieux, Carl Zeiss Jena Flektogon und Zeiss-Ikon Distagon) gibt es ein viertes (1963) aus Japan: Topcon Topcor 2,5cm f3.5, das unter diesen Optiken herausragt:

Topcor24f3,5_f3,5_Offen_sagittal — **Bild 20: Topcor 2,5cm f3.5 – sagittale Auflösung bei Offenblende im gesamten Bildfeld (443 … 618 LP/BH)**

Der Mittelwert der (sagittalen und meridionalen) Rand-/Ecken-Auflösungswerte beträgt hier 683 LP/PH. Das folgende Bild zeigt die Struktur von Target Nr.5 in der rechten unteren Ecke:

#TargetNr5_corr_Topcor24f3,5_f3,5 — **Bild 21: Topcor 2,5cm f3.5 bei Offenblende, Target Nr.5 – untere rechte Ecke bei sagittal 587 LP/PH (meridional 914 LP/PH) – also mit mäßigem Rest-Astimatismus – Vignettierung korrigiert**

Bei diesem Auflösungs-Niveau (mit mäßigem Astigmatismus und geringem Farbfehler (CA-Wert in der Ecke 1,5 Pixel!) liegt nun eine klare Bildstruktur vor – nur deutlich weicher als im Bildzentrum.

Dieses Objektiv ragt damit in der Bildqualität deutlich aus dem Feld der zeitgenössischen „Superweitwinkel“ zwischen 1957 und 1963 hervor.

Sehen wir uns noch den nächsten Qualitäts-Schritt am Beispiel des Minolta MD W-Rokkor 24mm f2.8 an:

#TargetNr5_corr_MD24f2,8_f2,8 — **Minolta MD W-Rokkor 24mm f2.8 Offenblende f2.8 – Target Nr.5 (untere rechte Ecke) – CA mit 3 Pixel deutlicher als beim Topcor – Vignettierung korrigiert**

Der Kontrast liegt hier deutlich höher mit einem Durchschnittswert der Auflösung Rand/Ecken von 1002 LP/PH.

Schließlich die gegenwärtige moderne Referenz – das Zeiss Distagon 25mm f2.0:

#TargetNr3_Batis25f2,0_f2,0 — **Bild 23: Zeiss Distagon 24mm f2.0 Offenblende f2.0 – Target Nr.3 (obere rechte Ecke) – sagittal 1.206 , meridional 1.897 LP/PH und CA von 0.5 Pixeln**

Das Objektiv ist mit der Auflösung bei Blende 2.0 in der Ecke mit durchschnittlich 1.517 LP/PH visuell kaum noch von der Bildmitte zu unterscheiden (Vignettierung auch hier korrigiert!).

Man sieht an diesen Beispielen deutlich, dass außer dem meßtechnischen Wert der Auflösung die Rest-Bildfehler die visuelle Wirkung wesentlich mit beeinflusst. Wobei man den Eindruck hat, dass ein größerer Farbfehler sich ggf. weniger zerstörerisch auf den Bildkontrast auswirkt als ein starker Rest-Astigmatismus.

FAZIT:

Man sieht, dass 200-300 LP/PH als Untergrenze einer bildgebend noch brauchbaren Auflösung gelten können (s. Bild 14), wenn Rest-Astigmatismus und Farbfehler im mäßigen Grenzen bleiben. Der absolute Auflösungswert entscheidet in diesem Bereich allerdings nicht alleine über das bildliche Ergebnis. Genauso entscheidend ist der Korrekturzustand – also die anwesenden Rest-Linsen-Fehler. Allgemein sind diese historischen Objektive in der Rand-/Ecken-Auflösung ab ca. 400 – 600 LP/PH als gut zu bezeichnen (s. Bilder 11, 12 und 21) – mit gewissen Abstrichen beim Kontrast.

