Good Day

[Andy Perrin]

Wow, that may not be a standard UV representation of a sunflower, but it’s a pretty photo!

[Andrea B.]

Here are the raw composites of the two files from the Raw Digger program. A raw composite is a demosaiced file which has had no white balance applied. (Some auto-scaling and contrast have been applied as is typical during demosaicing.) All raw composites look kind of dull like this.

 

But these particular raw comps show us that there is some contamination from unwanted visible or IR light. The program doesn't tell me that. It is rather that I recognize the "look". An uncontaminated raw composite made under a Kolari filter has dull pinkish/magenta tones uniformly across the photo. (I gave you a link to one of those above.) There would be absolutely no raw yellow or raw green in a Kolari UV-pass file.

 

 

[Nemo Andrea]

Forgot to say: On overcast bright days there is sufficient UV to make photos outdoors. And on cloudy bright days, same thing. Exposures just get longer, the more clouds there are. If the sky is really dark/cloudy/rainy, then you might not get much with non-dedicated UV gear. But if it is daytime, you will get something. Temperature has nothing to do with it other than when it is hot, it is summer and we have more UV. In the winter we have less UV outdoors to work with. But still enough to make photos.

 

Unless you are in a cave somewhere in the Netherlands, you can shoot reflected UV there !!

 

Please try my leak-covering experiment mentioned in Post #20. Your problem is not with the sensor. As mentioned, if the problem were with the sensor, then you would get a dull, dark, very noisy photo. The problem is that visible and IR light is not completely blocked.

 

edit: it is HOT here where I am and I am making goofy errors when I write. :D :D :D Pls excuse any typos or ogher stuff. I just wrote an absurdity about how close the sun is to the earth and started laughing and forgot to correct it for a bit. Fixed now.

 

Alright, thats good to know!

[Nemo Andrea]

UV under cloudy skies works fine. Sometimes it gives nicer photos for portraits because it's more diffuse lighting.

 

 

I don't think either of those is likely! We have members in the Netherlands who take UV pics all the time. Also, nearly all sensors have similar UV response once the blocking filter is removed.

 

The most likely explanation is that there is something up with the filter, in my opinion.

 

 

Good to know. Yeah it would surprise me if there was a fundamental difference between CMOS sensors, even older ones, but I suppose sensitivity differences are not unthinkable, but I suppose not to the extent here seen by an order of magnitude.

[Nemo Andrea]

If that is a Sunflower (which of course it appears to be :) ), then you are not recording UV at all or you are not recording very much UV. I say this because there is *no* UV-dark ring at the base of the petals. There is not even a darker yellow or orange colour at the base of the petals.

 

Would you please put a raw version into Dropbox and provide the link? Then I'll run it through Photo Ninja and Raw Digger and see what I come up with. Each UV-pass filter has a characteristic raw histogram. So we will see if your photo has that, or not.

 

When you post a photo for analysis, please also provide aperture, exposure time and ISO value. That info also helps tell us what's going on.

 

Experiment: Try making the sunflower photo on tripod with a dark thick towel wrapped all around the camera and lens mount. Let's see if anything changes at all. You *must* rule out contamination by light leak.

 

Ill will try to do so tomorrow if I can find a sunflower ;) (wont be able to go back to the ones from above). I was thinking that since the colours are at the correct locations, the leak must go through the optics, and cannot be through any side openings or even the lens mount? Then it could only be a faulty filter or perhaps somehow through the moving barrel of the lens, but from my earlier test that seemed to not be significant?

 

Either way, I will be sure to try tomorrow if the flowers permit.

[Andrea B.]

Here is the blue channel of one of the sunflower files. It does show a bit of the UV-signature. In the second version I increased contrast and light just a bit to make the pattern more visible. Remember these are raw composities and raw channels, so they have not had a complete conversion.

 

 

[Nemo Andrea]

Nemo, thank you for permitting me to work with your raw files in order to determine what is going on.

