Saturday, 19 October 2013

Comparing lenses - blur characteristics


The advent of digital cinema cameras has seen a revival of interest in older or "legacy" cine lenses, less than perfect in technical terms but often full of character and sometimes a welcome antidote to overly clinical digital imagery. While filters and manipulation in post can achieve similar ends, a lens creates its own unique effects due to its optical design, and responds to changes in real world lighting organically.

I thought it might be interesting to compare some modern 35mm format cine lenses with a few older examples, by looking at their projection characteristics. A lens test projector is a standard tool in lens repair and manufacturing facilities, it allows a technician to project a test reticle onto a large screen or wall and assess the lens for optical quality, aberrations or mechanical issues, to check the close distance scale and back-focus, and assists in making adjustments like element centering or zoom tracking.

Assessments are always done with a focussed image, but it's also quite interesting to see how lenses transmit light when they're defocused, which is mainly a product of how the aberrations have been corrected in the lens design, modulated by the f-stop, and to a lesser extent, by the shape of the iris aperture. The nature of the aberration correction will cause a different effect depending on whether the out-of-focus area is behind or in front of the plane of focus. The out-of-focus rendering plays an important part in the look of a lens, not just in the often discussed 'bokeh' but also in how sharpness falls off from the plane of best focus.

In the following comparisons, I took pictures of just the lower left quarter of the test image, with the test lens sharply focused (at 5 feet),  and then defocused (with the focus set to around 2 feet). All lenses were between 30 and 38mm in focal length and were set to maximum aperture (between f/1.7 and f/2.1, except the Zeiss Super Speed which opens to f/1.2). The defocused images represent how each lens transmits light from behind the plane of focus, ie the background blur. Disregard the orange dots, which are just horizon and tracking markers on my projection wall. Colour temperature variations and resolution loss are caused by the limitations of my camera (and my photographic incompetence) but I'm only really interested here in showing the overall qualities of contrast and flare and the shape and nature of the blur patterns.

Comparisons



Zeiss Ultra Prime 32mm f/1.7




Zeiss Ultra Prime 32mm f/1.7 defocused


Cooke S4 32mm f/1.8

Cooke S4 32mm f/1.8 defocused


Bausch and Lomb Super Baltar 35mm f/2



Bausch and Lomb Super Baltar 35mm f/2 defocused



Meyer Gorlitz Primoplan 30mm f/1.9


Meyer Gorlitz Primoplan 30mm f/1.9 defocused



Ross XPres 38mm f/1.9


Ross XPres 38mm f/1.9 defocused



Schneider Arriflex-Cine-Xenon 35mm f/2


Schneider Arriflex-Cine-Xenon 35mm f/2 defocused



Zeiss Super Speed 35mm f/1.2


Zeiss Super Speed 35mm f/1.2 defocused



Cooke Speed Panchro 32mm f/2.1


Cooke Speed Panchro 32mm f/2.1 defocused




Screen grabs

And now some screen grabs taken on an Arri Alexa to compare how each lens renders background blur in the real world (or at least out the window of Cameraquip). Each shot was taken at around f/2. Modern lenses like the Ultra Primes and S4s are quite similar, but it's surprising how different some of the older designs look. The Ross XPres came from a 1930s Debrie Parvo, the Meyer Primoplan from a WW2 Arriflex 35, the Schneider Cine-Xenon design dates from the 1950s, the Bausch & Lomb Super Baltar is from a 1960s Mitchell. Along with Baltars and Super Baltars, Cooke Speed Panchros were the most widely used cinema lenses in the decades after WW2. Zeiss Super Speeds became industry standards in the 1980s, Ultra Primes and Cooke S4s arrived in the late 90s.


Zeiss Ultra Prime 32mm

Cooke S4 32mm

Bausch & Lomb Super Baltar 35mm

Meyer Gorlitz Primoplan 30mm

Ross XPres 38mm

Schneider Kreuznach Cine-Xenon 35mm

Zeiss Super Speed 35mm

Cooke Speed Panchro 32mm

Wednesday, 19 June 2013

Inside a Bell & Howell Eyemo







Among wind-up cameras Bell & Howell's 35mm Eyemo (and its 16mm brother the Filmo) stand apart, as products of the highest technical standards. Their simple but precise mechanism, and extremely durable construction made them quickly popular among newsmen and military photographers, who used them to cover everything from the Spanish Civil War right through to Vietnam. In the movie industry Eyemos were often used as "crash cams", where their low cost, stable images and indestructability proved highly valuable for dangerous stunt work.