Ab Anfang der 1970er Jahre werden Auflösungs-Werte in den Ecken um 1.000 LP/PH bei Offenblende auch bei Weitwinkelobjektiven erreicht, womit zumindest in der Analog-Fotografie hervorragende Ergebnsise möglich waren.

Moderne Objektive erreichen dank asphärischer Linsenflächen hervorragend ausgeglichene Ergebnsise auch bei Offenblende über das gesamte Bildfeld – auch bei sehr großen Bildwinkeln (s. Bild 23).

Fotosauriers optisches Testverfahren für Objektive mit IMATEST

Ich messe die optische Qualität von Objektiven mit Hilfe des IMATEST-Verfahrens. (Imatest ist eine 2004 in Boulder, Colorado, USA gegründete Firma.)

Das durch das Objektiv mit der Digitalkamera aufgenommene Testbild (Target) stellt eine Datei dar (Bild-Daten + Exif-Datei). Diese Datei wird mittels einer (kostenpflichtigen) IMATEST-Software analysiert (IMATEST-Studio oder IMATEST-Master). Die Analyse liefert – abhängig von der Art des Targets – eine ganze Reihe von optischen Prüfergebnissen, die letztlich alle auf der MTF-Kurve basieren.

Das Basis-Verfahren wird Imatest SFR genannt (Imatest spatial frequency response), was man allgemein als „Modulation Transfer Function“ (MTF) bezeichnet. Analysiert wird eine Hell-Dunkel-Kante, die Imatest als „clean, sharp, straight black-to-white or dark-to-light edge“. Die hellen und dunklen Flächen, die an die Hell-Dunkel-Kante angrenzen müssen sehr gleichmäßigen (konstanten) Helligkeitsverlauf besitzen. Der Analyse-Algorithmus basiert auf dem Matlab-Programm „sfrmat“. Im Prinzip ließe sich dafür jede beliebige scharfe Kante verwenden. Imatest empfiehlt und verwendet eine Kante unter 5.71° Neigung und einem Kontrast von 4:1, da dies die am besten reproduzierbaren Ergebnisse liefert:

SlantedEdge — Analysefelder an einer als „slanted-edge“ bezeichneten Hell-Dunkel-Kante, Neigung 5.71°, Kontrast 4:1 horizontal (links)- vertikal (rechts)

Es versteht sich, dass die grafische Qualität dieses Testbildes/Test-Charts eine wichtige Rolle bezüglich der Reproduzierbarkeit von damit erzielten Prüfergebnissen spielt. Deshalb habe ich mir die große Test-Chart „SFRplus 5×9“ von Imatest aus USA liefern lassen (sie kostet derzeit $430,00). Der Abstand zwischen dem oberen und unteren schwarzen Balken beträgt 783 mm – die Gesamtbreite ca. 1.600 mm:

Die SFR-Messung erfolgt hier, wie vorstehend schon beschrieben, nicht etwa an den kleinen radialen Rosetten, die in die Quadrate eingebettet sind, sondern an den horizontalen und vertikalen Kanten der um 5.71° gedrehten grauen Quadrate.

Das Testbild kommt als eingerollter Druck und muss noch auf eine perfekt ebene, stabile, dauerhafte Unterlage aufgeklebt werden. Das habe ich von einem professionellen Laminier-Betrieb auf dem stabilsten Sandwich-Trägermaterial erledigen lassen ((Blasen/Falten würden das Testbild unbrauchbar machen!). Dazu habe ich auf der Rückseite zwei Al-Profile zur Versteifung und Wandmontage aufkleben lassen. Die genau vertikale und verdrehungsfreie Wandmontag habe ich mit einem Kreuzlaser unterstützt vorgenommen.

