[Filter Test SEU Gen2 #6] Monochrome Museum Comparison

[Cadmium]

I don't think it is UVIVF.

Ulf's graph shows a bit of blue in the mix, and Andrea's 400nm stack show blue also.

However, the tests I have seen her do with 420nm (and above) longpass stacks are black using the same exposure time.

For blue, it would be nice to check it somewhere between 400nm and 420nm.

I think if someone is wondering what they are seeing in a UV photo, UV, or UV + some other out-of-band light mixed in,

they they should do the same test, stacking whatever filter they are using with some longpass filter that cuts off the UV right above where they think that transmission should stop,

say 420nm in this case, and exposure the stack to the same exposure time as the UV shot. What remains is the out-of-band light that might be getting mixed into the UV shot.

 

If different results of that test are seen with various software, then that is up to people to decide which to use and judge things by.

[Dmitry]

Note to self: testing spectrum of UVIVF B)

[dabateman]

I think I will always say visible range is 400nm to 700nm as that is what I was taught in University. Some people can't see above 650nm, some can see just to 700nm. Some people probably can see below 400nm.

For a company to make a claim they would need to test their exact filter set and maybe even equipment. There are tolerance in everything.

I got a 405nm filter. It has range of 400nm to 410nm. Using it I saw similar uv image with a clover flower as with the Baader venus. Problem is the transmission peak on my filter is about 25%. So the exposure times really suck. But it does lead to question of where is the floral band range that we call a uv image. I still need to test multiple subjects with my filter series, 340, 370, 390 and 405, to make a statement.

[rfcurry]

Jonathan,

 

Thank you for your insights, testing, and sharing your experience.

 

When comparing two UV-bandpass filters, such as the BaaderU and the SEU Gen2, we should, imo, first look at the UV wavelengths that the filters optimally pass. Doing so in this case shows us that the BaaderU passes little in the 390-400nm wavelengths. However, this is the dominant solar UV range. The SEU Gen2 has its primary peak at 392, so it is getting most (?) of the photons inaccessible to the BaaderU. My first thought would not be that the SEU Gen2 was allowing visible light to pass, but that the SEU Gen2 was passing more longwave UV than the BaaderU.

 

The graph below is from my website and the data to create it was generated by Ulf

http://uvroptics.com/images/SEU%20Gen2%20Final%20Diabaticcrop.jpg

 

Clearly, there is some possibility of violet intrusion. However, the transmission in the 392nm-400nm range is much greater than that in the 400nm-405nm range. Certainly, if the target was solid extreme violet, the poor transmission in the 400nm-405nm range would be sufficient to create an image.

 

Just a thought.

[JMC]

Reed - yes, absolutely lots more in the 380-400nm region for the SEU, and I agree the relative amount of UV to visible in the image would be very high. I've prepared a number of evidence decks for court cases over the years, so unfortunately I look at things with a "what if?" mentality, as I've been asked to both defend scientific methods and find flaws in them for different cases.

[rfcurry]

Jonathan,

 

I understand that one. :( I can't count the times I have been in a conversation, deeply involved in problem solving, when the other party says "I didn't want you to fix anything, I just wanted you to listen." My bad.

 

Yes, the engineering mindset is very creative and fulfilling, but it never turns off. Not ever.

 

But it sure is fun...

[ulf]

The graph below is from my website and the data to create it was generated by Ulf

http://uvroptics.com/images/SEU%20Gen2%20Final%20Diabaticcrop.jpg

 

The transmission levels at < ≈ 300nm is not from the filter, but crosstalk in the measurement setup.

The small peaks around 1000nm might be from system-noise.

 

None of them are likely to cause any practical issues for UV-photo due to sensor sensitivity.

[Cadmium]

Violet, as defined in more than one Wiki, as being 380nm-450nm, which is a 70nm range, 20nm below 400nm, and 50 above 400nm.

I used the term 'blue', but I really meant 'anything above 400nm'.

So instead of saying blue or violet I am going to say 'anything above 400nm', or how about the term '400nm+'.

 

I agree that the peaks above 420nm don't show up with the UV blocking equal exposure tests, at least using more common and usual exposures times for the UV shot.

However, there is significant 400nm+, or what I sometimes refer to as 'blue edge', or 'blue leak'.

 

If we look as Andrea's test here:

http://www.ultraviol...__fromsearch__1

 

We see the GG400 stacked shot, which is shot with the same exposure time as the UV shot.

This does transmit some 400nm-, but transmits everything above 400 at 50% or stronger.

From what I see there, without being able to compare it directly to a UG1 2mm stack myself, it looks like there is a bit more 400nm+ present than even with a UG1 stack.

 

I tried playing around with Andrea's two shots, UV and 400nm+, and subtracting one from the other to see how it compared to the UV shot...

Probably the best test would be to perform the same set of shots using a UG1 stack to see what the 400nm+ difference is. And perhaps compare other stacks and UV-only filters this way also.

Then having a good comparison of how much 400nm+ is present in the UV exposures of different filters (and stacks).

Regardless of how much 390nm/upper UV-A range is present, the 400nm+ shot is still part of the UV shot at that same strength as shown, because they are both shot at the same exposure time.

 

I am not saying there is anything wrong with the 400nm+ ('blue/violet'), I am just saying it does influence the overall UV shot if it shows up in the same exposure time 400nm+ stack test,

unlike the rest of the peaks, illustrated by the 420nm+ stack tests, which all show only black.

[Cadmium]

Just tossing this out, hope you don't mind me playing with your test photos, Andrea.

