• Ultraviolet Photography
  •  

Are longpass filters useful for transmission testing?

48 replies to this topic

#21 UlfW

    Ulf W

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

Posted 02 November 2018 - 15:59

View PostCadmium, on 02 November 2018 - 07:15, said:

The fact remains that the SEU has a strong amplitude of visual range violet, above 400nm transmission, that is more than other usual UV filters.

Even the Moon U, which is designed to be a UV + Visual-Range-Violet filter, only crosses over from UV to visual at 5%.

I fully agree that the SEU is different, but not necessarily in a bad way, at least not in most cases.
For special scientific applications it might be a problem with a small amount of transition a few nm beyond 400nm, but then other types of filters might be required anyhow.

I do not think it is valid to have the same attenuation demands in the transition area for this type of filter .
It is not the same type of problem as we see when fighting IR-leaks that give fake colours among the normal false colours from the different wavelengths of UV-light.

At 402nm the SEU2 transmission is almost four stops down.
That kind of attenuation, compared to the transition of UV-light from wavelengths nearby makes this "visual" contribution irrelevant in most cases.

I think there are very few materials that change their reflectivity that quickly for a slightly changing wavelength.
Any blue or violet in a motif would only be visible if it was something that absorbed almost all UV, below 400nm and still had a good reflectivity above 400nm.

The blue tones in the white balanced images from this filter is due to a different colour composition from the filter seen by the bayer matrix.
The blue tones are created by the white balancing and not likely from any violet or blue light.

I chose to think of the SEU2 as a UV-pass filter, as the main part of the transmitted energy is almost all UV and then cuts down to a rather deep OD very close to 400nm.
This is not anything comparable to the UV-B-G-stacks that use light for the images much further into VIS.

For me the MOON U would be a UV-pass filter too as the light from the UV transmission by far is the dominant contribution to the image, even if there is less UV near 400nm to overwhelm the 5% at and beyond 400nm.
Ulf Wilhelmson
Curious and trying to see the invisible.

#22 JMC

    Member

  • Members(+)
  • 829 posts
  • Location: London, UK

Posted 02 November 2018 - 16:56

View PostReed F. Curry, on 02 November 2018 - 15:42, said:

Jonathan,

How do you feel about your Sparticle work as noted above? Do you think that the Sparticle offers wavelength-specific conclusions, i.e., does your 405BP prove that the Baader U, the SEU Gen2, and the La La U, leak violet, in varying degrees, in the 405nm wavelength?

Thanks.

Regards,
Reed
Going off some of my old data I have prepared the graph below for the SEU II.
Attached Image: SEU II and Spart filt.jpg

This has data for three filters (my 404nm and 405nm filters from the Sparticle, and the SEU Gen II). I have also multiplied each of the Sparticle ones by the SEU II and plotted those on the same graph.

Based on this, the combined filter transmissions for the 404nm and SEU II are spread evenly around 400nm, with contributions from the UV and visible (assuming we are defining UV as <400nm and visible as >400nm), and the 405nm and SEU II is spread evenly around 401nm. What it is not telling me is that the 404nm Sparticle filter is giving information on what is going on at only 404nm, and the same for the 405nm one. They are both spreads of data.

Keep in mind these are before any recalibration check of my spectrometer. Also, this does not take into account the nature of the light source being used for any image or the camera sensitivity - this is purely transmission. Flashes and sunlight will have more intensity at the longer wavelength side, and the camera sensitivity will also be higher at the long wavelength end. These would both tend to emphasize the contributions from the longer wavelength end as there is i) more light there, and ii) more camera sensitivity.

Hopefully that all makes sense.

EDIT - emphasised to readers this is before recalibration of my spectrometer. Hence the wavelength values will not be identical to Ulfs.

Edited by JMC, 03 November 2018 - 18:03.


#23 Reed F. Curry

    Member

  • Members
  • 333 posts
  • Location: New Hampshire, USA

Posted 02 November 2018 - 21:37

Thanks, Jonathan, that is how I visualized it as well. However, your illustration makes it quite clear.

Warm regards,
Reed
Best regards,
Reed
http://www.uvroptics.com

#24 JMC

    Member

  • Members(+)
  • 829 posts
  • Location: London, UK

Posted 03 November 2018 - 10:35

View PostReed F. Curry, on 02 November 2018 - 21:37, said:

Thanks, Jonathan, that is how I visualized it as well. However, your illustration makes it quite clear.