Eine typische Aufnahme dieses Testbildes durch das zu untersuchende Objektiv mit der Digitalkamera sollte so aussehen:

IMATEST stellt folgende Check-Liste für die Arbeit mit der Test-Chart auf:

IMATEST - hohe Abforderungen — **„Checklist“ für das reproduzierbare Arbeiten mit dem Imatest-Verfahren**

Vieles ist da zu beachten – und darüberhinaus entdeckt man in der praktischen Ausführung noch eine Menge Details, die einem eine sehr hohe Konzentration abfordern… zum Beispiel die Ausleuchtung:

LED-Lampen! … aber bitte nicht von Akkus gespeist – da ändert sich gegen Ende der Akku-Laufzeit die Beleuchtungsstärke. Unbeding Beleuchtungsintensität messen!

Bezüglich des Arbeitsabstandes als Funktion der Pixel-Anzahl der Kamera gilt, dass die große SFRplus TestChart für die 60 MP der Sony A7Rm4, die ich einsetze, gerade ausreichend ist.

Es kann im Prinzip jeder machen, der eine hohe Motivation dazu hat – aber es ist von äußerst großem Nutzen, wenn man viel von Optik und Physik versteht … damit man am Ende nicht Hausnummern misst! 😉

Ich werde jetzt nicht mehr in jedes Detail gehen. Natürlich ist die nächste wirklich wichtige Hürde, die man nehmen muss, die Ausrichtung der Kamera/Objektiv-Achse zur Mitte und zur Ebene des Testbildes. (Ich arbeite da mit zwei Kreuz-Lasern.)

Wenn man schließlich alles im Griff hat und man hat korrekte Aufnahme-Dateien des Testbildes erstellt, dann ist der Rest mit der Imatest-Software tatsächlich eine Knopfdruck-Aktion: mit dem oben dargestellten Chart SFRplus definiert das Programm automatisch 46 „ROI“ (region of interest) – also kleine Ausschnitte der „slanted-edges“ wie oben beschrieben – mal horizontal mal vertikal orientiert – und analysiert dann binnen weniger Sekunden die Auflösung an diesen 46 Stellen, die MTF-Kurve, ein (vorher festgelegtes) Kantenprofil und die Auflösungskurve über dem Bildkreisradius (getrennt nach sagittaler und meridionaler Orientierung.

Kantenprofil+MTF-Kurve — **Beispiel einer Kantenprofil/MTF-Auswertung an einer einzelnen ROI-Position (14% rechts vom Bildzentrum)**

Das wird in Graphen oder auch in Tabellenform ausgelesen – bzw. als Datei, mit der man weitere programmierte Auswertungen und Darstellungen durchführen könnte.

MTF30-Auflösungswerte in Linienpaaren je Bildhöhe (60 MP-Sensor!) in den ROI-Positionen mit überwiegend sagittaler Orientierung. Man erkennt, dass die Methode bis sehr weit in die äußerenen Bildecken hinein funktioniert!

Auflösungs-Daten kann man für mehrere MTF-Kontrast-Werte (MTF10, MTF20, MTF30, MTF50) ausgeben lassen. Dann wird neben den Einzelwerten in der obigen grafischen Darstllung auch der gewichtete Mittelwert der (z.B.) MTF30-Auflösung über das GESAMTE Bildfeld, der Mittelwert für die MITTE, der Mittelwert für den Übergangsbereich und der Mitttelwert für die Ecken ausgegeben:

MTF30-Mittelwerte — **Gesamt- (gewichtet!) und Zonen-Mittelwerte aus den Einzelwerten der darüber dargestellten Messung – die Mittelwerte enthalten ALLE sagittalen und meridionalen Meßergebnisse.**

Außer den Auflösungs- und MTF-Daten werden Chromatische Aberration und Verzeichnung ermittelt.

Ich messe stets bei ALLEN Blenden jedes Objektives und definiere als „optimale Blende“ der jeweiligen Optik die mit dem höchsten (gewichteten) Gesamt-Mittelwert der Auflösung über das gesamte Bildfeld. Es kann dabei sein, dass an diesem Blendenwert die maximale Auflösung in der Bildmitte schon überschritten ist, aber die Rand/Ecken-Auflösung noch deutlich steigt.

