• Ultraviolet Photography
  •  

Build thread - at home measurement of camera UV spectral response

107 replies to this topic

#81 JMC

    Member

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

Posted 15 July 2018 - 19:08

Thanks Andrea.

#82 dabateman

    Da Bateman

  • Members(+)
  • 1,309 posts
  • Location: Maryland

Posted 16 July 2018 - 01:28

This is great Jonathan!
I recently got a Pentax UAT lens and 25mm filters for band pass uv subject tests. Mainly to see if differences in uv reflected light.
I have the 313bp25, 340bp10, 370bp15, 390bp25 and 405bp10 filters. These I mounted in 25mm filters, except the 405nm I glued in a 28mm. I used 3m brand crazy glue and it will oxidize the filter surface. As the glue cures. I was hoping not, but observe this previously when mounting ZWB2 filters into 199a front adapter. So every couple hours I needed to clean it with just lens cleaner.
I wouldn't recommend that type glue to anyone. Too much hazel. Butter to just solder the filter in place.
With the SD14 I see nothing with the 313bp25 filter. I get good response with 340bp10 and above.

I don't think its the dyes that are limiting the less than 340nm wavelengths, but rather the glass microlenses on the surface of the sensor. However the only way to know that would be to find someone who could shave them off a Sigma camera. Best way to get a colour monochrome camera. That is monochrome in uv and IR response, but colour in the visible.

#83 OlDoinyo

    Member

  • Members(+)
  • 688 posts
  • Location: North Carolina

Posted 16 July 2018 - 03:39

Has anyone tried one of these filter arrays on a film camera? I am almost certain that one could breach the 340 barrier handily with b&w film. These results tell me that repeated failures with digital cameras below 300nm may be as much due to reseau or microlens issues as they are due to fundamental sensor response or lighting issues (though the latter factors undoubtedly play some role.)

View Postdabateman, on 16 July 2018 - 01:28, said:

Too much hazel. Butter to just solder the filter in place.

I assume some typos there--but it sounds tasty.

Edited by OlDoinyo, 16 July 2018 - 03:43.


#84 Andy Perrin

    Member

  • Members(+)
  • 2,684 posts
  • Location: United States

Posted 16 July 2018 - 05:10

OlDoinyo, it's my understanding that the Bayer filter is literally bonded to the sensor -- it is actually made of photoresist deposited using the same photolithographic methods as the rest of the sensor. You can't remove it without destroying it (and it's a wonder they manage not to destroy the sensor also).

Quote

Diazonaphthoquinone (DNQ)-novolac photoresist is one material used as the carrier for making color filters from color dyes. There is some interference between the dyes and the ultraviolet light needed to properly expose the polymer, though solutions have been found for this problem.[8] Color photoresists sometimes used include those with chemical monikers CMCR101R, CMCR101G, CMCR101B, CMCR106R, CMCR106G, and CMCR106B.[9]
A few sources[1][10] discuss other specific chemical substances, attending optical properties, and optimal manufacturing processes of color filter arrays.

For instance, Nakamura said that materials for on-chip color filter arrays fall into two categories: pigment and dye. Pigment based CFAs have become the dominant option because they offer higher heat resistance and light resistance compared to dye based CFAs. In either case, thicknesses ranging up to 1 micrometre are readily available.[1]

Theuwissen says "Previously, the color filter was fabricated on a separate glass plate and glued to the CCD (Ishikawa 81), but nowadays, all single-chip color cameras are provided with an imager which has the color filter on-chip processed (Dillon 78) and not as a hybrid."[10] He provides a bibliography focusing on the number, types, aliasing effects, moire patterns, and spatial frequencies of the absorptive filters.

Some sources indicate that the CFA can be manufactured separately and affixed after the sensor has been manufactured,[11][12][13] while other sensors have the CFA manufactured directly on the surface of the imager.[13][14][15] Theuwissen makes no mention of the materials utilized in CFA manufacture.

Quote

I don't think its the dyes that are limiting the less than 340nm wavelengths, but rather the glass microlenses on the surface of the sensor.
That is certainly possible (I always forget the damn microlenses exist). Why do you think it's the microlenses and not the dyes, though? I mean, how would you tell the difference?

Edited by Andy Perrin, 16 July 2018 - 05:17.


#85 JMC

    Member

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

Posted 16 July 2018 - 08:27

Answering the CFA vs Microlenses question will be tricky. However anyone have a friend with a Leica Monochrom camera? Apparently that has the micro lenses on it;

https://www.imaging-...ica-m-monoA.HTM

Be interesting to know how far into the UV that one could see.

EDIT - Actually scratch that. If the sensor has the IR absorbing S8612 filter on it, then it wont go down as far as 300nm anyway.

