Double Invisible Vision 308nm stack results (this is pure UV-B IMHO).

[lukaszgryglicki]

Hi, so I've made some tests of the IV308nm x 2 stack, transmission data is here (peak transmission 31.1% at 306nm, minimum distance with the biggest leak around OD6.2 at 500 nm - 502 nm, 1064 nm - 1116 nm, 2nd distance around OD7 for 712 nm - 732 nm): https://docs.google.com/spreadsheets/d/1zMXCT4vtTR6PGf-yzFgUJgUop3gfkpaRAZ50IqN2lsQ/edit#gid=925285003.

 

Overcast skies around noon (Fuji GFX 50R mono quartz + UV-Nikkor + 2 x IV308).

 

7 test exposures on a tripod all at ISO 100:

1) f=11, t=2 minutes.

2) f=16, t=4 minutes.

3) f=22, t=8 minutes.

4) f=32, t=15 minutes (this has some small light leaks in the bottom right corner, I'm happy that they're so small for 15 minutes of exposure in daylight).

5) f=8, t=50 seconds.

6) f=5.6, t=23 seconds.

7) f=4.5, t = 13 seconds.

 

RAW files were processed by AccuRaw Monochrome which is processing RAW data without demosaicing (no Bayer RGGB is applied).

 

First downscaled to 25% and JPEG quality at 75% full images:

- at f=32 (light leak in the bottom right, what is interesting is that there seems to be no or very small diffraction-related degradation at f=32 - in visible light and IR this is a lot more visible, so I'm thinking this is mostly UV-B, this is after saving RAW as JPG without demosaicing): small_32.jpg.

- at f=4.5 (sharp, even if this is wide open UV-Nikkor): small_4_5.jpg.

- at f=11 (11 or 16 is the best aperture when I was checking those files carefully): small_11.jpg.

 

The center or near the center crops (at bigger apertures < 11 looks like the ideal focus was sometimes slightly not in the center - it is very hard to focus this even with EVF magnification because there is so low UV amount to be able to see the image in realtime in magnified EVF):

- center at f=32: center_32.jpg

- center at f=4.5: center_4_5.jpg

- center at f=16: center_16.jpg

 

Corners or very near corners to show that this is also reasonably sharp far away from the focus point. The file name is corner_F.jpg where f is an aperture. Chosen areas near corners where I could see the best details (but never far away from the corner).

- f=32 has lower contrast due to light leak (exposure was 15 minutes in daylight around noon, but f=22 already looks very OK).

 

Summing up - I now think this will be my best UV-B stack. I also think it has waaaay lower shortest UV-A content (compared to Hoya U-340 4mm + single IV308 nm) because already at 340nm transmission is OD7 lower than the peak (31.1% at 306nm) so the signal to leak is OD7 already at 340nm and then almost everywhere, excluding two small ranges with ~OD6.3 (mentioned at the beginning) - not that this is OD from max signal to leak, not from 100%, from 100% it will be about 0.5 more (square root from 10 is about 3.16).

 

The chain in corner images looks best at f=22, at f=32 it seems to finally degrade from diffraction (but also maybe due to light leak and lower contrast due to this?) so aperture f=22 seems to be the best with this UV-B stack.

[colinbm]

Good on you Lukas.
Hope you get some sunny days soon.

[Stefano]

Hopefully that OD 6.2 leak at 500 nm isn't enough to contaminate your images (what really counts is the area more than the peak), if it does show than even a thin U-330-type filter will significantly reduce it.

 

To see if you are seeing slightly deeper in UV, try to image something that has a rapidly changing transmission curve at those wavelengths, like glass. Maybe even 5 nm can make it look noticeably darker. Probably other materials/objects (foliage, windows, cars, etc.) look the same.

 

You have an impressive setup. It makes taking UVB photos seem easy, which normally isn't.

[lukaszgryglicki]

I will stack with U-340 when there is sunlight, I don't have U-330 and don't plan to buy it, so there will also be one more stack U-340 4mm + 2 x IV308. But I *think?* this 2xIV308 (DIV308) is enough.

