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.
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).
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:
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:
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!)
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:
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.
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.
In the next picture you see the individual resolution-values in all these ROIs at f/0.95:
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!):
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:
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) :
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:
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:
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:
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.
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.
Happy „fotosaurier“, analog picture taken with Angénieux M1 50mm f/0.95 at f/2.8:
Berlin, March 21, 2024
Copyright Herbert Börger