M9 - Color Skopar 28mm f/3.5 - fast test

[Godfrey]

Decided to fit the tiny little 28 to the M9 this week.

Since I'd never used this lens on the M9 before, I took out the lens coding kit and created the code on the LTM->M adapter for 11804. Then I checked the M9 lens codes table that someone had posted (I think created by Sean Reid and someone else) and found the recommendation there for using 11604.

That led me to making a very fast set of three test images .. open blue sky at the same settings (ISO 160 @ f/8 @ 1/160 s) .. one with the auto lens code setting, one with the 11604 lens setting, and one with no-code setting.

Results are interesting. The 11804 setting produces the best blue of the three, the others are a bit cyan, and the least amount of corner edge fall off, but all three show some magenta shifting on the left.

Using the 11804 image as a calibration test target, CornerFix cleared up the magenta shifting on all three nicely.

Interesting results. I'll have to experiment more and use a gray card for more accurate evaluation, of course. Funny thing is that when I'm not taking a photo of a clear blue sky, the color shift is much less pronounced and in most cases almost unnoticeable. .

 

[Godfrey]

When I arrived home, I had a more deliberative test all mapped out for the Color Skopar 28mm f/3.5 on the M9.

- set up a flash rig in a dimly lit room to as evenly as possible light an off-white, textured wall.
- made test exposures at each of the lens codes 11804, 11604 and none, starting at ISO 80 @ f/4 and going to ISO 1250 @ f/16.

All the exposures were brought into Lightroom 3.6 and examined. There was marked evidence of color shifting and corner falloff (much more than with the GXR + A12 Camera Mount). However, corner to corner sharpness seems quite good, no untoward smearing or other artifacts.

I created calibrations for CornerFix at f/5.6 for each of the lens code settings and applied them to all the image files in that lens code group. Bringing the fixed .DNG files back into Lightroom 3.6, most of the color variations are removed and the corner falloff is almost eliminated as well.

I also looked at five or six exposures I'd made today of more average subjects (people, cafe scenes, etc). They looked quite good! Without looking at evenly lit, monochromatic subject matter, the color shifting and corner fall off is barely noticeable.

So: CornerFix is important for accurate edge-to-edge color work with this lens, but not so critical for more average, varied scenes.

Very interesting stuff. I'm sure many who've been working with the M9 and some of these near-symmetrical lens designs are well aware of this already ... I'm still on the "learning a new camera" thing with the M9. ;-)

fun fun fun

G

 

[Ario Arioldi]

Very interesting Godfrey.
I made myself some experiment in the past with other CV lenses (non retrofocus) combining coding an CornerFix and I found that in general the best results were obtained using CornerFix without any code.

 

[Godfrey]

Very interesting Godfrey.
I made myself some experiment in the past with other CV lenses (non retrofocus) combining coding an CornerFix and I found that in general the best results were obtained using CornerFix without any code.

Thanks for commenting.

That's why I tested both of the recommended lens code settings and the "no code" setting. At least with this lens, the differences are minimal using my white target. Here are the three tests. The ones in red were the calibration masters at f/5.6. Top is 11804, next 11604 and third row no codes. Highlighted frames are the CornerFix corrected frames.

Perhaps it's better to see them all against a white backdrop:

From that, any of the three settings should do fine.

What the lens codes might do to detailing, color balance, and geometric corrections, well, that I'll have to see.

G

 

[Ario Arioldi]

Well done.
As far as in-camera correction originated by the lens code, my understanding is that it will affect vignetting and color shift only.

 

[250swb]

Good work Godfrey.

Steve