Jump to content
UltravioletPhotography

[Filter Test] BaaderU: Looking for Red Leak


Andrea B.

Recommended Posts

I wondered if the Baader-U had any red leakage, so I stacked it with a Baader UVIR-Cut Filter and Baader 610 Red Longpass Filter. As mentioned in the 092IR Red Leak Test thread, I'm not sure how tight this filter stack is. It is entirely possible some visible light besides red could pass through. Indeed, the 30 second exposure result shows a magenta cast.

ADDED LATER: Magenta as profiled, reddish as shot.

 

Equipment: D600-BB + 105/4.5 UV-Nikkor + Filter Stack( Baader-U, Baader UVIR-Cut, Baader 610 Red-Pass )

White Balance: Preset3 which is an in-camera, visible, white balance preset made under the Baader UVIR-Cut filter in sunlight.

Conversion: Photo Ninja in which a D600-BB visible colour profile was assigned

but no other edits were made except for a bit of blowout control and some sharpening.

Exposure: f/11 for 30 seconds @ ISO-200

Comment: There is something leaking through after 30 seconds, but it is hard to see.

d600whiteStd_uvBaad+uvirCut+redPass610_sun_105uvNikkor_2015.10.06wf_40345pn01.jpg

 

Here is the preceding conversion with the Photo Ninja exposure slider pushed up +3.

It is evident that stacking 3 filters and shooting for 30 seconds in bright sunlight has induced some flare. The light is not at all even across the frame. Remember the Baader-U is dichroic which also doesn't help in this filter stack.

d600whiteStd_uvBaad+uvirCut+redPass610_sun_105uvNikkor_2015.10.06wf_40345pnExp+3.jpg

 

Here is the RawDigger histogram. All R, G and B peaks are between 6-7 EVs below EV0. So we can say that this is not much of a visible leak.

d600whiteStd_uvBaad+uvirCut+redPass610_sun_105uvNikkor_2015.10.06wf_40345histogram.jpg

 

 

I'm trying to make an educated guess about the colour from this histogram.

Can we do that ???

At EV-6, the histogram shows about 25% less green than there is red. Ditto for the blue. So the RGB ratios at E-6 are something like 4::3::3. Would this represent an unsaturated red, that is, a red at about 25% saturation?

In the converted file, the colour appears to have equal parts of red & blue, thus producing a magenta. But white balance is at play in the conversion.

 

Here is the same photo As Shot, converted in Photo Ninja but with no colour additional colour profiling or additional white balance added. Still dark. This does however probably better show the actual leakage?

d600whiteStd_uvBaad+uvirCut+redPass610_sun_105uvNikkor_2015.10.06wf_40345asShot.jpg

 

And the As Shot photo with the +3 exposure push, but still with no additional profiling or wb.

Here you can see what I mean by the reminder in a later post that if you are patient enough you can force almost any light thru a given camera/lens/filter combination. Other colours were beginning to be recorded.

d600whiteStd_uvBaad+uvirCut+redPass610_sun_105uvNikkor_2015.10.06wf_40345asShot+3.jpg

Link to comment

Andrea, do you have considered to use a laser pointer as the only light source?

 

With a laser, you get really monochrome light. When yo run a "normal" WB with "normal" white light, you get the color of the laser (almost). When you do a WB on your reflectance standard (on the picture taken with the laser light only) you get ... a black & white picture (as long as there is no luminescence).

 

I think lasers can help judging the filters as well. I have experimented with a red, a green and a blue laser. With the blue one, you can see the differences of some UV filters close to 400 nm (sorry, no pictures).

 

Many of the difficulties (looking into filter and sensor performance) resulting from the multispectral light sources are easily eliminated with a monochrome light source.

