• Ultraviolet Photography

The importance of using a common relative reference level

Filters UV Lens
5 replies to this topic

#1 UlfW

    Ulf W

  • Members
  • 249 posts
  • Location: Sweden, Malmö

Posted 11 March 2018 - 14:58

It is very important to use a common relative reference level when comparing cutoff wavelength-values for filters and lenses.
If different levels are used any comparison become meaningless.

I was amazed and disturbed when I finally realised that this is not always common practice on this site.

In documents for optical components I have seen the convention of using 50% of the passband maximum level, as the reference level.
As an example see under Filter Nomenclature for Longpass and Shortpass at

In general 50% of the maximum level is a very commonly used reference level in the optical industry.
In special cases other levels can be used, but then this level information must follow the wavelength value to avoid misinterpretations.

An numerical example:
Assume a filter or lens, measured and having a passband attenuation with a transmission of 80%.
The transition level to define the cutoff wavelength will then be 40%.
( 50% of 80%, not just the 50% - level )

In the Lens Sticky and on his own website, Klaus has for some unknown reason used a different much lower reference level.
I have not found a definition in the Lens sticky what level he used used, but at Photozones 24 January 2010, in a diagram, he used -3 stops.

The wavelength-values derived from different levels are not possible to be compared.

The decrease of 50% correspond to one f-stop.
In combination with the quickly decreasing sensitivity of the camera-sensor in the UV-range even that is a quite significant decrease.

I prefer to continue using 50% to define cut-off wavelengths in my measurements of filters and lenses.

Edited by UlfW, 11 March 2018 - 15:03.

Ulf Wilhelmson
Curious and trying to see the invisible.

#2 JCDowdy


  • Members
  • 1,119 posts
  • Location: Arlington, TN

Posted 11 March 2018 - 19:23

The 50% cut-off is a standard metric but not perhaps all that one might wish to consider. The shape of the cut-off, degree of blocking and maximum transmittance is also standard information.

Source: schott-optical-filters-2015-catalog-complete-us.pdf, page 13, section 3.6.1.
Attached Image: LAMBDSCP.JPG

λc: Edge wavelength or cutoff wavelength at which point the spectral internal transmittance has a value of 0.5.
λs: The limit of the blocking range. Below this wavelength, the internal transmittance has a value below τi,s for a certain spectral region.
λp: The limit of the pass band. Above this wavelength, the spectral internal transmittance does not fall below τi,p within a certain spectral range. The pass band can be divided into several sub-ranges, e.g. into two ranges with τi,p1 = 0.90 and τi,p2 = 0.97.

I have understood Klaus's metric of -3 stops to be a general descriptor of what he considers the limit of useful transmittance for photography, or a τi,pKDS = 0.125 if I may. B)
I seems intuitive to me that the slope of the cutoff may well impact the false color UV white balance.

#3 Cadmium


  • Members(+)
  • 1,547 posts

Posted 12 March 2018 - 05:20

I like to use "FWHM" = "Full Width Half Maximum". My Baader U has the term HWB on the side of it, which they use to mean "Full Width at Half Maximum".
I refer you to the the Baader U page, and the description they use:

"80% at 350nm, bandwidth 60nm (320-380nm), completely blocking the entire spectral range from 200nm to 1120nm"
Here we are told that the "Center Wavelength" (CWL) is 80% at 350nm.
They are describing the bandwidth as 60nm (320nm to 380nm).
My particular Baader U has the following on the box, "T=300-390", and the following on the filter ring, "HWB=325-369nm"

I prefer a graph, but it is a bit hard to put a readable graph on the edge of a 4.5mm tall filter ring.
So the term I like to put there is FWHM, for a stack anyway.
But the graph is important for people to see, have, and understand, because the FWHM does not fully define the characteristics of a filter.