The Eyemo was first released in 1925, as a single lens 100 foot spool load model, based very much on the 16mm Filmo that had come out a year or so earlier. Despite the quality of its construction, it was intended to serve the amateur or semi-professional market. Later models came with triple lens turrets and viewfinders with their own matching lens turret. The model Q pictured, also nick-named "spyder", had the lenses spaced out to avoid a neighbouring lens hood sticking in to the frame, and came with an additional critical viewfinder and the ability to take magazines. The internal mechanics however remained virtually unchanged since the first model.


A few different Eyemos, including a PL mount "crash cam" front and centre.                                                                                                                                                                                                                          


Because they use an uncased spring, dismantling Eyemos and Filmos can be very dangerous if you don't know what you're doing. The force unleashed by a coiled length of spring steel can be tremendous, even when they have been run right down, and the edges are very sharp. For this reason I would advise inexperienced people not to attempt to completely dismantle one of these cameras. 


With certain precautions it is safe to remove the front housing and access the shutter and intermittent mechanism. Many times this will be the area that is causing a camera to stop running.
Once the turret has been removed, 2 screws on either side of the front plate and one inside the film chamber above the gate can be undone and the whole front mechanism comes out. A stop pawl in the front (activated by the release lever protruding through from the camera side at the bottom of the cavity pictured) is all that is keeping the camera from running, so if the spring is wound at all (and the jam is indeed in the front mechanism), it will start running the moment the front is removed.


So to be safe, before removing anything the spring should be run down as far as it will go. A 13 tooth idler gear visible through a hole in the film chamber controls the duration of the run (the left gear in the picture). It runs directly off a 14 tooth gear connected to the spring shaft, and after 12 revolutions a shallow valley on the main gear hits a long tooth on the idler gear and locks. To completely wind down the spring, the idler gear can be depressed with a screwdriver to allow the spring to keep turning until there is no driving force left. 
If the camera is jammed (as is often the case when a service is required) the spring cannot be run down, so it may fire up when the front is removed. In this case I set the speed to the lowest setting.


The front plate houses the shutter, stop pawl, intermittent mechanism and gate. When removing the gate be careful to note the spring beneath the side rail (at the top of the picture) that applies pressure to the side of the film and controls lateral steadiness.
The pulldown mechanism is a marvel of simple yet effective design. The cam should spin quite freely when the stop pawl (on the left in the picture) is activated. I lightly grease the cam surface, follower, shuttle and shuttle pins. The shutter bearing can be oiled from the other side.
Often the shutter needs some straightening to avoid scraping. To prevent excessive vibration it is made of very lightweight but easily deformed tin plate. 


According to the service manual, any further dismantling requires the spring to be fully wound 
up (with the front fitted) and then run at the lowest speed until the long tooth of the idler gear is visible. It should then be depressed and the camera run again until the shallow valley of the main gear is visible, at which point the two gears can be locked together to prevent any further unwinding. Now the front can come off, the 4 smaller screws around the edge of the top mechanism plate can be undone and the entire mechanism, tightly wound spring and all, can be carefully removed. At this point a special jig to hold the spring needs to be fitted so that it can be detached from the rest of the mechanism. Without a jig do NOT attempt this procedure!



I don't have a jig, and don't really like the idea of removing the spring fully wound, so I take a different tack. With the spring completely wound down (the idler gear must be depressed as described earlier) I remove the curved strip connecting the speed selector to the governor, and the 2 sprockets. Then I remove all of the screws - the 4 smaller ones holding the mechanism into the body, the 6 (or sometimes 5) larger ones holding the two mechanism plates together, and the 'special' screw (with 2 holes rather than a slot) next to the feed spindle that controls the feed tension. If there is any spring tension remaining,  separating the top plate from the bottom plate will cause gears to slip and the spring to suddenly unwind at full speed. It makes a terrifying sound, and will strip the teeth from gears. So you can't do this if the mechanism is jammed and the spring not fully unwound.




With some careful jiggling the top mechanism plate can be separated from the bottom plate. The bottom plate is loose but still connected to the spring beneath, so it needs to remain in place or it will lift the spring out with potentially nasty consequences.
On the underside of the top plate you can see the rather compact speed governor with its gearing, the feed and take-up spindle clutches and the two pawls attached to an eccentric shaft which drive the footage counter ratchet.




A close-up of the top mechanism plate with some of the gears removed. At left, the take-up clutch with its gear, at right the feed clutch. Just off centre you can see the idler gear that limits the spring wind in both directions, mounted with a small spring that allows it to be depressed from the film chamber. At the top is the speed governor - a worm gear driven shaft with arms that expand out with centrifugal force, housed within a sliding case that limits how far the arms can extend. The oil holes in the film chamber lubricate the governor - its two bearings, the sliding casing, and the worm drive gear.





