• Ultraviolet Photography

[Filter Test SEU Gen2 #4] Dealing with the Usual Dichroic Effects

8 replies to this topic

#1 Andrea B.

    Desert Dancer

  • Owner-Administrator
  • 7,541 posts
  • Location: USA

Posted 17 July 2018 - 18:40

[Last Update: 2018.07.19 12:08 EDT First post complete.]

[Filter Test SEU Gen2 #1] Introduction to the SEU Gen2
[Filter Test SEU Gen2 #2] White Balance, Raw Histogram & Andrea's "White Signature"
[Filter Test SEU Gen2 #3] Filter Speed & A Windy Bull's-eye
[Filter Test SEU Gen2 #4] Dealing with the Usual Dichroic Effects
[Filter Test SEU Gen2 #5] Landscape Interlude
[Filter Test SEU Gen2 #6] Monochrome Museum Comparison
[Filter Test SEU Gen2 #7] Measured Filter Transmission
[Filter Test SEU Gen2 #8] Dichroic Reflection Detour
[Filter Test SEU Gen2 #9A] Longpass Stack Wandering Discussion. See #9B for results.
[Filter Test SEU Gen2 #9B] Longpass Stack Results
[Filter Test SEU Gen2 #10] What good is a filter test without a Rudbeckia?
[Filter Test SEU Gen2 #11] A Fascinating New Rudbeckian View
[Filter Test SEU Gen2 #12] Summary

Dichroic Filters and Concentric Discoloration

We all know that a dichroic UV-pass filter can produce some concentric discoloration when used over a wide-angle lens. I wanted to directly address that so that there would be no surprises for anyone using any of the currently available dichroic filters. UVR Optics directly mentions the wide-angle issue on its product page for the new dichroic SEU Gen2 (scroll to the bottom) and discusses the dichroic effect on a dichroic FAQ page. The Baader Planetarium does not mention the problem because its BaaderU is promoted for astronomy use not photographic use.

What did come as a surprise later -- to me -- was that converting dichroically discolored photos to a monochrome or split tone rendering mitigated some or all of the problem areas in the photograph. If you are already aware of this, good! If not, then I'm going to show some alternate monochrome or two-tone conversions to illustrate this in the second post. Monochrome conversion of dichroically discolored photos needs further exploration to determine how well it holds up for various types of scenes and subjects.

DICHROIC TEST: Use dichroic filters over a wide-angle lens to investigate dichroic concentric discoloration.

Question: Does the dichroic SEU Gen2 produce dichroic concentric discoloration when used on a wide-angle lens?

Answer: Yes, the SEU Gen2 produces dichroic discoloration, sometimes more than does the BaaderU on a wide-angle lens. But I'm going to show a work-around which seems to mitigate the worst of the problems for some scenes.

REMINDER: Those new to dichroic filters, please remember that dichroic problems occur with dichroic filters used over wide-angle lenses. I have seen no dichroic problems using the SEU Gen2 on my UV-Nikkor 105/4.5 or my EL-Nikkor 80/5.6 (metal version).

Subject: Hitachi backhoe

Location: Southwest Harbor, Maine, USA

Gear: D610 + Novoflex 35/3.5 Noflexar + UV-Pass Filter + Sun

UV-Pass Filters: I chose to use commercial, currently manufactured filters in these tests.
  • BaaderU :: 350fwhm60 :: approx 80% at peak :: OD > 3.5??
  • KolariU :: 365fwhm40 :: approx 50% at peak :: OD > 3.5
  • SEU Gen2 :: 392fwhm50 :: approx 74% at peak :: OD > 4.5
Exposure: f/16 at various speeds @ ISO-100

Conversion: Photo Ninja. White balance and some sharpening was applied. Black and white points were adjusted as needed. Clearly I did not crop out the dark corners or lift any vignetting. The non-white-balanced raw composites were converted in Raw Digger.

Comment_1: The first thing you will likely notice in these photos are the Dark Corners. Those are my fault, so kindly overlook them. I am up here at the summer rental in Maine without my drawer(s) full of step-rings and other useful fittings. So properly mounting the narrow 48mm Baader filters over the 35mm Noflexar was impossible. I also got a little bit of the Dark Corner effect with the KolariU and the SEU Gen2.

Comment_2: Falloff vignetting (optical, natural, mechanical, pixel) is always more obvious with wide-angle lenses. That's just physics. I think that the dichroic filters make any falloff perhaps a bit more obvious? For more about the various types of vignetting, see Vignetting, where digital pixel vignetting is also discussed.

Comment_3: Does your full-frame camera have a DX option? Use that when shooting with a dichroic filter over a wide-angle lens. The instant crop will get rid of some of the dichroic or vignetting problems.

NOTE: These photos are all sharper when clicked up to their maximum 1200 pixel width in an enlarged browser. Click the photo, then click again on the pop-up.

