Dichroic Filters & Wide-Angle Lenses

[bvf]

I recall posts on UVP saying that one of the problems with Dichroic filters like the Baader U is that they have problems with wide-angle lenses. I assume that these problems should manifest them as you get away from the centre of the image.

 

Have I got that right?

 

Assuming that is right, I tried to see if I could detect the problem. My widest angle UV-friendly lens is the 28mm Lithagon. So I took comparison shots at centre and corner of the frame (APS-C sensor) using a Baader U and a U340+S8612 as an absorptive filter.

 

I just can't see any difference - either between centre and corner or between filters. So have I misunderstood the issue, or is the problem only present at wider angles?

 

Top images are Baader U, bottom images are U340+S8612.

 

Left image is a crop from the centre of the frame, and right image is a crop from extreme corner.

 

Images WB-ed against PTFE, specifically for the filter being used.

 

 

[Stefano]

They are both pinkier at the edges, but that's probably the lens, as this shows up in both tests.

[Andy Perrin]

Bernard- try it with one of the omega filters. It’s more obvious with them because they were never intended for photography I bet.

 

In particular, the issue is caused by light hitting the filter at a steep angle rather than head-on. Those are the conditions needed, as well as a background that will show the issue. (I would think you’d want a mostly blank background without much clutter?)

 

Finally you should be taking two crops from the same image. It looks like you moved the camera in those two shots if that was the same building. That will never show the issue!

[bvf]

Bernard- try it with one of the omega filters.

 

...

 

Finally you should be taking two crops from the same image. It looks like you moved the camera in those two shots if that was the same building. That will never show the issue!

 

Unfortunately the Omega filters are too small (only 25mm) to go on the front of a 28mm lens, and I can't fit them to the back because they're cemented into a 49mm filter mount.

 

Yes, I moved the camera betwen shots so I could get the same building. I'll try doing what you say.

[nfoto]

With an FX (24x36mm frame) camera, the uneven colours are plainly visible even with the not-so-wide 35mm Noflexar. One really needs to put the filter to the rear of the lens to (partially) mitigate this issue.

[ulf]

You have got it right except that a 28mm lens on an APS-C-sensor is not that wide.

Dichroic filters work by interference between many layers.

The thickness of the layers is essential for the filter's characteristics.

When the light is hitting the filter at an angle the traveling distance in the layers change, changing the transmission curve.

 

These are some transmission curves for Baader U and SEU2 at different filter angles:

This was done hand held in a not very controlled way and I cannot remember the rotation angles.

[rfcurry]

Here is a generic representation of wavelength shift due to AOI. Obviously, some lenses may not pass, and/or, some sensors will not be able to record, the lower wavelengths. That is lost data.

http://uvroptics.com/graphics/Bandshift400px1.jpg

 

Birna, imo, presented the best solution to the problem - collimate the light before it strikes the filter.

[dabateman]

Bernard the dichoic effect is a phenomenon that occurs when Andrea takes a UV photograph.

Sorry Andrea, but all the examples I know of on the board are from you.

 

See:

https://www.ultravioletphotography.com/content/index.php/topic/2877-filter-test-seu-gen2-4-dealing-with-the-usual-dichroic-effects/

https://www.ultravioletphotography.com/content/index.php/topic/2884-filter-test-seu-gen2-10-what-good-is-a-filter-test-without-a-rudbeckia/page__view__findpost__p__23872

 

[nfoto]

Andrea talks about a real effect that is there for anyone to see. In fact, it makes processing wide-angle captures rather difficult at times.

[Andrea B.]

Bernard the dichoic effect is a phenomenon that occurs when Andrea takes a UV photograph.

Sorry Andrea, but all the examples I know of on the board are from you.

 

Apologies not necessary because it's my job to try to show everyone what can happen !!! All those photos were purposefully posted and side effects purposefully mentioned.

 

HOWEVER, I'm not sure the effect is entirely from dichroic filters. There is a lot of effect from light fall-off due to filtering a wide-angle lens. Whether it is a single cause or a combo of causes, the point is to be aware that this can happen with wide-angle lenses.

 

In one of those topics somewhere I show a nice method for dealing with the miscoloration.

LINK: Drop blue channel, stack R with G, convert to Mono, clean-up, Tone with some color or split tone with 2 colors.

https://www.ultravio...dpost__p__22595

 

---

 

Bernard, the photos are somewhat underexposed.

 

I think I'd show the entire photo in order to look for problems? You must compare the center with concentric areas around it.

 

What camera with the APS-C framing? It is entirely likely that the lens coverage over a smaller sensor mitigates the problem.

[bvf]

Bernard, the photos are somewhat underexposed.

 

I think I'd show the entire photo in order to look for problems? You must compare the center with concentric areas around it.

 

What camera with the APS-C framing? It is entirely likely that the lens coverage over a smaller sensor mitigates the problem.

 

Andrea - not sure I agree with the comment about the exposure, but how would a higher exposure impact on this discussion?

 

The camera was a Sony A6000.

 

Here are the full images - as always, happy for anyone to have the RAWs of they want:

 

 

[Andrea B.]

Andrea - not sure I agree with the comment about the exposure, but how would a higher exposure impact on this discussion?