Warm regards,
Reed
No problem Reed. If you think it may be useful, I can also prepare similar graphs for the Baader U and LaLa U, and I do have a set of three Sparticle photos, one for each of the SEU II, Baader U and LaLa U, all taken under the same lighting conditions and camera settings which might be useful to look at alongside the transmission plots.

#25 Andrea B.

    Desert Dancer

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

Posted 03 November 2018 - 13:26

I'm catching up with the reading of this very interesting thread. Not finished with it yet, but I'd like to thank Ulf for his excellent transmittance chart of the SEU2 in Post #17 and Post #18 along with the discussion of the importance of spectrometer
calibration.

I have always encouraged and applauded the creative and interesting informal measurements made by our members. Such work has greatly increased our knowledge about our filters and lenses. But it is important for certain filters or lenses (or cameras?) to have some careful, formal measurement made. So that we know for sure.




These would both tend to emphasize the contributions from the longer wavelength end as there is i) more light there, and ii) more camera sensitivity.

I am thinking that in the case of the interval 395 - 405 nm, there is not much dropoff in sunlight (for example). So if there would be about equal contributions on both sides, yes? But suppression is being applied from 400-405 nm and transmittance is being promoted between 395-400 nm.
Andrea G. Blum
Often found hanging out with flowers & bees.

#26 Reed F. Curry

    Member

  • Members
  • 333 posts
  • Location: New Hampshire, USA

Posted 03 November 2018 - 14:06

Yes, Jonathan, I think it would be very useful to have similar graphs for the Baader U and LaLa U. We are discussing the general concept of the precision of the longpass filters and the Sparticle filters in identifying discrete wavelengths of transmission, not just the potential and actual transmission of one UV bandpass filter. So, please, give us more! :)

Thanks.
Best regards,
Reed
http://www.uvroptics.com

#27 JMC

    Member

  • Members(+)
  • 829 posts
  • Location: London, UK

Posted 03 November 2018 - 15:14

Andrea, the effect I mentioned about sunlight and camera sensitivity changing will for sure be small, and I'm not even sure over that range of wavelengths whether it would have any effect. But I wanted to mention them for completeness.

EDIT - I have deleted my spectra, to avoid confusion.


Now for the Sparticle images. These were taken on my multispectral EOS 5DSR, with 105mm Rayfact lens, ISO400, f5.6, 2s exposure, with the Sparticle aimed at a blue sky, away from the sun. Taken as RAW files, and opened in Raw Digger as RAW composite files. Screen grabbed and cropped. The 404nm filter is on the bottom row, in the middle. The 405nm filter is on the bottom row on the right hand side.

SEU II Sparticle image
Attached Image: SEU II.jpg

Baader U Sparticle image
Attached Image: Baader U.jpg

La La U Sparticle image
Attached Image: LaLa U.jpg

Edited by JMC, 03 November 2018 - 18:00.


#28 Reed F. Curry

    Member

  • Members
  • 333 posts
  • Location: New Hampshire, USA

Posted 03 November 2018 - 16:57

Jonathan,

I just noticed that the SEU Gen2 data you used in your two recent posts was far different than the data Ulf gathered. Looking at 400nm, the SEU Gen2 on your graph is approx. 40%; Ulf has 24.1%. Looking at the 404 filter, at approx. 40% at 400nm, the Light-to-lens (product of SEU Gen2 and 404nm filter) is shown as 16%, whereas, it should be 9.6% by Ulf's measurement. That is a big difference. The result is that the potential transmission of the SEU Gen2 by 404nm filter curve appears much greater in amplitude than it would be in fact.

Do you have Ulf's data on the SEU Gen2?
I hope you can correct that, as it is significant.

Thanks.
Best regards,
Reed
http://www.uvroptics.com

#29 JMC

    Member

  • Members(+)
  • 829 posts
  • Location: London, UK

Posted 03 November 2018 - 17:07

As mentioned before Reed (a few posts ago) my system has not been re calibrated since buying it, so I'm not hugely surprised there is a difference to Ulfs. I was very clear in pointing that out. I have a calibration light source on order but it won't be with me until the end if the month. If you would prefer I can delete the post, the last thing I want to do is add confusing data again.

What I would say though is the Sparticle is at least trending how I would expect it too, based on the transmission data for the different filters.

EDIT - Based on the concerns over the spectra I am deleting the spectra from my post comparing with the Sparticle images in this thread, to avoid potential confusion I'll leave the Sparticle images up though, as they are independent of any spectrometer calibration. I will revisit any filter transmission measurements, once my new calibration setup has arrived, although I would need to destroy my Sparticle to get the filters out, unfortunately, as they were glued in when making it.