Link to comment

Most laser pointers emit little or no IR--so if the point is to test specifically for IR leakage, that does not help much. There are IR diode lasers, but you won't find them in a pointer. You could use a red laser to test for visible leakage, if you aimed it directly at the camera. The range 610-700 is more visible than IR, so if your IR-cut is good, the images above show mostly red light. This is in accordance with my experience where images clearly showed taillights at night, even though the photons came through thick red plastic and are unlikely to have been UV.

 

To test for IR leakage, you could delete the UVIR cut filter and replace the 610LP with something like an R72 (720 nm) or 093 (820 nm.) Don't worry about any UV leaking through that!

Link to comment

I have IR-leakage tests posted somewhere here on UVP for the BaaderU and other UV-pass filters. Those tests were made (variously) with 092, 093 and gg420.

It has gotten to the point where I should think about indexing our technical posts so that they can be found more easily!!

 

In this test, I was just trying to see if any high red got through the BaaderU. I would say there was not any significant red if the photo was that dark after a 30" exposure. We could quibble with the test being at f/11 instead of more open, but I typically shoot UV floral signatures at f/11 and so was being practical. :)

Link to comment
What was the light source/intensity for this test? I find it difficult to assign a significance to this if I do not know that. And why does the color cast matter, for these purposes?
Link to comment

Clark, I was outdoors in strong summer sunlight with my back to the sun and the standards facing it.

 

The visible colour cast under that filter stack is shown under a proper in-camera white balance.

So if I see the colour +big(R )+little(g)+little(b ), which is basically a desaturated, dark red that tells me I PROBABLY have a red leak, but only a very very small one if the visible photo is that dark after 30 seconds.

 

This is not exactly a rigorous experiment, and we cannot (nor must not) read too much into it. A rigorous experiment would be conducted in the laboratory with proper spectral equipment. As a photographer, however, I am concerned about the actual effect on my photos if a filter leaks outside its stated range. I can conclude from this trial that I don't need to really worry about visible Red contamination with a BaaderU when making actual photographs. We pretty much already know that but it is always good to see for oneself.

 

Keep in mind that you can force almost any wavelength through a camera/lens/filter combo if you are patient enough. At what point does it become significant that some wavelength leaks through? Knowing I rarely make UV exposures with the BaaderU over 15 seconds long, I'm happy stopping there. But I took it out to 30 seconds just to see what would happen.

 

I'm always open to suggestions about how to better test filters under actual field conditions !!

 

[Added: I forgot to mention that the flare is slightly misleading the colour interpretation. I couldn't seem to get rid of it. But that's a known hazard with shiny filter stacks.]

 

I'll put a couple of more photos up there to make things clearer.

Link to comment

I don't think I wrote this one up very clearly!

Sorry 'bout that. I'll try to do better next time. :)

 

****

 

Werner, yes a monochrome light beam would be useful were I to try this again. "-)

Link to comment
The UV to visible ratio is very favorable in sunlight. For some of my night exposures, it seems less so, and I am doing longer exposures; so perhaps that is why I notice certain things.
Link to comment
  • 1 month later...

Problems like this one make me wish I still had access to a good spectrophotometer.

 

Has anyone tried an LED array? That is just stick a series of LEDs of known spectra and (hopefully) balanced intensities in a breadboard. Then, in a light controlled environment take a series of pictures with various neutral density filters, say OD 0,1,2,3,4 and compare the images to that obtained with the filter (s) in question. That should not only tell you which wavelengths are leaking but how bad the leak is.

Link to comment

This information exists and is presented in the filter's transmission chart from the manufacturer.

 

However, it is sometimes difficult to get a detailed transmission chart.

 

For a BaaderU, one alternative is to look at its substrate DUG11X chart. (.....did I get those letters correct??.....)

Link to comment
Transmission charts, due to their usual format, are not the best for evaluating out-of-band leakage: .1% and .0000001% look just the same on the graph, but the practical difference can be very large if the in-band signal is weak.
Link to comment
true, but I meant the good spec charts not the marketing junk they typically show in adverts. "-)
Link to comment

LED's are not a good method for this kind of test. They have too strong of intensity in many cases, unnaturally strong, and also have their own out of band emittance even though their specifications may show a very narrow band.