Zur Charakterisierung einer Optik habe ich mich entschieden, folgende Auflösungswerte anzugeben – und zwar einmal bei Offenblende, einmal bei optimaler Blende:

Mittelwert gesamte Bildfläche (gewichtet mit 1/0.75/0.25)
Mittelwert der Meßpunkte in Bildmitte (bis 30% Bildradius)
Mittelwert der Meßpunkte Rand/Ecken (außerhalb 70& Bildradius)
MTF-Kurve (über der Frequenz aufgetragen)
Kurve der Auflösung über dem Bildradius (Mitte=0 …. Ecke=100)

Außerdem Verzeichnung und CA. In meinen Vergleichstabellen kann das dann so aussehen:

Gelegentlich kann die 3D-Darstellung der Auflösung über der Bildfläche noch zu weiteren Erkenntnissen beitragen. Hier ein Beispiel (dasselbe Objektiv, wie in den anderen Beispielen weiter oben und unten!):

3D-Darstellung der meridionalen (links) und sagittalen (rechts) MTF50-Auflösungswerte

Alle Messungen erfolgen an derselben Digitalkamera Sony A7Rm4 mit 60 MP-Sensor und E-Mount-Objektivanschluß unter stets gleicher Einstellung von Auflösung und kamerainterem RAW-Converter (z.B. Schärfung auf Wert „0“).

Soviel zur konkreten Messung der Qualität der optischen Systeme. Und damit wäre für fabrikneue Objektive an einer Kamera, für die die Optik hergestellt wurd, eigentlich alles gesagt.

Bei meinen Untersuchungen an HISTORISCHEN Objektiven treten allerdings folgende Einflüsse auf:

a) Ich nehme hier die Messungen an historischen Objektiven vor, die bis zu 100 Jahre alt sein können. Die meisten davon sind in einem normalen Abnutzungs- und Alterungs-Zustand, wobei ich festhalten möchte, dass nur Objektive in ein Vergleichsprogramm aufgenommen werden, die keine starken Ablagerungen, Beläge und Separationen an Linsenflächen zeigen, die schon als „Schleier“ in Erscheinung treten. Staubpartikel im Inneren und mäßige Putzspuren sind nicht auszuschließen – aber alle geprüften Optiken erscheinen – auch mit einer LED-Punktlampe durchleuchtet – weitgehend klar! Welchen Einfluss die Alterung und „normale“ Verschmutzung auf die Messergebnisse haben kann ich nicht klären – ich schlage vor, dass man die Ergebnisse pragmatisch eben als das ansieht, was sie sind: nämlich die Eigenschaften (unterschiedlich) gealterter historischer optischer Geräte! Die Ergebnisse liefern allenfalls einen orientierenden Eindruck vom Auslegungs- und Neu-Zustand dieser Objektive. Da die Ergebnisse in vielen Fällen überraschend gut ausfallen, darf man die Dinge auch gerne so bestaunen, wie sie jetzt erscheinen. Ich kann mir kaum vorstellen, dass die Optiken durch die Alterung BESSER geworden sind…

b) Um die Objektive der unterschiedlichsten historischen Kamerasysteme (wie Exakta, Alpa, M42,…) an die Kamera mit E-Mount anzuschließen, wird ein ADAPTER benötigt. Damit tritt ein rein mechanisch-geometrisches Problem im Versuchsaufbau auf: nach meinen bisherigen Erfahrungen ist genau das die zweitgrößte Fehlerquelle bei den Versuchen, über die ich berichte. Weil der Adapter nun ein Bestandteil der Fassung des Objektives ist, verschlechtern sich oft Zentrierung und Ausrichtung der optischen Achse relativ zum digitalen Bildsensor.