Edited by JMC, 16 July 2018 - 09:39.


#86 dabateman

    Da Bateman

  • Members(+)
  • 1,309 posts
  • Location: Maryland

Posted 16 July 2018 - 10:35

@oldoinyo, yes sorry for the typos. But since your comment made me laugh. I will not fix them now.
@Andy,
I assume the micro lenses are the issue based on the Sigma cameras having the same response problem. The Sigma cameras have micro lenses, but no dye layers.
However this requires that the same silicon response occurs between the Sigma cameras and other. Which may not be a fause approximation. If MaxMax could shave off the glass microlenses on a Sigma Camera then you could directly compare the effect of the microlenses.

#87 JMC

    Member

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

Posted 16 July 2018 - 19:53

I thought it might be interesting to play around with exposure time and see how the different filters come into play. I took my Multispectral 5DSR and Rayfact 105mm lens (f5.6, ISO400) no filters on the lens, and took a sequence of shots between 1/4s and 8s exposure. This is what they look like as RAW composites in RawDigger (top row L to R, 303nm, 321nm, 341nm, middle row L to R, 355nm, 364nm, 382nm, bottom row L to R, 396nm, 404nm, 405nm).
Attached Image: 1a quarter second.jpg

Attached Image: 2a half second.jpg

Attached Image: 3a 1 second.jpg

Attached Image: 4a 2s.jpg

Attached Image: 5a 4s.jpg

Attached Image: 6a 8s.jpg

As exposure time goes up, the longer wavelength filters obviously let in so much light that the camera whites out, but we do start to see something even at 321nm. I get the feeling to see anything down there, you'd need extremely long exposures, and very very good blocking of other wavelengths. I don't think OD4 blocking would be good enough any more if you wanted a pure 320nm image.....

The filter transmission curves are shown below.
Attached Image: Sparticle filters.jpg

Interestingly the 303nm to 382nm filters all let roughly the same amount of light through. So these exposure pictures above give an idea of how the sensor sensitivity (and Bayer filter/microlens transmission) drops with decreasing wavelength, especially when looking at the top two rows of circles in each image.

#88 OlDoinyo

    Member

  • Members(+)
  • 688 posts
  • Location: North Carolina

Posted 16 July 2018 - 22:38

Lest my earlier comment be misunderstood, when I referred to a filter array I did not mean the reseau bonded to the sensor. I meant the device JMC constructed as described above--but I wondered what would happen if one loaded a film body with b&w film and then mounted it on said device for testing.

I am skeptical of the role of the glass microlenses because they are perforce extremely thin---much thinner than a microscope cover slide. It would be difficult to get much attenuation in a pane that thin without some special doping of the glass.

#89 Andrea B.

    Desert Dancer

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

Posted 17 July 2018 - 00:49

OK, now that JMC has shown the 303 to 382 nm filters all have roughly (very) the same transmission, then I'm gonna say that it is simply the relative lack of 300 and 320 nm UV in sunlight as compared to the amount of 360, 380 nm UV in sunlight.

(Must note here that even that small 5% difference in transmission rates on your chart can mean a LOT in that 300-340 nm range.)

That the 340 filter does not show up in JMC's photos until 8 seconds does not surprise me. Here are two 340 nm shots with the 340fwhm10 which transmits at an astounding 95%+. These are just total-junk, oof setup shots, but the long exposures make the point. The 2.5" shot over the water was in good sunlight. The 15" photo of the path up the hill from the deck was in hazy sunlight. Both f/5.6 @ ISO-100.

You need time to capture 340 nm sunlight. You need even more time to get shots in 320 or 330 nm sunlight.

Jonathan, thank you for satisfying my curiosity by showing the transmission rates and making those longer exposures. :D

I was junking around trying to figure out which helicoid would be best for the 105 EL.
Attached Image: 610_5292pn.jpg

Attached Image: 610_5299pn.jpg
Andrea G. Blum
Often found hanging out with flowers & bees.

#90 Andy Perrin

    Member

  • Members(+)
  • 2,684 posts
  • Location: United States

Posted 17 July 2018 - 02:17

Quote

OK, now that JMC has shown the 303 to 382 nm filters all have roughly (very) the same transmission, then I'm gonna say that it is simply the relative lack of 300 and 320 nm UV in sunlight as compared to the amount of 360, 380 nm UV in sunlight.
Then how do you explain how much brighter they are in the monochrome camera test? The light is the same, the only difference is the microlenses and Bayer, right?