I have no idea exactly how to define this, but this is a photoshop layer subtraction,

layer 1 = SEU shot

layer 2 = SEU + GG400 stack shot

layer 2 is changed from normal to subtract.

your thoughts are welcome, because I don't know exactly what that subtraction creates.

Just playing around...

 

color

 

mono version (original two shots desaturated before subtraction process)

 

To reverse that process, take the center frame and add it to the right frame, "Linear Dodge (Add)".

And then you are back to the original.

[Andrea B.]

Subtraction appears to be a kind of cancellation effect.

I can't figure out how it is relevant here?

 

If you subtract blue (0, 0, 255) from purple (128, 0, 255) then you get medium red (128, 0, 0). But what does that even mean with respect to my two photos, one of which is kind of purple and one of which is kind of blue ??

[Andy Perrin]

Are you trying to subtract off the out-of-band parts of the image to leave behind the UV only?

[Cadmium]

Andy, yes, that was the idea. Know how to do that??

[Andy Perrin]

Yeah, that is what ICA is supposed to do essentially. But in this case it’s not needed because the long pass filter has already separated out the OOB stuff so straight subtracting should work. The only issue is that the LP will not pass 100% of the passband, so you may want to adjust the opacity by a few percent to adjust for that.

 

ETA: You need to be using the raw, though. You can't do it after it's been JPG'ed because JPG adds a nonlinear curve to it. (This is the same issue as when I do the subtraction for the EIR images -- it's why I work with the RAW-converted TIFF in MATLAB instead of using JPG in photoshop.)

[Cadmium]

Cool... I think?

So is the above subtraction any good at all then?

[Andy Perrin]

Based on my experiment with the Aerochrome, probably not. The first time I tried with them I did it in PS with white balanced JPGs and got garbage. It was only when I went back to the unWB'ed RAW that I got good results. You don't need MATLAB, only the RAW to get it to work though. The procedure is fine otherwise.

[Cadmium]

So... what might your version of the idea look like?

[Andy Perrin]

Same thing, just get Andrea to send the RAWs over, convert in PN with all the options including WB unchecked, save to TIFFs, then do the subtraction. If you want me to try, send me the original RAWs.

 

Options in PN like so:

 

I don't own any GG filters or I'd give it a go myself.

 

The issue with using JPG is this step "Color Space Transformation":

https://en.wikipedia..._transformation

 

It alters the colors in the image permanently, which prevents simple subtraction from working right.

[rfcurry]

I now use an electric shaver but I always keep Occam's Razor handy. :)

 

Just looking at the GG400 2mm per Schott, I noticed a few things. Here is the transmission per wavelength of the GG400 and SEU Gen2 from 392nm-407nm.

http://uvroptics.com/images/GG400overlay.jpg

 

 

Here is what it looks like graphically:

http://uvroptics.com/images/SEU%20Gen2%20and%20GG400.jpg

and here is the result of the two filters:

 

http://uvroptics.com/images/SEU%20Gen2%20and%20GG400%20overlay.jpg

 

Now, I acknowledge my bias, but does anyone else detect much more light <400nm than the light >400nm?

 

We know from Ulf's very extensive testing that the SEU Gen2 curve is accurate. We know nothing about the particular melt of the GG400, but Schott advertizes the above transmission.

 

Does the simple premise above have any merit?

 

Thanks.

[Andy Perrin]

Looks fine to me, Reed. Since the graph is diabatic, only the stuff right near 400 on either side will contribute much, but it still looks tilted to the <400 side.

 

Is anyone really concerned about this filter, or are we all just geeking out for the heck of it? Which I have no trouble with — I'm quite curious to see if the <400nm stuff could be subtracted off using the TIFFs. From the spectrum Reed shows here, one may need something with a straighter edge (heh) than the GG400 to be able to extract just the >400 part. Maybe two GG400's?

[Cadmium]

I wanted to subtract an image from an image, man... if anyone can do that, it has to be you, Andy. You are the image guru of MatLab.

They should employ you, I think.

[Andy Perrin]

I wanted to subtract an image from an image, man... if anyone can do that, it has to be you, Andy.

Heh. I told you what you have to do? What's the problem?

They should employ you, I think.

I literally considered applying one time. Opted not to, since I am pretty allergic to working for companies. (I like to be free.) But I am their greatest cheerleader. :)

[Cadmium]

Well, I am not as smart as you are about that kind of thing.

[Andrea B.]

The equine's soul has long since departed for the Pearly Gates so we can stop performing CPR on it now.

:lol:

 

(I couldn't resist.)

[Andy Perrin]

Aw, Caddy, you're so sweet!

[dabateman]

@Andy,

This is totally just geeking out. I shot with my 405bp10 filter yesterday and saw the same uv typical pattern on a clover flower. The filter is not as good as the subtract above only 25% transmission, but tight at 405, max lower 400nm maximum upper 410nm and blocked for all else.

 

I have really been staring at Andreas flower shots taken with Baader venus vs SEUmk2 on a perfect lens. I don't see any difference in the uv pattern regions.

 

@Andrea,

If you up for some subtract photo edit work. Can you take an image with exact exposure for both Baader venus and SEUmk2 and subtract them. My feeling now is there will not be any difference in the floral uv highlights. That is for all intensive purposes the SEUmk2 with peak max 392nm, provides the exact same UV image, in regards to flowers as the Baader venus.

 

This obviously will not apply to glasses though as Jonathan has shown here in the canopy shot.