Micro Nikkor-P Auto 55mm f/3.5

[enricosavazzi]

There is an entry on a Micro Nikkor-P 55mm f/3.5 lens in the UV lens sticky, reporting -2 EV with respect to the Noflexar 35 mm.

 

There are several versions of this lens, so my results may not apply to all versions. Mine (see picture below) is one of the later non-AI version (home-converted to AI) with single-layer coating, so in theory more likely to work as a UV imaging lens than subsequent multicoated versions. The optical scheme has remained constant through the years, with 5 elements in 4 groups moving as a single assembly during focusing. The lens elements are quite thin because of their small diameter, another potential plus-factor. The single cemented doublet is probably glued with Canada balsam, a minus-factor. These lenses are known as reasonably sharp and still useful in macrophotography (albeit limited to 0.5x without an added extension ring) in spite of the simple optical formula.

 

 

My simple test with Baader U and full-spectrum Sony A7 II suggests that this lens performs marginally in UV imaging. False-red overwhelms false-blue in the test image, which looks quite monochromatic. I tried pushing the false-blue up dramatically to show some chromaticity, but the result shown below is more of an artefact than real false color (interesting expression, isn't it?). Nothing is recorded in the green channel, so there is no way to recover any false-yellow, except perhaps with a massive addition of 365 nm LEDs. Solar UV transmission is roughly 3 stops less than I would expect with a reasonably good accidental UV lens, and limited to 380-390 nm judging from the results. This lens would probably show UV nectar tracks on flowers, but none are available at the moment. As a whole, one is probably better off with a 50 mm El-Nikkor on a focus helicoid.

 

[ulf]

Or a Canon FD 50/3.5 Macro:

https://www.ultravioletphotography.com/content/index.php/topic/3447-canon-fd-5035-macro-a-simple-uv-macro-lens/page__hl__%2Bcanon+%2Bmacro__fromsearch__1

[Stefano]

Nothing is recorded in the green channel, so there is no way to recover any false-yellow, except perhaps with a massive addition of 365 nm LEDs.

Do you need the green channel for false-yellow? I think you get it from the red of the red channel (before WB).

[enricosavazzi]

Do you need the green channel for false-yellow? I think you get it from the red of the red channel (before WB).

To get yellow you need both red and green. Red being a primary color on Bayer sensors, from red alone you cannot get yellow without in some way putting some non-zero data in the green channel. The same applies to violet, you need both red and blue non-zero data.

 

One way to create non-zero green data where there is none is color remapping, e.g. remap the blue channel by swapping its data with the green channel.

 

Another way is pushing up the zero level of the green channel (by adding the same offset to the green channel in all pixels) to artificially introduce non-zero green data where there are only zeros in the original picture.

 

Both methods are forms of post-processing and create data where there is none in the original picture.

[Stefano]

Yes, what I meant before is that you don't need green in the RAW image to have yellow in the white-balanced image. Usually in UV the reddish portions in the image become yellow when you WB. Of course to have a yellow pixel you need red+green.

 

Below is an image I made in Paint. The colors are (128, 0, 255), (255, 0, 255) and (255, 0, 128). Those colors are colors you could see in a RAW UV image. Probably that much color separation is a dream and nobody will ever see that, it would be awesome, but the colors are roughly what I would expect. I can't white-balance it, I don't have any software to do that. Can someone do that for me, clicking in the middle square? I want to see if my "theory" is correct. You should see, from left to right, blue, gray and yellow.

 

[nfoto]

Like this?

 

[Stefano]

Like this?

 

Yes, exactly.

[Andy Perrin]

I think the point is proved. It's interesting that the click white balance does NOT simply re-weight the channels by multiplying them with different numbers, which is how I've always heard it explained. There is clearly addition involved also or the green channel would not be able to go from 0 to non-zero.

[nfoto]

This result was a 'UV click-white' done in Photo Ninja, using the middle sector as specified. I had to convert the original .png file into something PN would handle, so saved it as jpg in Photoshop. Otherwise no alternations introduced.

 

The change of colour space from Adobe RGB to sRGB might have introduced a minor change of pixel values say from (128,0,255) to (150,0,255) for the left and (255,0,128) to (255,0,131) for the right sector, respectively. I doubt this has had much influence on the balancing in PN. One could of course make the appropriate prior bias before going to sRGB, then submitting the file to PN, but the main lesson of this exercise remains unaltered.