Another important factor is out of band transmission, out of band suppression, out of band optical density (OD).
And where do we set the limit for out of band suppression? At the boundaries of the FWHM? No, we need to designate an edge to the full band which is defined by the nanometer of the OD we choose.
The minimum out of band OD will refer not only to the the red/IR peak suppression, but also to a blue edge (in this example).
I like to use the 1E-03 line for this as a danger zone, but 1E-04 or more is better.
FWHM is good info, fits on the side of the ring, but dose not inform us about the out of band suppression, and with photography this can be quite important information.
FWHM doesn't express the peak transmission %, which doesn't help us compare the peak strength to other such filters, unless we look at a graph of course, and this can be valuable information for our photography.
Here is an example below, this is not a recommended stack, just showing it as an example.
CWL = 364nm, Peak Transmission = 60% (5.98E-01), 364FWHM50
Minimum Out of Band Suppression = 3.46E-04 for the defined band of 312nm - 408nm
Why do I designate this filter band as 312nm to 408nm? Because it is specifically designed to transmit some blue edge (or violet, or whatever color you want to call that),
the transmission crosses over the 400nm line, it is about 1E-03 at 407nm.
Using the FWHM definition alone would tell me nothing about the out of band suppression, and nothing about how strong the peak UV transmission is, and it would give me no idea how much blue to expect from the filter.
I need a lot more information than FWHM, we all do.
There may be abbreviated terms to express more, but a graph shows it all.
Attached Image: graph_FWHM_b.jpg

Edited by Cadmium, 12 March 2018 - 06:09.

#4 Andrea B.

    Desert Dancer

  • Owner-Administrator
  • 6,287 posts
  • Location: USA

Posted 14 March 2018 - 17:45

I was amazed and disturbed when I finally realised that this is not always common practice on this site.

No, there is unfortunately no uniformity in the way transmission data is presented. Whatever we are given is what is posted. Or whatever a manufacturer provides is what is posted. I really have no way to force standards on manufacturers or Dr. Klaus or anyone else.

I seems intuitive to me that the slope of the cutoff may well impact the false color UV white balance.
John makes a good point about the shoulder of the transmission curve. For example, the right shoulder of the StraightEdgeU is very important because of its extreme steepness. For this filter you would not want to use a 50% cut-off measurement on the right side because it would leave out important information. (The SEU is a filter which produces a lot of false blue but does not leak any blue/violet.)

Any suggestions for "cleaning up" the Stickies are always welcomed. I am happy to try to do so. We can attempt to provide the full-width half-maximum numbers if I can get some help with it. :D :D :D I've tried in the past but many times I think I have a FWHM nailed down and some new filter iteration changes that.

Creating the Chart post was another way I have tried to help with this. Showing an entire transmission chart is important. Then the FWHM can always be determined. http://www.ultraviol...mission-charts/

The Chart post needs a serious effort to bring it up to date. It should have charts for the important filters, filter glass, BG blockers and filter stacks. A lot is currently missing.

BTW, the manufacturers do not always label their filters according to any common standard.
Andrea G. Blum
Often found hanging out with flowers & bees.