 

Just a passing observation. It probably does not impact the dichroic discussion.

 

I was actually thinking that it might be very useful to make the observation somewhere in one of the Stickies that using an APS-C sized sensor can have some benefits when using dichroic filters. I'm going to go look now to see what is there.

[Stefano]

A bit out of topic: I was now looking at the full-size images, and one thing became more apparent: the Baader U image has a pinkier hue. I earlier noticed that the edges were pinkier, and also the Baader U images were pinkier than the U-340 + S8612 stack, but didn't really give it too much importance. But now it is really apparent, looking at the sky. Is this an IR leak? The Baader U is sometimes at the thereshold of leaking, with its ~OD 4.8 (If I remember correctly). There are examples here of leaks due to a poor UV lens (this is an example, there are surely many other on this forum), and Bernard's images may be another example?

[OlDoinyo]

I have posted a number of shots that show the effect here; it is not just Ms. Blum. And that is after I have tried to minimize the effect in post-processing.

 

Here is a less-processed image, taken with the Baader through a pinhole about 45mm from the sensor surface. Display intent BGR.

 

 

Is this an IR leak?

 

The Baader can definitely leak IR and that can become significant when the UV signal is weak, such as at night or with bright IR sources. But I have not seen IR leaks that look pink--most appear greenish.

 

The FF vs APS-C business is relevant only inasmuch as it affects field of view. The smaller sensor will have a narrower field of view with the same focal length optic and thus will show the effect less.

[Andy Perrin]

The Baader can definitely leak IR and that can become significant when the UV signal is weak, such as at night or with bright IR sources. But I have not seen IR leaks that look pink--most appear greenish.

In BGR or RGB?

[OlDoinyo]

In BGR or RGB?

 

Actually, both. It just alternates between yellow-green and bluish-green.

[bvf]

... the Baader U image has a pinkier hue. ... Is this an IR leak?

 

I think it's more likely to be WB issue. The photos were taken in weak sunshine - note the weak shadow on the tiled roof. When I took the PTFE WB shot shortly afterwards the PTFE was in full sunshine.

 

According to RawTherapee, the colour temperature for the WB I used was 3028. If I WB against the fir tree (which should be reasonably rich in IR) the colour temperature is reported as 2996, and the image is still pinkish. If I WB against the clouds the colour temperature is 2933 and the pinkness has largely gone.

 

Here are both images WBed against the same bit of cloud (Baader U on top). There is still a slight pinkness in the Baader image.

 

However, I would have thought that if there wasany noticeable IR leakage this would have made the source image somewhat pinker and the WB would then have upped the blue in the output image. So I would expect IR leakage to manifest itself with a blue tint rather than a pink one.

 

 

 

PS: I've increased the contrast in the U340+S8612 image as it was rather flat, even though both filters had just been cleaned to get rid of surface clouding - and the sky detail is still poorer. And ther U340+S8612 exposure was 2 secs. compared with 1 sec. for the Baader U (auto-exposure with an exposure compensation of +2 stops).

[Andrea B.]

IR leakage in a reflected UV photo causes a general overall washed-out effect especially noted in areas which are supposed to be UV-dark. For example, if you are making a reflected UV photo of a sunflower, you would see the expected UV-dark bullseye as lighter than it should be. How much lighter that bullseye is depends, of course, on the strength of the IR leakage.

 

I don't think there is any particular false color associated with IR leak. I've seen those sunflower centers look more of a false dark green with IR leak (after WB applied). If you [that is a general you, not a specific you] have any dual-bandpass filters, experiment with them without using a blocking S8612. Then you can see the interesting range of false colours possible with IR invading UV. (Not really "leakage" in a dual-bandpass filter.) Photos made with unblocked U-360, UG11 and so forth can be quite false-colourful depending on what white balance setting you choose.

 

 

As Bernard has noted, his photos are not entirely white-balanced hence the pink cast. Everyone knows my often repeated statement that we do not *have* to white-balance our UV photos. We only do it for a standardized look. And white-balance is *not* the only way to render UV photos. OlDoinyo uses a very nice channel switch which is also a great way to render UV. Many UV photogs leave a bit of pink tone in their work.

---

 

I've increased the contrast in the U340+S8612 image as it was rather flat, even though both filters had just been cleaned to get rid of surface clouding - and the sky detail is still poorer.

 

With filter stacks you can sometimes get a bit of flare or veiling -- causing lower overall contrast -- but it is often cured by changing the shooting angle. It is *tough* to shoot into a bright sky like that in any wavelength.

 

It's not a surprise that a well-blocked U-340 stack would have a longer exposure than the BaaderU with OD3.5-4.0.

 

[bernard, I know you know most of this. Sometimes I add comments which I hope are educational for less experienced folks.]

[Andrea B.]

.....well, andrea rattles on....

 

I forgot to mention that I was looking through a bunch of my photos last night and observed that using a dichroically filtered wide-angle lens on an FX sensor at a close range (say within 5 feet) did not induce concentric discoloration. Probably because when I'm photographing flowers with a 35mm lens, I'm down on the ground and the light is not entering the filtered lens at quite the same combination of angles that it would be if I were photographing a large wide-angle landscape. Make sense??