ARRI/Zeiss Ultra Prime 8R T2.8 Test
This is not a full review Evin style but rather looking closer at important aspects of the lenses in terms of Misterium 12MP resolving power and the demands that such resolution puts on the lenses.
Since I didn’t have the 12MP still picture camera with the PL mount, I had to compromise and use the pictures done by D20 6MP movie camera, just for now.
Also, for the resolving power test, I used test pattern that was projected on the wall ~4’ away from the lenses. The horizontal size of the picture on the wall was ~11’ wide.
All tests were done at T2.8
First, this is the 8mm (112 deg at 24mm horizontal sensor size) wide angle lenses that really do keep the vertical lines vertical.
The cushion distortion at the left or right edge of the frame are not noticeable, estimated less then 3 deg. However, if you move lenses off 5 deg from the horizontal plain you get 20 deg or so tilt in vertical lines at the edges.
See the picture below, the houses on the right are falling on its back ~20 deg, camera was pointing bit up and no shift/tilt adapter was used.
So make sure that your lenses are pointing perfectly horizontal or you have to use shift&tilt adapter like this:
Underwater shooter, do not need to be concern.
The focus adjustment on this lenses is extremely precise and smooth, lending itself to the high precision of adjustment required at 4K standard.
I could go in to the details of MTF curves and resolving power of Lp/mm but for the simplicity I converted the results in to the LCD screen or projection RGB pixels.
So the lenses do resolve more then double in terms of the required horizontal and vertical Misterium pixel count, even on the far periphery of the frame. 200Lp/mm at estimated 50% modulation.
It is what one should expect from aspherical lenses. No spherical aberration problems influencing the focus is visible.
I was unable to test 10% modulation (change between white and black) since the nature of the test chart used and the fact that I don't have 12MP sensor PL mount camera here.
I have noticed that it takes just a small movement of focus ring (~1/16”) to change resolving power from 200Lp/mm to 50Lp/mm so the focus adjustment will be the most demanding factor to keep the 4K images 4K, instead of 2K. Remember the tests were done at T2.6 changing it to T8 or higher did increase resolving power but I didn’t have 400Lp/mm marks on the chart so can’t speculate what was the magnitude of increase.
Also we have to remember that focus plain is the sphere around the lenses not a flat surface.
Chromatic aberration is the whole different thing with this lenses.
I still didn’t get the clear answer from ARRI/Zisse if this lenses is a full achromatic design but as compare to the focus, I noticed, one pixel width (80Lp/mm) chromatic aberration at the 80% and farther to the side of the picture. See the fragment of the left edge of the picture below, 100% crop, one to one pixel mapping.
Chromatic aberration is almost not visible.
Note that the picture was done by 2K camera (1920 horizontal pixels) so one pixel equals 40Lp/mm on 35 format sensor.
(1920/24mm = 80; 80/2=40LinePairs/mm) Where ~24mm is the horizontal size of the 35 mm format
IMHO this alone brings the 4K format down to the 2K format in terms of color resolution.
We will repeat this test with a different unit to make sure that we are getting well aligned lenses out of the box and there is no film versus lenses, specific optimization required.
We have to remember that film has 3 color layers, spaced ~4 micron apart each, in terms of distance from the lenses (film deep).
We adjusted back focus though, 15 microns back and forth and there was no improvement, the resolving power actually dropped to 50Lp/mm.
Chromatic Idiosyncrasies TBC.
I took quick look on the 16mm and 27mm ARRI/Zeiss Master Primes and I will test these lenses as well, in light of the Misterium demands.
This test do not addresses lens breathing test or bokeh test, since these lenses are very popular and plenty of lens specific tests were done and published.
I am more concern with the film versus CMOS differences, in light of 5.5 micron photo-sensor format of Misterium.
Also the influence on the resolving power and chromatic distortion, when film versus CMOS is considered.
Blair (RED reservation #19) was so kind, expressing his interest to participate in the tests here as soon as he will get his camera. Then we will be able to test the whole system resolving power, Lenses + Misterium and even the REDCODE RAW together.
We want to use the best possible lenses for Misterium so poor lenses quality will not lower the combined camera + lenses performance.
The questions to be answered:
Is there a difference in lenses optimized for film and for the CMOS sensor to the point that will influence in a major degree the resolving power at 4K format?
What is the best type of lenses for 4K CMOS format?
Do we need a special lenses design to be more forgiving on the focus setup?
Feel free to add more questions here.
Pictures of the res charts are below in post #27