 

BTW: the actual lowest signal-to-noise is: log10(31.1 / 0.000016) = log10(1943750) = 6.2886.

This is not by area but just peak transmission/peak leak.

 

 

So I will be referring to my stack as UV-B 300-320 ( > 3% transmission), peak 306nm with at least OD 6.28 blocking from 340 nm to 1.2 um. 

 

 

EDIT: did a lot of testing with my other cameras and I'm quite sure leaks from about 500-502 nm,  1064-1116 nm (OD 6.28 signal-to-noise), and 446-462 nm, 712-732 nm (OD 6.89 signal-to-noise) are not contributing at all (or very almost at all), why?

 

- when I close the window and point the lens directly into the window from inside then Fuji 50R mono quartz, Nikon D600 full spectrum mono, and Nikon D600 full spectrum (Bayer) all record nothing at 30 seconds of exposure and max ISO (12800 Fuji, 6400 Nikons). My windows block UV-B - this was already checked with the U-340+IV308 stack.

 

Outside:

- Fuji is barely able to use EVF - very laggy and hard to focus, with a lot of noise. Able to focus when doing it very slowly and takes a long time.

 

- Nikon D600 mono - the live view is all black even outside - Hoya U340 + IV308 stack was displaying a very dark image almost unusable (only outside). If I do the 30 seconds of exposure at ISO 6400 by guess - then the image *IS* recorded very overexposed - so the DIV308 stack (double invisible vision 308 nm filter) can be used by Nikon using guessing or maybe *something will be visible* when there is strong sunlight (today is overcast).

 

- Nikon D600 full spectrum with Bayer CFA untouched - the same pitch black live view image, but 30s ISO 6400 exposure records a little to medium underexposed image mainly in yellow color.

 

I'm now quite convenient this stack is really UV-B only.

 

 

[Stefano]

Well, if you can't see the leak with a 30 s exposure at ISO 12800, it probably won't be an issue.

 

Your image looks yellow, things often look green there. Was that the raw color? It's entirely possible that your camera records UVB in an unusual way.

[lukaszgryglicki]

Your image looks yellow, things often look green there

That was JPG, next time I'll record RAW and see in AccuRaw EXR.

 

BTW: is 310 nm expected to be recorded in green - is this the "magic green area" - especially the 300-320 nm range?

 

I have one more idea to check.

I'll put that stack on Fuji (seems to be recording the best image and the shortest exposure is needed) and replace UV-Nikkor with some normal lens (like Nikkor 50/1.4 AF-D) - I'm expecting not to see anything with 1-minute exposure at ISO 12800 in this case. Then I can check something UV-capable but not dedicated like EL-Nikkor 105/5.6.

[Stefano]

55 minutes ago, lukaszgryglicki said:

BTW: is 310 nm expected to be recorded in green - is this the "magic green area" - especially the 300-320 nm range?

Based on what me and other members got, I would say yes. Here I linked some examples posted on this forum: https://www.ultravioletphotography.com/content/index.php?/topic/4782-hellow-from-poland-minsk-mazowiecki/&do=findComment&comment=59208

 

Based on our collective experience, most cameras record UV wavelengths with essentially the same colors. Sometimes this is not the case, as can be seen here: https://www.ultravioletphotography.com/content/index.php?/topic/5751-revisiting-the-340bp15-on-a-gh-1

 

 

[lukaszgryglicki]

OK so it can be from orange, yellow to green (in most cases), well Nikon's JPEG come out as mostly yellow/orange - will investigate more on a sunny day, I didn't include it here because it was out of focus (focus by guess) - Live View is black.

2nd question - are you sure other source really record 310nm as green - because there are so many factors there and almost all of them put wavelength in a higher number than expected.

My stack should *really* be around 306 nm (peak) offset by declining sunlight maybe to 310nm but not more because transmission drops very very quickly later.

RAW file should tell the truth.