Edited by JMC, 03 November 2018 - 18:06.


#30 Reed F. Curry

    Member

  • Members
  • 333 posts
  • Location: New Hampshire, USA

Posted 03 November 2018 - 18:59

Hi Jonathan,

I think we are always going to have problems when, as a forum, we can't agree on data sources. At 400nm, your SEU Gen2 is 166% greater than Ulf's; whereas, at 400nm, your Baader U is 58% of Ulf's.

What this means is that your data is spreading the difference between the SEU Gen2 and the Baader U unrealistically. Ulf's data shows a ratio of 23.4:1, SEU Gen2 to Baader U at 400nm. Your data spreads that to 65.1:1, SEU Gen2 to Baader U at 400nm. That vast difference clashes with the evidence of the Sparticle itself.

Your recent Sparticle images should illustrate the differing transmissions of the filters. However, if we look at them, they do not show the spread that the graphs indicate. The GS images below were rendered grayscale by brightness.

Baader U (400px wide) Baader U (grayscale 400px wide)
Posted Image Posted Image

SEU Gen2 (400px wide) SEU Gen2 (grayscale and 400px wide
Posted Image Posted Image

The 405nm Sparticle in grayscale is 10 of 256 for the Baader U and 97 of 256 for the SEU Gen2. Change to percentages and the ratio of SEU Gen2 transmission to Baader U transmission is 12.7:1. Your graphs would indicate a ratio of 65.1:1... 5.13 times greater than the Sparticle images show.

It looks like the problem is greater than your spectrometer calibration.

I do appreciate your work. Thanks.

Edited by Reed F. Curry, 03 November 2018 - 19:02.

Best regards,
Reed
http://www.uvroptics.com

#31 Andy Perrin

    Member

  • Members(+)
  • 2,677 posts
  • Location: United States

Posted 03 November 2018 - 19:35

What is the formula for getting "brightness" from the color images, Reed? Several members have attempted to derive the Sparticle colors from the spectra graphs (with different data sets) and nobody has managed to get it right in that direction. It seems really iffy to me to attempt it in the other direction also. The spectroscopy data may indeed need calibration, but calling in question everything Jonathan has done seems a little much?

Edited by Andy Perrin, 03 November 2018 - 19:36.


#32 JMC

    Member

  • Members(+)
  • 829 posts
  • Location: London, UK

Posted 03 November 2018 - 19:44

The filters have been measured at different times on different days, so I can understand that drift has occurred, even simply as a result of that. As the values around 400nm are already very low, very small changes in wavelength will have a big percentage difference when looking at the values.

At least the camera images of the Sparticle aren't subjected to spectrometer drift. What they do emphasize though is the effect of sensitivity drop and light intensity drop as a function of wavelength. On your greyscale images above, the 404nm and 405nm Sparticle filters (bottom middle and right) are darker than the rest of the Sparticle filters other than the top left and top middle ones (303nm and 321nm respectively) on the Baader U image. By eye, as I'm not measuring it, the SEU II greyscale images, the 404nm and 405nm Sparticle filters look to be somewhere around the same brightness as the middle Sparticle filter which is the 364nm one.

But (and it is a big but) even this is an over simplification. It is all further complicated by the Sparticle filters not all transmitting the same amount - the 404nm and 405nm ones transmit much more than one at 382nm and below, over emphasizing potential issues. Just goes to show how many variables there are when looking at these things, and how some things can over emphasize results. I prefer to use the Sparticle for relative tests, not absolute ones - change one variable at a time and see the effect.

I'm not sure to read this comment, "It looks like the problem is greater than your spectrometer calibration.", Reed. Perhaps I'm missing the context, but I'm not clear on that, but it is definitely a complex area.

Edited by JMC, 03 November 2018 - 21:00.


#33 UlfW

    Ulf W

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

Posted 03 November 2018 - 19:49

View PostReed F. Curry, on 03 November 2018 - 16:57, said:

Jonathan,

I just noticed that the SEU Gen2 data you used in your two recent posts was far different than the data Ulf gathered. Looking at 400nm, the SEU Gen2 on your graph is approx. 40%; Ulf has 24.1%. Looking at the 404 filter, at approx. 40% at 400nm, the Light-to-lens (product of SEU Gen2 and 404nm filter) is shown as 16%, whereas, it should be 9.6% by Ulf's measurement. That is a big difference. The result is that the potential transmission of the SEU Gen2 by 404nm filter curve appears much greater in amplitude than it would be in fact.