Here is a Photo Ninja conversion of both files which seems to show that you are getting a little bit of UV. But not much.

 

This was a non-standard conversion for reflected UV files. The false-blue and false-yellow are too intense. The false-yellow in these photos should be very dark in a typical reflected UV photo of a sunflower. The tips of the flower should be false-yellow or pale false-yellow. The background foliage should be greyish or maybe have some grey-glue tones.

 

(If you want to try this conversion in your DxO app, then white-click on the tips of the flowers and turn off any camera profiling. Adjust exposure/highlights/shadow sliders as needed.)

 

Now I'm going to run the files through Raw Digger. BRB.

 

 

 

I can replicate this look from the raw files, save for a little bit of extra purple (but thats just a matter of colour profile). The difference between base and tips of the sunflower might also purely be due to the difference in visible (see visible image for reference).

[Andrea B.]

For anyone reading along here as we try to figure out Nemo Andrea's unusual gear behaviour, here is what a Kolari UV-pass raw composite file should* look like. And here are also sunflower photos made under a Kolari UV-pass filter.

 

*When I say "should", take it to mean when shooting UV floral patterns in close work.

 

 

[Nemo Andrea]

Thanks for your help so far Andrea and Andy,

 

I will try to rule out light leak definitively tomorrow. If that doesnt solve the problem, then the most likely candidate would be the filter.

I don't think its worth spending more time on the rather blurry raw files, but you managed to extract quite a bit of information. I should see if I can open these raw files in ImageJ; that should allow me to replicate Raw Digger results and would be a good tool for me to have.

[Andrea B.]

You are very welcome! It is an interesting problem. It will be good to have the knowledge of the solution (if we can find one!) so that help can be offered again sometime to someone else.

[Nemo Andrea]

Alright, I got myself another sunflower and a towel and did the test.

 

Here are the results (white balanced on leaves + slight colour cast change to blue background):

Images were taken at f/4-5.6 at ISO (800) at exposure times around a second.

[ Not Covered ]

 

and

 

[ covered ]

 

The camera and lens were covered with a dense towel. I also did the same using my ''converted'' canon 350D, whose sensor is the direct predecessor of the one in the 1000D (and 400D). It yielded more or less the same result. I covered the whole body and the lens, draping over parts of the lens, but since this is a lens designed for full frame, this is not visible in the images (1000D is APS-C).

 

And the extracted blue channel from the RAW using imageJ: (from the covered case)

 

 

Based on the greens, and lack of noticable difference between the two cases, I think we rule out light leak.

 

Then I suppose there must be something wrong with my filter...

 

Do you think there is any chance it is due to the sensor? I expect that sending the filter back or getting an older nikon is in the same price class.

[Andy Perrin]

No, it’s not the sensor. It’s the filter.

[ulf]

No, it’s not the sensor. It’s the filter.

I agree with Andy, it is definitely the filter.

I have a just slightly younger Canoe EOS 60D that works quite well into UV after modification.

[dabateman]

My thoughts are that the filter maybe leaking. However, you may have been "lucky" and got a better coated Nikon 50mm D lens.

Do you have any other lenses in the sticky page of known good lenses?

Or if you don't have too many lenses, just list them.

Jonathan as shown that the Canon 40mm pancake is good in UV. Do you have that lens?

 

I am thinking you have a good camera, a filter that may be leaking IR and some visible. But also a lens that may only allow 380nm and more through.

[dabateman]

Yes I am saying its both the filter leakage exasperated exacerbated by the low transmission of the lens.

[Nemo Andrea]

Hey dabateman.

 

It would surprise me if they dramatically change the lens coating on such an older lens design without revising the name (considering they already have a more modern replacement), but I suppose its not impossible.

 

I also have two canon other canon lenses:

- Canon EF-S 35mm f/2.8 Macro IS STM (this is a new lens design, so I suspect it will do very poorly)

- Canon EF-S 18-55mm kit lens. (this one might work, I have never used it)

 

I might as well test it. It may still be possible to see some effect. If those lenses are indeed bad for UV, then I should see even less of the faint UV signature I suppose.

