I set out to create profiles for three lenses: Zeiss 21 f/2.8 ZE, Zeiss 35 f/2.0 ZE, and the Canon 90 TSE. The bad news is I’m not done. The good news is what I did do now seems to work quite well. For me, it was quite a learning experience. I had several problems along the way that slowed me down dramatically. These problems stem from two issues: First, I kept trying to out-think this process. Second, maybe it is just me, but the documentation is mysterious in several places.
The first thing I did wrong was print the wrong charts. I followed the directions, which suggest shooting at 1x the minimum focus distance (which is less than a foot), leading to a very small target. Thing is, because the Zeiss lenses don’t send distance information, I realized after the fact that I’ll be manually selecting the specific distance profile. Therefore, I can choose the distance that makes the most sense for me; I don’t have to follow those recommendations (I hope this is correct – Eric?). Not many of my photos have a subject within a foot of the lens, so I didn’t go there.
Ultimately I chose two distances: Roughly 2 feet and 8 feet. I’ll add a third very close set later when I get a chance. I didn’t go farther because the largest chart I can print is the 24x36. The actual distances I used were determined by the chart sizes I printed. I just moved back until the chart was ~1/3 of the frame.
I decided to shoot outside against the garage; in the morning the ambient light is very even there. This turned out to be good and bad. Good because the vignetting adjustments seem to be spot-on, so the even light turned out just right. Bad because it was a very humid weekend. Anyone try to keep a 2' x 3' print flat going from an air-conditioned house to 90% humidity? Ugh. In fact I had to give up on the 35mm at 8 feet, because I was sure the profile would add warping from the print as if it was a lens issue.
The shooting guide reads as follows:
i. First shoot centered straight on.
ii. Tilt the camera up, thereby effectively framing the chart at the bottom-center of the image.
iii. Tilt camera down, thereby effectively framing the chart at the top-center of the image.
iv. Move camera a bit to the left (so that when turning to the right to face the chart, it is about 10 to 30 degrees). Take a series of shots similar to the first three, above, except that the chart is framed at the center-left, top-left, and bottom-left areas of the image.
v. Move camera to the right, and do the same for the center-right, top-right, and bottom-right areas of the image.
This seems backwards (assuming the reference is standing behind the camera). If I move the camera to the left, the target will be on the right side of the frame, not the left side! I suppose they could be talking about the actual image on the sensor, which is upside down and backwards, but if that is the case then the up/down is wrong. Anyway, I’m presuming they don’t want us to put an extreme angle to the film plane on purpose. At the minimum focus distance there would be no depth of field. I certainly didn’t do it as described above. I moved the camera left and framed on the right, then vice versa.
I also was paranoid about simply tilting the camera. The first time I raised the tripod up to minimize the angle on the film plane thinking this helps. It doesn’t. I repeated the 35mm without raising or lowering the tripod and got results that to me are indistinguishable. It’s certainly much quicker to simply tilt. Especially because my tripod doesn’t have a center column.
It really is quite an application. It doesn’t matter whether the film plane is parallel and it’s not critical to have all the images lined up precisely. Just make sure all the images in each series have the same aperture and shutter speed, and get as close to the edge of the frame as possible. From there it is point and shoot, running through a series of f-stops/shutter speeds then reposition and repeat.
Where I did get confused was making the profile. At first I thought I needed to rename each series as it relates to the f-stop. That created a mess of profiles that didn’t work at all. For these lenses there’s a profile created for each distance. Each profile has all the f-stops. Just put all the dng’s at each distance in the same folder, and then select them all with the check boxes. At the end of each series it asks how to save the profile; just keep the name the same each time as it works through the f-stops.
Here are some images showing before and after for the 21, which is more dramatic than the 35. Sorry for the ugly subjects, but at least they highlight the difference.
The first is the Vignetting
[attachment=22973:Zeiss_21_moustache.jpg]
[attachment=22974:Zeiss_21_shaved.jpg]
Note: The stone is not evenly lit. Bottom and top are supposed to be darker
Second is moustache distortion
[attachment=22975:Zeiss_21_vignette.jpg]
[attachment=22976:Zeiss_21...adjusted.jpg]
Third is CA
[attachment=22977:Zeiss_21...orrected.jpg]
[attachment=22978:Zeiss_21...orrected.jpg]
Ciao,
Dave