Do you have Ulf's data on the SEU Gen2?
I hope you can correct that, as it is significant.

Thanks.

The SEU2 has a beautifully steep transition close to 400nm.
Even a small deviation in calibration will give big shifts in transition value at 400nm.
My original measurement was ca 0.7nm off at that wavelength giving a transition value of ca 24%.
When the wavelengths scale was calibrated the second measurement that I posted in this thread gave ca 32% at 400nm.

If Jonathan's currently uncalibrated spectrometer is off slightly more than 1nm in opposite direction, compared to my first measurement that would produce a transmission of ca 40%, assuming that the two filters we measured are completely identical.
A drift of around 1nm after more than a year is not impossible.

Edited by UlfW, 03 November 2018 - 20:12.

Ulf Wilhelmson
Curious and trying to see the invisible.

#34 Reed F. Curry

    Member

  • Members
  • 333 posts
  • Location: New Hampshire, USA

Posted 03 November 2018 - 20:47

Jonathan,

When I wrote "It looks like the problem is greater than your spectrometer calibration" I was referring to the divergence in your data from that of Ulf's. If the problem was only calibration, it would seem that your transmission curves for both the SEU Gen2 and the Baader U would be parallel (roughly) to Ulf's. However, as there was not parallelism but divergence, then one or both of you have problems other than calibration. Does that make any sense?

I'll measure the 400nm of the SEU Gen2 on my Hitachi U-1500 and get back to you. :) I still haven't gotten my diode array working. Hey, it's only been a year. :(

Edited by Reed F. Curry, 03 November 2018 - 21:10.

Best regards,
Reed
http://www.uvroptics.com

#35 Reed F. Curry

    Member

  • Members
  • 333 posts
  • Location: New Hampshire, USA

Posted 03 November 2018 - 20:58

Andy,
You asked "What is the formula for getting "brightness" from the color images, Reed? " Back in the day, I wrote graphics and imaging SW for a SW company. Great fun! We had a lot of different algorithms we wrote for filtering bitmaps. Here is a link to three simple algorithms for color to grayscale conversion - https://www.johndcoo...olor-grayscale/

I used luminosity.
Best regards,
Reed
http://www.uvroptics.com

#36 Andy Perrin

    Member

  • Members(+)
  • 2,677 posts
  • Location: United States

Posted 03 November 2018 - 21:06

I don’t understand how you can say he is doing his spectra incorrectly for reasons beyond calibration and not be questioning all the prior work? It did seem kind of harsh, but I don't understand how else to reconcile things?

Edited by Andy Perrin, 03 November 2018 - 21:11.


#37 Cadmium

    Member

  • Members(+)
  • 3,033 posts

Posted 04 November 2018 - 03:57

In my humble opinion, this topic is not really about longpass filters, or bandpass filters, or spectrometers, or how to use them, or who has the best one,
but instead, the original intention of this topic seems to be centered on dispelling any idea that the SEU is not a 'UV-only' filter.
I leave it to others to define what a 'UV-only' filter is, should be, or what you want it to be.
Instead I will present a graph overlay, using Jonathan's spectrometer graphs of the SEU, and several stack plots using the Schott program.
First let me say to Jonathan, I hope it is OK to use your graphs in this overlay, please let me know if it isn't and I will gladly remove this post, this graph, or put your name on it, however you like.
Here are Jonathan's original graphs he made using his spectrometer. I used his top graph in this overlay graph below.
http://www.ultraviol...dpost__p__22664

I believe Jonathan used a Perkin Elmer Lambda 650S UV-Vis spectrometer for his graphs:
https://www.google.c...wiz.75JIgqrjt1o

Also, note: This is a combination of what I assume is a T graph (made by Jonathan) and Ti graphs made with the Schott program.
Because Jonathan's graph is linear, I didn't have any way to make a combined (stacked) T graph that was in linear form with the program.
Therefore the peak amplitudes of the two calculated stacks would be slightly lower in T form than they are presented as in Ti form on the graph below,
and the 400nm cross over points might be slightly lower also, again, if those graphs were presented in T form.

OK, so this graph compares the SEU with a U-360 1mm stack that is intentionally designed to be UV+VV (visual-violet),
and also with Ulf's UG2A stack, which is a UG2A 2mm + S8612 2mm stack, and it is also designed to be UV+VV, with even more VV content.