#5 Cadmium


  • Members(+)
  • 1,547 posts

Posted 15 March 2018 - 01:27

FWHM applies to bandpass filters, but ignores much of the information we want to know that makes a lot of difference to our results.
FWHM doesn't really apply to such filters as longpass filters, which are designated by the 50% transmission point, more like the graph John shows from the Schott catalogue above.
Schott and Hoya base their longpass filters on the 50% transmission point, nothing to clean up there.
UV bandpass filter (or what you could call dual band UV + IR filters, such as UG1, UG5, UG11, and the Hoya equivalents, U-360, U-330, U340)
are designated by Schott and Hoya by the peak transmission nm-point of their UV band. Nothing to clean up there, but doesn't tell you a thing about the rest of the story unless you look at the graph.
UV only filters (such as absorptive stacks and dichroic interference filters), nothing to clean up there, because you will need other info about the transmission more importantly than what the half maximum is.
Calling a filter whatever FWHM doesn't tell you anything about a lot of important things.
You need a graph, or else a lot of other data to tell you the whole story.
Take UG5, as an example, or even a more extreme illustration like a UG5 stack, which transmits UV+Blue+Green,
If we were to make a FWHW for a UG5 stack, this would be virtually useless for what we want to know about that stack, or even the UG5 filter alone.
I am sorry if I sound like I am not agreeing with anyone here, but I don't think too many filters are wrongly designated here, especially by Schott or Hora, and I don't see how FWHM helps anyone with any of these filters when it comes to UV or IR photography. Take a look at these UG5 graphs, and tell me exactly how FWHM helps you know what you need to know when you have no graph?
The same thing goes for UV-only filters, FWHM is an interesting designation, but tell me how that helps you know how much blue is in the mix, or where the plot drops below your OD line?
Attached Image: UG5_fwhm.jpg

Attached Image: UG5_stack_fwhm.jpg

Edited by Cadmium, 15 March 2018 - 01:29.

#6 Andrea B.

    Desert Dancer

  • Owner-Administrator
  • 6,287 posts
  • Location: USA

Posted 19 March 2018 - 20:46

Yes, it is tricky providing a simple description of the various filters and stacks we use.
But when things are given a designation I need to make sure it is correct -- or at least add some information to make sure everyone understands what the numbers represent.

Particularly troubling -- for me, anyway, -- are the cut-on and cut-off values. I have been prowling around this afternoon and do find that different manufacturers designate this in a different way. By that I mean that some brands label the point where a 5% absolute transmission begins/ends as the cut-on/off points -- and some brands do not because they use a different value, or neglect this altogether.

The charts are the most valuable way to pick up info about a filter or a stack.


I was playing with the Schott program today and from the chart and data table I gathered the following information about a Schott stack consisting of a 2mm thick UG11 together with a 2mm thick S8612.

Stack: UG11 x 2.00mm + S8612 x 2.00mm

Center Wavelength, Peak Transmission: 355 nm
Maximum transmission of 54.66% (OD 0.26) @ 355 nm.
Designating center wavelength makes sense for this particular stack. But for the UG11 alone, it makes no sense because there are two "peaks" in a dual band-pass filter. (Of course, it is easy enough to mention those two peaks.)

Half Maximum Points, Half Power Points
The region where the filter transmits at half of its peak transmission.
.2724 = 27.24% @ 377 for OD 0.57
.2701 = 27.01% @ 329 for OD 0.57

Full Width at Half Maximum: 377 - 329 = 48 mm
Filter Designation: 355fwhm48
That is a fair designation for this filter stack which has a nice symmetic transmission curve. And I think we all understand what this means. However, for our UV photography we all seem to want to know more about the transmission endpoints.

I'm calling these "endpoints" instead of cut-on/off because of the lack of uniform definition for cut-on/off.

If we are discussing IR-leak or Blue-leak in a stack, then 1% (OD2) is a LOT.
However, if discussing UV transmission only, then we probably don't care about it
until we get to at least a 5-10 % transmission?

Is knowing only where 50% transmission begin enough?
What endpoints are meaningful to you?
How would you designate the absolute bandwith of this stack? At the first hint of any UV transmission at all 304 - 398 nm? Or would you wait until the 5% transmission 317 - 386 nm to designate the absolute bandwidth?

(Sorry this editor does not permit neat lining up of data.)
OD 4.0 .0001 = 0.01% 304 398 nm
OD 3.0 .001 = 0.10% 307 395 nm
OD 2.3 .005 = 0.50% 310 393 nm
OD 2.0 .01 = 1.00% 312 391 nm
OD 1.3 .05 = 5.00% 317 386 nm
OD 1.0 .10 = 10.00% 321 383 nm

Please notify me of typos or arithmetic mistakes. My eyes get glassy after staring at Excel charts and tables for a long while. :D
Andrea G. Blum
Often found hanging out with flowers & bees.