A more natural and accurate method of testing is to stack the U filter with longpass filters to incrementally isolate a leak to above or below the cutoff points of different longpass filters,

using unnaturally long exposures to force the exposure of extremely low transmissions, using whatever lighting you use in real situations.

 

For example just now I tested using a D7000 full spectrum camera, Kuribayashi 35mm, Baader U stacked with GG435 on the lens, and an MTE torch, using live view,

I could see some small point of light, actually the same both with/without a 2mm U-340 torch filter added (which would change more with/without added torch filter using GG400).

I performed the same test using U-360 2mm + BG40 2mm in place of the Baader U and saw the same small point of light.

I then performed the same set of tests replacing the GG435 with RG610, and saw no point of light with either U filter/stack, and with/without added torch filter.

This really only illustrates one thing to me, that there is some out of band light emittance by the MTE/Nichia, but this is not the LED one would pick out for such a test.

With other LED's you will see more, but again this is not a good method of testing.

 

A Baader U leaks at a higher point, so its leak should look like a 'white' leak.

Shane has the only graphs that I am aware of, using a spectrometer, showing Baader U leak ranges:

http://www.beyondvis...BV3-filter.html

 

Absorptive U filter stacks (such as UG1/U-360/UG11/U-340 + BG40/S8612/BG39) have their peak Red/IR leak centered at the 720nm (UG1 and U-360) and 750nm (UG11 and U-340) points,

therefore the leak color will appear more red/brown, however when adequately suppressed (below 1E-03/1E-04) these have similar suppression as the Baader U leak in the 900nm range.

 

Diabatic transmission graphs show the transmission, the leaks, and suppression in more detail than a linear graph.

I have not seen diabatic graphs published for Dichroic filters, and I have not seen per nanometer data published for such filters either.

Linear graphs are nice, but diabatic graphs reveal the hidden truth. There is a lot you can hide below the line of a linear transmission graph alone.

Schott provides the most inclusive transmission data, Hoya is far behind Schott in that regard, but still second best.

Such data is extremely useful for intelligent calculation of expected results, showing where and how strong Red/IR suppression is for any part of the transmission range.

 

Photo 1

Perhaps I am contradicting what I said above in showing this old test using a typical remote control LED (which transmits in the range of the Baader U leak, but not in the range of the stack leak),

but this test illustrates the difference between the leak range of the stacked U and Baader U. Note the extended exposure times and stacked 610nm longpass filter.

Settings: D7000 UV/IR, Kuribayashi 35mm f/3.5 lens, @f/3.5, Aperture Priority, +2.0EV, ISO 640, Spot, Live View.

These use the same un-optimized in-camera white balance.

 

post-87-0-09803900-1449457845.jpg

 

 

Photo 2

A second test of the same kind.

Note: UG11 2mm is used in these tests, with S8612 1mm (equivalent Red/IR suppression as BG40 2mm).

Settings: D7000 UV/IR, Kuribayashi 35mm f/3.5 lens, @f/3.5, Aperture Priority, +1.0EV, ISO 640, Matrix, Live View.

These use the same un-optimized in-camera white balance.

 

post-87-0-46008100-1449457938.jpg

 

 

Photo 3

A third test of the same kind.

Settings: D7000 UV/IR, Kuribayashi 35mm f/3.5 lens, @f/5.6, Manual, 0EV, ISO 640, Matrix, Live View.

These use the same un-optimized in-camera white balance.

 

post-87-0-87191600-1449465729.jpg

Link to comment

Steve, good tests which also illustrate the point that it is good to test filters in an actual shooting situation (imho, of course).

 

*********************

 

In the 2nd set of tests you have the first row BaaderU stacked with only the Red Longpass filter.

But the second & third rows have the UG11 stacked with an IR blocker in addition to the Red Longpass.

This may have been what you intended?