By the way, we all have Ulf to thank for the UG2A, because he asked for it, and pushed for it, and the UG2A is a welcome and useful addition to our repertoire of filters,
so that is why I have labeled his stack the "ULF U" on the graph below. :)
He has posted results using that stack several places on this board.
http://www.ultraviol...__fromsearch__1

http://www.ultraviol...__fromsearch__1

My handy dandy UV and Violet reference list:
UVA = 320nm to 400nm
Violet = 380nm to 450nm
VV = Visual Violet = 400nm to 450nm
UVAV or UVV = UVA Violet or UV Violet = 380nm to 400nm

Regardless of whether or not the SEU crossed the 400nm threshold at 45%, 24%, or 20%, I consider ALL of these filters/stacks to be UV+VV filters.
Please note how all of these filters on the graph below clearly transmit across the UV-only 400nm separation line, and have one foot solidly planted in VV.
There is nothing wrong with using any of these filters, and there is nothing wrong with UV+VV,
however, to define or classify any of these filters as being UV-only would be erroneous.
Attached Image: UVA_plus_Visual_Violet_filter_comparison.jpg

This version may be a little easier to read:
Attached Image: UVA_plus_Visual_Violet_filter_comparison_NEW.jpg

Edited by Cadmium, 04 November 2018 - 08:13.


#38 UlfW

    Ulf W

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

Posted 04 November 2018 - 08:35

View PostCadmium, on 04 November 2018 - 03:57, said:

I believe Jonathan used a Perkin Elmer Lambda 650S UV-Vis spectrometer for his graphs:
https://www.google.c...wiz.75JIgqrjt1o

I think that is not correct.
He has been using a Ocean Optics FX-spectrometer for quite some time now.
It is either a OCEAN-FX-UV-VIS or a OCEAN-FX-UV-VIS-ES.
https://oceanoptics....oduct/ocean-fx/
That is the one he and I have discussed how to calibrate with a reference lamp for some time now.

I'm shure Jonathan will let us know what he used.

I see that you honor me by your naming of one of my favourite UV-stacks. Thank you very much, Steve. :)

View PostCadmium, on 04 November 2018 - 03:57, said:

My handy dandy UV and Violet reference list:
UVA = 320nm to 400nm
Violet = 380nm to 450nm
VV = Visual Violet = 400nm to 450nm
UVAV or UVV = UVA Violet or UV Violet = 380nm to 400nm

The list here is an excellent idea! I like it very much.
I like to have clear definitions and that give a good common understanding when discussing things.

I have no opinion against the chosen names. They work well for me.
Another one might be UV-VIS border (400nm)

Together with that there should be some definitions of common decision levels to use for defining wavelength widths and cutoff wavelengths
Normally 50% of full transmission for cut on, cut off and FWHM.
https://www.google.c...haracteristics.

With the commonly used 50% decision level, by definition all filters we have discussed above will with margin fall into the UV-pass category.
There might be some real reason to use a different definition of decision level in this case, but I cannot see any.

For analytical purposes the amount of light passed from the VIS or NIR might be too big.
That varies from case to case, but do not change the fact that by normal definitions they are UV pass filters.

To define what would be a UV-only filter, one must decide what "only" means in OD terms both directly at the UV-VIS border and further into the VIS-NIR area.

I think the OD close by 400nm is application-dependent and would preferably be based on practical effects for a "normally" intended usage.
This would be a very interesting subject to discuss, maybe in a new thread.

A more important factor, for our type of photography might be some kind of measure of the total amount of transmitted light beyond 400nm.
I would then suggest some kind of weighting-factor dependent on how far away the light's wavelength is from 400nm.
Ulf Wilhelmson
Curious and trying to see the invisible.

#39 JMC

    Member

  • Members(+)
  • 829 posts
  • Location: London, UK

Posted 04 November 2018 - 08:45

Thanks for explaining about the, "It looks like the problem is greater than your spectrometer calibration" comment Reed, yes that makes sense.

Steve, I originally used a Perkin Elmer system for filter measurement, but I lost my access to that in September 2017. I now use an Ocean Optics FX UV-VIS spectrometer, in combination with one of the OO DH-2000-BAL light sources. The SEU work was done with that.

#40 Cadmium

    Member

  • Members(+)
  • 3,033 posts

Posted 04 November 2018 - 09:16

OK, then how does this idea sound, would you all like me to classify the U-360 1mm stack as a UV only filter?

.