Die IMATEST-Software liefert bezüglich dieses Problemes allerdings eine wichtige Hilfestellung:

Analyse der Geometrie der Imatest-Bilddatei: in der untersten Zeile stehen die „Convergence angles“ in horizontaler und Vertikaler Richtung: wenn die Zahlenwerte hier „Null“ sind, ist die Ebene des Testbildes relativ zur Sensor-Ebene ideal parallel ausgerichtet (d.h. die Linien des Rasters schneiden sich im „Unendlichen“. Die Bildmitten müssen sich dann nicht exakt decken (s. dritte Zeile von unten: central square pixel shift).

Man kann dieses Analyse-Ergebnis benutzen, um den Meßaufbau mit dem jeweiligen Adapter optimal auszurichten. Ich habe mir derzeit eine Toleranz von <0.1 Grad bei den Konvergenz-Winkeln gesetzt.

c) Die größte – und leider nicht sicher abzuklärende – Fehlerquelle bei diesen Messungen an historischen Objektiven, die für die Benutzung mit „Analog-Film“ konstruiert wurden, ist die unbekannte Wechselwirkung zwischen Optik und Digital-Sensor („Digital-Tauglichkeit“).

Hier sehe ich aufgrund meiner Erfahrungen drei Haupt-Probleme:

c1) Mögliche Reflexionen zwischen einer oder mehreren Linsenflächen und der Sensoroberfläche. Das kann sich zonenweise als Kontrastminderung auswirken oder auch das Bild ganz gravierend stören. In meinem Blog-Beitrag über das Ernostar 100mm f2.0 habe ich eine solche Erscheinung beschrieben (mit dem 42 MP Sony-Sensor).

Ernostar (die 2.) – Streulicht-Problem auf Anolog-Film?

Dort bildete sich beim Abblenden über f5.6 ein großer, milchig aufgehellter Bereich in der Bildmitte. Am 24 MP-APSC-Sensor in der Fujifilm-X-T2 (bzw. X-Pro2) trat dieselbe Erscheinung nicht auf. Dabei habe ich auch untersucht, dass diese Erscheinung auf Analog-Film bei diesem Ernostar-Objektiv nicht auftrat.

c2) Anti-Aliasing-Filter als zusätzliche optische Elemente können einen nennenswerten Einfluß auf die Bildqualität nehmen. Das ist sehr anschaulich im Artikel von H.H.Nasse unter lenspire.zeiss.com beschrieben.

https://lenspire.zeiss.com/photo/app/uploads/2018/11/Nasse_Objektivnamen_Distagon.pdf

Allerdings besitzt die verwendete Sony A7Rm4 kein Anti-Aliasing Filter, sodass ich nicht davon ausgehe, dass es in meinen Untersuchungen diesen Einfluss gibt.

c3) Hintere Schnittweite (Abstand zwischen hinterstem Linsenscheitel und der Film/Sensor-Ebene) und daraus möglicherweise resultierende sehr flache Einfalls-Winkel der Strahlen auf den Sensor. Was der Film verkraftet (und zwangsweise mit starkem Helligkeitsabfall im Außenbereich des Bildes quittiert = starke Vignettierung) bekommt dem Sensor nicht: es kommt zu schlimmsten Einbrüchen der Auflösung und Farbübertragung! Auch das ist im Nasse-Artikel sehr anschaulich beschrieben!

Diese Erscheinung gilt grundsätzlich für alle (symmetrischen) Weitwinkelobjektive der Brennweite <35mm an Digitalsensoren, also meistens für die Weitwinkelobjektive mit Bildwinkel >70°, die für analoge Meßsucherkameras gebaut wurden. Für Retrofokus-Objektive gilt das nicht.

Ich rechne aber damit, dass es auch noch andere, unbekannte Wechselwirkungen zwischen Analog-Objektiv-Strahlengang und Digitalsensor gibt. Deshalb ist für mich die wichtigste Voraussetzung für die VERGLEICHBARKEIT von Messergebnissen mittels Digitalkamera, dass immer dieselbe Kamera dafür verwendet wird – mit immer gleichen Einstellungen des RAW-Converters.