But the results between the BaaderU and the UG11 are therefore not comparable?

Not a criticism, simply an observation - because I think I was looking for the same type 'leak' in the second & third rows and was temporarily confused as to why I didn't see it.

 

I should try to find my BaaderU IR leak tests.

Did that. They are in this post.

http://www.ultraviol...-precision-uold

 

ADDED: It is time to recall

Bjørn's Filter Maxim: Given long enough exposure, any filter will show leaks outside its designated passband.

Although I will mention that I had a darned hard time getting through the D810 internal filter to record UV. Only got a muzzy mess.

Bjørn's Lens Maxim: Given long enough exposure, any lens will pass some UV or IR light.

Link to comment

LED's are not a good method for this kind of test. They have too strong of intensity in many cases, unnaturally strong, and also have their own out of band emittance even though their specifications may show a very narrow band.

A more natural and accurate method of testing is to stack the U filter with longpass filters to incrementally isolate a leak to above or below the cutoff points of different longpass filters,

using unnaturally long exposures to force the exposure of extremely low transmissions, using whatever lighting you use in real situations....

 

...Diabatic transmission graphs show the transmission, the leaks, and suppression in more detail than a linear graph.

 

 

I hadn't heard heard the term "diabatic" to describe a semi-log graph before. Interesting the things I'm learning here.

 

I'm not sure that filters are much better solution than LEDs. I'm having a hard enough time to get a diabatic graph from the manufacturer of my UV filter and as you point out Schott is one of the few who do so. Unless one has a toolkit of Schott filters one has to make do.

 

Does anyone on the forum have access to a decent spectrophotometer and be willing to be a tester? If members are willing to part with a filter for a week or two to be tested and have the results published on this forum in a diabatic graph that would be a useful resource.

Link to comment

Schott uses that term 'diabatic' to describe a particular kind of log graph which shows peaks & leaks so nicely. Schott points out somewhere that it is not a typical usage of the word.

 

For about half a year I read it as 'diabetic' and was rather cornfuselled by it all. :lol: :rolleyes: :D

 

I cannot seem to find the Schott reference to 'diabatic' tonight, but it will show up eventually.

Link to comment
The diabatic Schott scale is (1-log(log(1/t)) where t=transmittance. So has the property of allowing a very wide range of values into a single graph.
Link to comment

where did you find that? I'll go look again.

 

So it is just a loglog graph and we get a double expansion - one increasing upward for better illustration of the peaks

and one increasing downward for better illustration of the leaks.

Expansion apparently around .5 so they are using base 10.

 

ok, cool.

Link to comment

See the definition given in section 3.2 Spectral diabatie, page 16-17 of this Schott brochure on Interference Filters & Special Filters.

 

The generic definition of diabatic indicates an origin from Greek diabat (ós) able to be crossed, fordable (equivalent to dia- dia- + batós passable, verbal adjective of baínein to walk, go). My German is extremely limited but I suspect the term has something to do with the "folded" nature of the scale in which "both the passband (at high transmittance) as well as the blocking band (with low transmittance) are stretched."

 

I prefer the simple simi-log format.

Link to comment

I JUST now found that exact a similar link!! Thanks John.

 

Schott Optical Filters 2015 Catalog Complete English (PDF)

Page 77 contains the discussion about the transmission charts:

 

It is advisable to use a derived form of the spectral transmittance, the so-called spectral diabatie Θ(λ). The spectral diabatie is defined as

Θ(λ) = 1 - lg{lg[1/t(λ)]}

where “lg” denotes the logarithm to base 10. The diabatic form offers a significant advantage over the linear form: both the passband (at high transmittance) as well as the blocking band (with low transmittance) are stretched. Thus both the passband and the blocking band can be seen clearly.