Backhoe Shoveling Up My Red Car
Aside: Those advocates for using a prime with foot zoom rather than using a lens with prime zoom must never encounter the situations I find myself in like this one where the hill of dirt was too steep and crumbly for me to back up any further to get room to crop for the filter vignetting produced by those snarky little 48mm Baader filters. Oh well. I just HAD to shoot this backhoe!
Attached Image: d610_novoflex_vis_sun_backHoe_20180713swhME_11824pn.jpg

BaaderU Conversion
It's almost as though there is a reverse hotspot here. The BaaderU dichroic effects don't show so much in the next two nearly monochrome versions.
Attached Image: d610_novoflex_uvBaader_sun_backHoe_20180713swhME_11829pn.jpg

BaaderU As Shot
Attached Image: d610_novoflex_uvBaader_sun_backHoe_20180713swhME_11829asShot.jpg

BaaderU Raw Composite
Attached Image: d610_novoflex_uvBaader_sun_backHoe_20180713swhME_11829rawComp.jpg

SEU Gen2 Conversion
The discoloration, more obvious in the next two versions, happened to work out to give a kind of less saturated center after white balancing from a white balance preset made for the SEU Gen2 using Spectralon. Using an average white balance over the entire frame produced a colored central area, as did making a marquee white balance over an edge sky area.
Attached Image: d610_novoflex_uvSeuGen2_sun_backHoe_20180713swhME_11831pn.jpg

SEU Gen2 As Shot
Attached Image: d610_novoflex_uvSeuGen2_sun_backHoe_20180713swhME_11831asShot.jpg

SEU Gen2 Raw Composite
Attached Image: d610_novoflex_uvSeuGen2_sun_backHoe_20180713swhME_11831rawComp.jpg

KolariU Conversion
Note that there is some obvious falloff vignetting even with this non-dichroic filter.
Attached Image: d610_novoflex_uvKolari_sun_backHoe_20180713swhME_11838pn.jpg

KolariU As Shot
Attached Image: d610_novoflex_uvKolari_sun_backHoe_20180713swhME_11838asShot.jpg

KolariU Raw Composite
Attached Image: d610_novoflex_uvKolari_sun_backHoe_20180713swhME_11838rawComp.jpg
Andrea G. Blum
Often found hanging out with flowers & bees.

#2 Andrea B.

    Desert Dancer

  • Owner-Administrator
  • 7,541 posts
  • Location: USA

Posted 17 July 2018 - 18:40

[Last Update: 2018.07.19 15:43 EDT Second post now complete.]

Monochrome Conversions of Dichroically Discolored Photos

MONO CONVERSION TEST: Convert dichroically discolored photos to monochrome and investigate the result for dichroic artifacts. Experiment with toning and split-toning.

Question: Do dichroic artifacts lessen or disappear?

Answer: Happily, yes!

Comment_1: I experimented with several types of monochrome conversion. I'm just going to show you one unusual method here.

Note: I am going to use the preceding photos again, so I will not repeat all the details. Please refer to the first post above.

The SEU Gen2 Backhoe As Shot
Let's repeat the SEU Gen2 ultraviolet photo here, for reference.
Attached Image: d610_novoflex_uvSeuGen2_sun_backHoe_20180713swhME_11831asShot.jpg

As Shot Inversion
Reed reminded me the other day that inversions sometimes better show problems areas. Keeping in mind that there is some falloff vignetting, you can see that the problem area is mostly inside the circle I drew. In particular, note the brightness difference in the cloud-free sky areas under the backhoe swing arm. (On the right, the cloud free area is to the upper left although not many low lying clouds were left by this time.)
Attached Image: d610_novoflex_uvSeuGen2_sun_backHoe_20180713swhME_11831asShotInversion01.jpg

Comment_2: It seemed natural to wonder whether the dichroic effects from the SEU Gen2 on a wide-angle lens were worse in one channel than in another. And indeed that is what I found.

As Shot Red Channel
Needs a bit of desaturation to bring back the clouds. "-)
Attached Image: d610_novoflex_uvSeuGen2_sun_bxackHoe_20180713swhME_11831redChanBwPts.jpg

As Shot Green Channel
Attached Image: d610_novoflex_uvSeuGen2_sun_bxackHoe_20180713swhME_11831greenChanBwPts.jpg

As Shot Blue Channel
So, can we conclude that the discoloration seems to lie in the blue channel? I don't know yet if this occurs regularly, but it was interesting to see that the answer is certainly 'yes' for this example. So I decided to eliminate this pesky blue channel.
Attached Image: d610_novoflex_uvSeuGen2_sun_bxackHoe_20180713swhME_11831blueChanBwPts.jpg

As Shot Red + Green Channels :: Banish the Blues!
It surely does look to me like the dichroic problem is now mostly gone once the blue channel was eliminated. There is still some falloff vignetting, but that could be dealt with.
Attached Image: d610_novoflex_uvSeuGen2_sun_backHoe_20180713swhME_11831redGreenChan.jpg

Comment_3: So how did I get from there to the next photo?? B) Took a TIF of the Red + Green Channels into Photo Ninja, adjusted the some exposure sliders to very slightly open up the dark areas and then used PN's Black & White tool to tone the photo with a low saturation blue. The split tone slider was adjusted to taste. I chose not to deal with any remaining vignetting for this demo.