#91 dabateman

    Da Bateman

  • Members(+)
  • 1,309 posts
  • Location: Maryland

Posted 17 July 2018 - 08:13

Jonathan,
Can you run a similar exposure series using your SD14? I would be interested if its similar to the full spectrum Canon.
Thanks,
David

#92 JMC

    Member

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

Posted 17 July 2018 - 08:19

View Postdabateman, on 17 July 2018 - 08:13, said:

Jonathan,
Can you run a similar exposure series using your SD14? I would be interested if its similar to the full spectrum Canon.
Thanks,
David
Should be able to David, but it may not be until next week, as I have a few things to sort first. I also need to check whether I have got a Sigma to Nikon adapter yet - I ordered one without a lens in it ages ago, so I can use the Rayfact lens but I'm not sure whether it has arrived yet. If not then I'll have the use the UAT lens on them both to make it a fair comparison.

EDIT - I'll have to use the UAT anyway, as the sensor on the SD14 is smaller. All cloudy here at the moment so will need those blue skies back....

Edited by JMC, 17 July 2018 - 09:44.


#93 dabateman

    Da Bateman

  • Members(+)
  • 1,309 posts
  • Location: Maryland

Posted 17 July 2018 - 10:30

View PostJMC, on 17 July 2018 - 08:19, said:


I also need to check whether I have got a Sigma to Nikon adapter yet - I ordered one without a lens in it ages ago, so I can use the Rayfact lens but I'm not sure whether it has arrived yet. If not then I'll have the use the UAT lens on them both to make it a fair comparison.

EDIT - I'll have to use the UAT anyway, as the sensor on the SD14 is smaller. All cloudy here at the moment so will need those blue skies back....

What Nikon to SA adapter will you use?
I have a Nikon to M42 adapter and then use a M42 to SA mount adapter. But only good for macro.
In your case with this light test I don't think that would matter nor would the 1.7x crop factor. Or does you wood box not allow for the filters to fit in full view of the Sigma?

Anyway thank you. I look forward to a direct comparison. Mainly to just tell me how good my full spectrum E510 camera is. I get 4 stops better at 340bp10 and 1 stop better at 370bp15 with the SD14 than the E510. I never found out how it was converted as bought super cheap used. But with my 1000 lines/mm grating, I do see green spectal light with Baader venus filter and sun as a light source and the 340bp10 works, so can't be a horrible conversion.

Edited by dabateman, 17 July 2018 - 10:34.


#94 JMC

    Member

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

Posted 17 July 2018 - 12:49

View Postdabateman, on 17 July 2018 - 10:30, said:

What Nikon to SA adapter will you use?
I have a Nikon to M42 adapter and then use a M42 to SA mount adapter. But only good for macro.
In your case with this light test I don't think that would matter nor would the 1.7x crop factor. Or does you wood box not allow for the filters to fit in full view of the Sigma?

Anyway thank you. I look forward to a direct comparison. Mainly to just tell me how good my full spectrum E510 camera is. I get 4 stops better at 340bp10 and 1 stop better at 370bp15 with the SD14 than the E510. I never found out how it was converted as bought super cheap used. But with my 1000 lines/mm grating, I do see green spectal light with Baader venus filter and sun as a light source and the 340bp10 works, so can't be a horrible conversion.
I'll have to use the UAT for both, so just need the M42 adapter on both cameras. When I tried with the Rayfact the cropped sensor on the Sigma meant I couldn't see all the filters - something I didn't consider when making the box was cropped sensors, and needing to move further back to see all the filters. I actually have like you Nikon to M42 (but with no lens) and then M42 to Sigma. Will let you know when I have something.

#95 JMC

    Member

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

Posted 17 July 2018 - 14:00

View PostJMC, on 17 July 2018 - 12:49, said:

I'll have to use the UAT for both, so just need the M42 adapter on both cameras. When I tried with the Rayfact the cropped sensor on the Sigma meant I couldn't see all the filters - something I didn't consider when making the box was cropped sensors, and needing to move further back to see all the filters. I actually have like you Nikon to M42 (but with no lens) and then M42 to Sigma. Will let you know when I have something.

David, but a of rush here, but I managed to get it done. I have 2 sets of images. 1. EOS5DSR multispectral conversion, 85mm UAT lens (f5.6, ISO400, exposure from 1/4 to 8s). 2. SIgma SD14, hot mirror removed, 85mm UAT lens (f5.6, ISO400, exposure from 1/4s to 8s). Sparticle box pointed at blue sky away from the sun.

EOS - full sized raw files, Sigma SD14 - low res raw files (to reduce noise). Both opened in RawDigger as RAW composite files. Black level set of 15 for all the EOS shots. Black level set to just below the average black area with the SD14 files - the black level changed with longer exposure with the Sigma files, so everything became washed out as the exposure got longer. To be honest I am not sure whether this is the right way to deal with the change in black I saw with for SD14 files, but it makes the visual comparison easier.