 

Or as annedi says "the peaks & leaks". :lol:

 

Light Measurement Handbook: http://homepages.inf.../RYER/ch09.html

The handbook tells us that:

 

The Diabatie scale is a log-log scale used by filter glass manufacturers to show internal transmission for any thickness. The Diabatie value, q(l), is defined as follows according to DIN 1349:

q(l) = 1 - log(log(1/t))

Linear transmission curves are only useful for a single thickness. Diabatie curves retain the same shape for every filter glass thickness, permitting the use of a transparent sliding scale axis overlay, usually provided by the glass manufacturer. You merely line up the key on the desired thickness and the transmission curve is valid.

Link to comment

http://dictionary.re...com/browse/dia-

dia in greek

A prefix occurring in loanwords from Greek (diabetes; dialect) and used, in the formation of compound words, to mean “passing through” (diathermy), “thoroughly,” “completely” (diagnosis), “going apart” (dialysis), and “opposed in moment” (diamagnetism).

 

As a prefix

1. through, throughout, or during: diachronic

2. across: diactinic

3. apart: diacritic

4. (in botany) at right angles: diatropism

5. in opposite or different directions: diamagnetism

 

"passes through" or "in opposite directions"

So diabatic graph passes through that horizontal line on the y-axis where stretching is in opposite directions on either side of that line.

 

ok, whew ! Enough already. Go make photos.

Link to comment

Schott uses that term 'diabatic' to describe a particular kind of log graph which shows peaks & leaks so nicely. Schott points out somewhere that it is not a typical usage of the word.

 

For about half a year I read it as 'diabetic' and was rather cornfuselled by it all. :lol: :rolleyes: :D

 

I cannot seem to find the Schott reference to 'diabatic' tonight, but it will show up eventually.

 

You and Google both :lol:

Link to comment

The diabatic Schott scale is (1-log(log(1/t)) where t=transmittance. So has the property of allowing a very wide range of values into a single graph.

 

Bjorn is correct.

http://ftgsoftware.c...es_diabatic.htm

 

The diabatic graph is available in the Schott filter program, as well as linear graphs and 'normalized' combined graphs.

The Schott filter glass catalog uses the diabatic graph format, not linear.

If you have a data sheet for any filter brand / glass type, not just a graph, but actual data per nanometer, or per every so many nanometers (such as with Hoya data sheets that provide every 10nm data),

then you can enter that data into the Schott program as an added filter type. However it can be a bit more complicated and limiting if you only have T data to enter (as apposed to Ti data),

unfortunately the Optima data sheets I have found don't provide Ti data, so you may find some workaround for that, but combined graphs may pose a problem.

Hoya data sheets provide both T and Ti data at every 10nm, but also have long data dropout ranges, like for most of the visual range with Hoya U glass, meaning there is NO data for most of that range.

Schott data is already in the program for most of their filter glass types, at 1nm resolution.

You need to enter the Ti data per resolution of the data sheet, the reference thickness, and reflection factor for those glass types you wish to add.

You can then generate diabatic graphs.

It can take a while, so like Bjorn use to say, put on your favorite CD... ;-)

If you want to more precisely calculate what to expect with filter stacks, then entering glass data into the Schott program is like having a flashlight in the dark.

Link to comment

Here is another comparison of BG glass, all 2mm thick, T (not Ti), linear graph.

There seems to be very little difference between QB21 and QB21S, other than mostly visual range amplitude, which is not a concern with UV-only stacking.

Keep in mind, any increase in thickness to suppress Red/IR will also suppress UV, requiring longer exposure time and reduced UV bandwidth.

Also note that Optima refers to QB21 as BG38 equivalent, and most glass sites also show it as BG38 equivalent,

however it looks like QB21 and QB21S should have stronger Red/IR suppression than BG38 according to their graph.

I think this is a fair comparison and as exact as I can make without better Ti data.

Also note, that what looks like unaligned lines on the graph, are simply the difference between 25nm increments being used on the QB21S graph, and 20nm increments being used on all other graphs.

 

post-87-0-63051200-1449717267.jpg

Link to comment

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
×
×
  • Create New...