Finish: Cyanotype Backhoe :: Come Back Baby Blue (a rather desaturated and split tone type of blue)
This is such a fine backhoe!
The dichroic discoloration is gone in this monochrome, blue-toned version produced from the Red + Green channeled version above. This is more of a bluetype than a cyanotype I suppose. As I tend not to let my reflected UV photographs languish in the dark obscurity to which they are naturally prone, I have taken the liberty of locally enhancing some contrast in certain areas as well as performing a bit of judicious dodging, both in NX2 and after the PN adjustments and toning.

Attached Image: d610_novoflex_uvSeuGen2_sun_backHoe_20180713swhME_11831pnCyanoCrop.jpg
Andrea G. Blum
Often found hanging out with flowers & bees.

#3 Andrea B.

    Desert Dancer

  • Owner-Administrator
  • 7,541 posts
  • Location: USA

Posted 19 July 2018 - 16:10

[Last Update: 2018.07.19 12:08 EDT First post seems to be complete now.] 6 new photos were added. Now working on 2nd post.
Andrea G. Blum
Often found hanging out with flowers & bees.

#4 Andrea B.

    Desert Dancer

  • Owner-Administrator
  • 7,541 posts
  • Location: USA

Posted 19 July 2018 - 19:45

[Last Update: 2018.07.19 15:43 EDT Second post now complete.]

This topic is now open for discussion.
Andrea G. Blum
Often found hanging out with flowers & bees.

#5 dabateman

    Da Bateman

  • Members(+)
  • 1,295 posts
  • Location: Maryland

Posted 20 July 2018 - 05:27

Very interesting, my first thought is why in the blue channel. My gut guess would have been the red channel as its seems boosted in uv images.

Very helpful general article. Also makes me wonder if its always that way.
I also wonder if its a Sony/Nikon sensor thing. As I haven't seen that on four thirds, as typically more telecentric design. But than again there is really no wide angles for four thirds as all are x2.
I wonder if its due to the micro lenses off angle focus on sensor to help with sharpness in corners. Something that was a problem on the first Sony E mount full frame cameras, until they changed the microlenses. Then maybe always in blue channel due to nature of wavelengths and resolution.

Edited by dabateman, 20 July 2018 - 05:35.

#6 dabateman

    Da Bateman

  • Members(+)
  • 1,295 posts
  • Location: Maryland

Posted 20 July 2018 - 07:01

Another thought,
I see you shot this at f16. This will be diffraction limit on your D610. I think the average pixel pitch is 5.97um for that camera. At 360nm average wavelength the limit is about 7.12um, where as at 395nm its 7.71um. You maybe seeing the offset of airy discs between the Baader shot (bright exterior), vs the SEU (bright spot).
If you wanted to investigate this it would be helpful to know if the whole issue is absent at f11. Where you are not diffraction limited with the D610.
Again I don't know why this would still only appear in the blue channel though.

#7 UlfW

    Ulf W

  • Members(+)
  • 848 posts
  • Location: Sweden, Malmö

Posted 20 July 2018 - 09:02

It is logic that the difference is in the blue channel only for the SEU-filters.
It is the dominant channel where the dichroic effect is for this filter.
That will not be the case for all filters.

Here is a comparison of the responses by the SEU 2 and Baader U for different angles.
The rightmost curves are for normal position, then increasing angles to roughly 20°-25°.
Attached Image: Screen Shot 2018-07-20 at 10.52.14.png

The slopes that are not affected are crated by the ionic substrates/components in the filters.
If a set of spectrograms with more accurate angle steps is interesting I can redo the measurements.
Ulf Wilhelmson
Curious and trying to see the invisible.

#8 dabateman

    Da Bateman

  • Members(+)
  • 1,295 posts
  • Location: Maryland

Posted 20 July 2018 - 10:36

That makes a lot of sense. However what I am surprised by with your curves is the difference between the spectral character. Your SEUmk2 ends about 405nm, but at sharp angle to subject looks to end below 400nm. The Baader shift is in the UV end, were due to camera sensitivity we may not see the effect. The peak maximum changes but not by huge amounts, possibly not noticeable.
Very interesting response.
After an hour of sleep I remember that diffraction shouldn't play much role here. My brain has just be stuck thinking about it for other purposes. And when your using a hammer, you suddenly want to use it on everything.
Remembering Jonathan post:
The red channel has strongest sensitivity below 380nm and based on Ulfw spectrum, can you tease out the Baader image and see if the effect is only in the red channel?

#9 UlfW

    Ulf W

  • Members(+)
  • 848 posts
  • Location: Sweden, Malmö

Posted 20 July 2018 - 12:45

The accuracy of low level values for these spectrograms is limited by the crosstalk of the spectrometer with the used settings.
The main thing I wanted to show was the difference in where the dichroic filtering happened and approximately how much the change were.
Ulf Wilhelmson
Curious and trying to see the invisible.