In the image;
Top row, left to right, 303nm, 321nm, 341nm
Middle row, left to right, 355nm, 364nm, 382nm
Bottom row, left to right, 396nm, 404nm, 405nm

Firstly the EOS5DSR multispectral camera images;
Attached Image: 2 Quarter second.jpg

Attached Image: 3 half second.jpg

Attached Image: 4 1s.jpg

Attached Image: 5 2s.jpg

Attached Image: 6 4s.jpg

Attached Image: 7 8s.jpg

Now the Sigma SD14 images;
Attached Image: 2 quarter second.jpg

Attached Image: 3 half second.jpg

Attached Image: 4 1s.jpg

Attached Image: 5 2s flattened black.jpg

Attached Image: 6 4s flattened black.jpg

Attached Image: 7 8s flattened black.jpg

Overall I think there is lower sensitivity for the SD14 compared to the multispectral EOS5DSR based on the brightness of the circles at the same exposure times. More striking than that though is the Sigma sensitivity to UV is basically all in the blue channel, unlike the 5DSR which shows different colours at different wavelengths.

#96 Andrea B.

    Desert Dancer

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

Posted 17 July 2018 - 14:29

Andy Then how do you explain how much brighter they are in the monochrome camera test? The light is the same, the only difference is the microlenses and Bayer, right?

Yes, agreed. Probably. :lol: But if one is going to use a Bayered conversion, then exposures will certainly be longer to get below 340 nm. That's all I was tossing out there.
I have gotten a noisy, hazy recording around 300nm with the old D600. So I think that 320 or 330 is not impossible. Just difficult.




I think maybe I ought to test this 320/330 thing for everyone, but I can't figure out what 320 or 330 filter to get. Edmund makes those narrowband (10 nm) filters but there is that untidy little business of unwanted reflections and flare from the very mirrored surfaces. Even rear mounting is not entirely doing away with some veiling. I had to hammer the contrast on that blue & yellow thing posted up there.




Jonathan, great experiments. Thanks for posting here!
Andrea G. Blum
Often found hanging out with flowers & bees.

#97 Andrea B.

    Desert Dancer

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

Posted 17 July 2018 - 14:30

Andy Then how do you explain how much brighter they are in the monochrome camera test? The light is the same, the only difference is the microlenses and Bayer, right?

Yes, agreed. Probably. But if one is going to use a Bayered conversion, then exposures will certainly be longer to get below 340 nm. That's all I was tossing out there.
I have gotten a noisy, hazy recording around 300nm with the old D600. So I think that 320 or 330 is not impossible. Just difficult.

I think maybe I ought to test this 320/330 thing for everyone, but I can't figure out what 320 or 330 filter to get. Edmund makes those narrowband (10 nm) filters but there is that untidy little business of unwanted reflections and flare from the very mirrored surfaces. Even rear mounting is not entirely doing away with some veiling. I had to hammer the contrast on that blue & yellow thing posted up there.
Andrea G. Blum
Often found hanging out with flowers & bees.

#98 Dmitry

    Dmitry K.

  • Members(+)
  • 236 posts

Posted 17 July 2018 - 16:17

I have 300-330nm filter with peak at 315nm, but have no proper lens to check it. It have IR leak less than 1%, but its enough to ruin UV image.

#99 dabateman

    Da Bateman

  • Members(+)
  • 1,309 posts
  • Location: Maryland

Posted 18 July 2018 - 07:23

Out standing, thank you Jonathan. The SD14 seems to be about 2.5 stops slower than a correctly full spectrum converted camera.
So this means my E510 is horrible and getting a proper conversion done will greatly improve my exposure settings.

Now for me to think which would be best, getting my Nikon DF converted with best over all sensor or my Olympus EM1 with capabilities to use any lens?
I tested the UAT on the DF today and with M42 to nikon adapter I can focus to 20 feet. The 1mm or so adapter issue didn't impact it as much as I thought. I just don't know if the DF is plaqued with the IR shutter monitor problem.
Another advantage of the EM1 is upto 3 hour exposure time is allowed in live composite mode. I could try my 313bp25 filter overnight with a compact fluorescent black light. I wonder what the noise would look like.

#100 JCDowdy

    JCDowdy

  • Members
  • 1,165 posts
  • Location: Arlington, TN

Posted 18 July 2018 - 21:13

View PostJMC, on 15 July 2018 - 12:37, said:

..... - this amused me, I needed a tube 12-12.5mm external diameter, so I cut down a 45ACP shell case as it was the right diameter (I'm one of the few people in the UK that still shoots so have a whole box full of these).

Bravo! I understand that is not so easy in the UK.
Seems we have more that UV photography and photobiology interests in common.