The bottom mechanism plate. In the centre are the large main drive gear and (connected to the same shaft) the smaller gear with one shallow valley that meshes with the limit idler. Underneath, the shaft hooks into one end of the motor spring. By rotating the shaft anti-clockwise it will unhook from the spring and allow the bottom plate to be removed.







The motor spring revealed. The outer end (at top right) is held in place by a long pin that extends up through both mechanism plates. The spring is lubricated with graphite powder.
Reassembling everything in situ can be a little tricky. Certain gears need to be fitted to either one or the other mechanism plate. It usually takes me several tries to eventually get everything together with all the bearing shafts properly located. Before assembly, every bearing is lubricated with a drop of oil.


To reset the idler gear, the spring is wound several revolutions with the idler depressed before allowing it to re-engage. It should stop the spring from being wound before the point where it becomes very tight, and stop it unwinding before it slows down excessively. It often takes a few goes to get the meshing right.










Friday, 22 February 2013

Bolex H16 RX5 disassembly





Bolex H16 RX5


The RX5 first released in 1967 is for many 16mm afficionados the high point of the Bolex range, the last of the reflex C-mount turret series before single bayonet mount models arrived on the scene, which were designed to hold the heavier zooms that were all the rage at the time. But the zooms never matched the optical quality or speed of the C-mount Kern Switar primes, particularly the Macro-Switars pictured above. And with its flat base, triple lens turret and sleek magazine port like a Greek helmet all in tuxedo black leatherette and chrome, no other camera ever looked so handsome.

The following post documents the dismantling and servicing of one of these beauties, primarily for the curiosity of those interested in the procedure. It's not intended to be a guide for people hoping to do it themselves, because it requires certain tools and years of accumulated knowledge that are beyond most non-professionals. If your Bolex needs a service, take it to a professional!




Having thoroughly tested the camera to ascertain what may be out of tolerance, the first step is to remove the viewfinder. The next step is to remove the front. This loses the shutter to pull-down synchronisation, which must be reset during reassembly. It also breaks the light seal.

On reflex cameras the shutter assembly remains in the front, as opposed to non-reflex models. This makes removing a reflex front a considerably easier prospect than removing a non-reflex one.





After removing all the knobs and levers on the control side that are necessary, the entire mechanism can now be removed. The light seal is now completely broken.












The mechanism removed. The speed governor, release mechanism and claw drive gear are visible at the front.













After disengaging the spring limit gearing and winding the spring all the way down, the upper and lower plates can be separated.












A close up of the upper plate.










Both plates with the motor removed.













Complete disassembly of the gears, spindles and drive belts.













After ultrasonic bath cleaning.














The mechanism reassembled after lubrication, and play tolerances checked.













The claw assembly after ultrasonic cleaning. Claws are checked for wear and replaced if necessary.












Setting the sprocket height with a gauge. Later the rotational position will be set for optimal loop forming.
















The mechanism is now fitted back into the body.








If necessary, focal flange depth adjustment and reflex prism alignments are done to the front before fitting. Also any work required on the variable shutter mechanism or viewfinder prism. The gate is cleaned, and if it has been removed, reset with a jig. Shutter/pulldown sychronisation is set when the front is fitted.









Having lubricated and fitted the claw assembly, the forward and reverse claw torques are measured. If they are out of tolerance, new spring washers are fitted.











Using a gate jig, the claws are checked for alignment and adjusted if necessary. The sprockets are adjusted rotationally for smooth loop forming.












The light seal is reapplied.













Various speeds can be checked with Bolex stroboscopic discs fitted to the 8:1 driveshaft.













As they spin, an upper and lower tolerance band seem to spin in opposite directions if the speed is within tolerance. They work off the 50 Hz frequency of electrical lighting, much like old record player speed displays.







A more accurate technique is to use a strobe gun that can be set to three decimal places of a pulse rate per second, which I use for modern film cameras but is a bit of an overkill on spring-powered Bolexes. But it shows me that they are very stable for at least the first 15 or 20 seconds. This one slowed down considerably after about 15 seconds before service, now it maintains 24fps (within 0.3 fps) up until the last few seconds.









The take-up torque is checked.












Using a Bolex RX test lens designed to check infinity incorporating the prism in the light path, the camera focal flange depth is double-checked on a bench collimator.












After a final scratch test with fresh film, where film is run through the camera and checked for scratches using a loupe, the camera service is complete.