 

EDIT:

also see my parameters, I usually have a LOT more exposure needed, even if I use UV-Nikkor - that (IMHO) means that my blocking is better (sorry but I really think this is the most important)

 

See that I need 4 minutes at f=16 at ISO 100 outside around noon on an overcast day (but not rainly and total dark). Most examples listed elsehere have way shorter exposures indoor. Now I start to think how it is possible?

 

 

 

[Stefano]

As for other sources, I cannot guarantee if the exact peak is in UVB, but it's probably at least close.

 

My UVB photos are done at an aperture probably larger than f/8 (I can't precisely tell on my homemade lenses), ISO 25600 and in a sunny day require at least 8-15 s exposures.

[lukaszgryglicki]

Well I did all at ISO 100.

OK will update on a sunny day.

 

[Andy Perrin]

Stefano, you are also much much closer to the equator. Not sure how much extra UV that's giving you.

[lukaszgryglicki]

OK one more experiment:

Today is a sunny day.

Tried DIV308 on Nikkor 50/1.8 AF-D said (by Kolari) to have a good UV reach.

 

Fuji 50R mono quartz + Nikkor AF-D 50/1.8 at f=1.8, ISO=12800, time=30s - recorded only some fog and some light leaks (looks like internal) - can't distinguis any details of outside scene (daylight in sunlight).

If I use the similar settings as with UV-Nikkor (f=1.8, ISO=100, time=30s) then I record black frame, ISO 12800 with 30s and f=1.8 is really PUSH - Nikon records all possible leaks from everywhere then - even with lens cap on (looks like a one big leak) Fuji just records gray fog then.

 

My conclusion is that when i use UV-Nikkor (that passes UV-B) - I'm getting mostly UV-B image and leaks does not contribute - they start to be visible with f=32 maybe f=22.

 

Also note that I use rafcamera's special lens shade. It has male 52mm thread (to mount to a lens) and then 2 outputs: female 52mm thread (where I mount DIV308) and female shade with 67mm (areound that filter which extends maybe 6cm out of filter and makes a lens shade - this is a custom tool made by rafcamera for me).

If I mount DIV308 directly on the UV-Nikkor then there is a bit less leaks, byut then filter is unprotected and I cannot put lens cap on it or any other filters on it - it has too much glass inside so its external female 52mm thread is unusable, this is why I use rafcamera's shade and put 67mm cap on it (the filter is inside).

 

Will later record Nikon RAW image.

 

[lukaszgryglicki]

Ok so I've recorded Nikon RAW, this still looks yellow to me. Nikon D600 full spectrum with Bayer CFA untouched.

 

Settings:

- Max exposure time possible on Nikon (30s).

- ISO - I had to bump it to 3200.

- UV-Nikkor f=8.

- Focus by guessing multiple times using 3s and ISO=Hi2 (25600).

 

Link to RAW file: https://teststats.cncf.io/backups/div308/raw.nef

 

 

[Stefano]

3 hours ago, Andy Perrin said:

Stefano, you are also much much closer to the equator. Not sure how much extra UV that's giving you.

He lives at 52°N, I live at 45°N, I'm not sure how much difference this makes. You are actually slightly below me at 42°N.

[Stefano]

I opened your raw image in Photo Ninja, removed all ticks, saved as .tif, picked a pixel in Paint in the sky region, and brightened that color to 100%. Here's what I got:

 

(255, 217, 110):

 

That's yellow. Given the strength of your filter stack, as well as your testing, you should be seeing UVB. Maybe your camera simply records it differently than usual. After all, cameras weren't designed to see UVB, so colors may differ.

 

[enricosavazzi]

14 hours ago, lukaszgryglicki said:

[...]

BTW: is 310 nm expected to be recorded in green - is this the "magic green area" - especially the 300-320 nm range?

[...]

The false-green band is quite a bit wider than that. With xenon flash I get a bright yellow with a 330WB80 bandpass filter (which is quite wideband, so most of the recorded wavelengths in practice are around 350-370 nm), a bright green (probably mixed with a little yellow) with Thorlabs FB340-10 and a "greener green" with 325BP10. See Figure 12-13 and Figure 17 at https://www.savazzi.net/photography/uv-pass-filters.html

 

Of course you must use artificial UV sources like non-coated xenon flash tubes. There isn't enough UV at these wavelengths in sunlight, it is typically swamped out by longer UV wavelengths that punch through the tail of these bandpass filters. If the source does not emit enough shortwave, you end up with an image dominated by 360-380 nm even with a narrowband 325 nm-pass filter.

[lukaszgryglicki]

But see https://docs.google.com/spreadsheets/d/1zMXCT4vtTR6PGf-yzFgUJgUop3gfkpaRAZ50IqN2lsQ/edit#gid=925285003 - I can't actually be over 338nm as blocking there is the same as everywhere in longer wavelengths, see that log chart - I *understand* that sunlight will move peak to longer wavelengths than filter peak (306nm) but not much, maybe just to 320-325 IMHO, what do you think?

[enricosavazzi]

11 minutes ago, lukaszgryglicki said:

But see https://docs.google.com/spreadsheets/d/1zMXCT4vtTR6PGf-yzFgUJgUop3gfkpaRAZ50IqN2lsQ/edit#gid=925285003 - I can't actually be over 338nm as blocking there is the same as everywhere in longer wavelengths, see that log chart - I *understand* that sunlight will move peak to longer wavelengths than filter peak (306nm) but not much, maybe just to 320-325 IMHO, what do you think?

As I explained, I think that you can punch through a narrow-bandpass filter with much longer wavelengths than the nominal bandpass. With strong enough VIS, for example, it is entirely possible to punch through a narrow-bandpass 320 nm filter. The out-of-band OD of a filter is never infinite.

[lukaszgryglicki]

Yes, I know that, saw that with a single filter. But this time I hope this really is UV-B only, I have no other ideas to check if there are longer wavelength leaks - all I tried so far indicates that there are no leaks contributing to the final image - even with that color image. I can be wrong of course, but then I would like to know why and how to check this.

Assuming there is no leak in the 500s, 700s, and 1100s then the peak cannot move further than to 336-338nm because blocking there is already the same, so from 306nm filter peak and declining sunlight I still think that peak is somewhere around 318 nm or even at shorter IMHO (looking at the chart).

 

So it looks like 315-320 nm is yellow with Nikon D600.

 

[Stefano]

1 hour ago, enricosavazzi said:

The false-green band is quite a bit wider than that. With xenon flash I get a bright yellow with a 330WB80 bandpass filter (which is quite wideband, so most of the recorded wavelengths in practice are around 350-370 nm), a bright green (probably mixed with a little yellow) with Thorlabs FB340-10 and a "greener green" with 325BP10. See Figure 12-13 and Figure 17 at https://www.savazzi.net/photography/uv-pass-filters.html

Keep in mind that I'm talking about raw colors, which can be very different from white balanced colors. For example, I get raw orange at 340 nm (like other people do), but it becomes green when applying a normal broadband UV white balance.

[lukaszgryglicki]

Right, this is why I've included the RAW image, I did NO white balance at all - didn't touch colors in AccuRaw EXR processing. I will have a hard time white balancing this because what should I white balance to? I have no PTFE and I don't know what is white/gray in UV-B.

 

[Stefano]

I would keep your UVB images in black and white since they are narrowband and there's really no false colors to be seen. White-balancing them may force some colors, but they would probably look noisy.

[lukaszgryglicki]

Actually I only did color images because of:

Your image looks yellow, things often look green there. Was that the raw color? It's entirely possible that your camera records UVB in an unusual way.

 

My original iages were from monochrome Fuji and I will mostly use this stack with Fuji (the only one where I ca actually focus using EVF not by guessing).

 

[Stefano]

Yes, I too would use a monochrome camera to take these images, unless you want to use the same camera to take VIS/UVB comparison photos.

[photoni]

Thanks for sharing @lukaszgryglicki
This is the Raw of the D600 developed with Capture One, I did the white balance on the cloud in the sky. the file is underexposed by 2.5 stops

100% crop

.

 

 

300DPI>100

.