 

I will give it a shot just for fun, since I am currently waiting for a response from kolari to hear their thoughts.

[Andy Perrin]

Hey dabateman.

 

It would surprise me if they dramatically change the lens coating on such an older lens design without revising the name (considering they already have a more modern replacement), but I suppose its not impossible.

That’s exactly what happened with the EL-Nikkor 80mm/5.6, though— there are two versions, one metal and one plastic, and only the metal one is good in UV!

[JMC]

Nemo Andrea,

 

The best Canon lens I've come across so far for UV is the Eos 40mm f2.8 pancake lens. Not as good as the earlier uncoated manual lenses, but good enough.

[Nemo Andrea]

So i've repeated the process for the Canon EF-S 18-55mm USM kit lens.

 

[ white balanced ]

and the blue channel

 

 

I'm not sure if much can be concluded from this image. The contrast is a lot lower in the blue channel, but since the sunflower is also slightly more orange near the base of the petals in the visible, I find it hard to conclude if this is due to the visible or UV contribution.

 

I have not been able to find much on the nikon f/1.8D receiving a revision since its introduction in 2002, it had Nikon's Super Integrated Coating since its introduction, but it is theoretically possible the formula has changed. I did buy the lens new, but I still suspect it would not be the lens?

[Andy Perrin]

Yeah, I think you need another filter and some blocking glass (S8612 2mm).

 

That white balance is not anything like a UV one. There are never reds in UV-A photos.

[Nemo Andrea]

I don't have any additional filters at hand, but I figured it might be informative to have a look at the hot mirror that was removed from the camera? So here are the images, white balanced on this white paper cup. Images shot at f/8 with Nikon 50mm f/1.8D, at base ISO (100) on canon 1000D.

 

[ Just the hot mirror ]

 

[Hotmirror + Dust shaker/LP]

 

[Just hot mirror, but NO kolari UV pass filter.]

 

The last one is of course mostly how the hotmirror looks to the human eye. Exposure time without filter was 1/500s, while with the filter it was 2 sec.

 

The colour channels from the raw files show significantly more transmission without the filter (relative transmission), but I would expect less transmission than this, which seems consistent with some unexpected amount of visible light making it through the filter. Do you agree?

 

[ Raw channels, with filter, (just hot mirror)]

 

[ Raw channels, no filter]

[Andy Perrin]

The hot mirror looks normal in visible light. It looks much too light in the filtered shots.

[Andrea B.]

Nemo - In your Post #36 you showed a covered and an uncovered image. May I please look at those in the raw CR2 format? Thank you.

It initially appears to me that there is some light leak in the first photo in Post #36 which is affecting your photos along with the faulty filter. If there were no light leak, then the two photos in Post #36 should look the same after white balancing.

---

 

I want to show you what it looks like when you make a UV photograph in the 380-400 nm range. (I think I have probably posted the following chart somewhere here on UVP in 2018.)

 

The StraightEdge-Gen2 is a filter I tested in 2018. It has a peak very far up in the near-UV band. The white Spectralon portion of the photo on the lower right -- made with the SEU-2 -- shows a distinct raw magenta colour before white balance (as shown in the upper right). And the Color Checker card shows raw violet colours prior to white balance. By contrast, the two white Spectralon photos on the lower left -- from the BaaderU and the KolariU -- show the typical raw orangy-red colour before white balance. And the lower left Color Checker cards show raw orangy-reds also, but they are very unsaturated (dull).

 

Now my point is this: if you happen to be shooting with a lens which only captures UV in the waveband between 380 - 400 nm, then you are going to get magenta and violet raw colours of varying intensity like what is seen in this little chart. You will not see the raw red-oranges which are produced in the 350-365 nm area.

[Nemo Andrea]

Dear All,

 

I have just got word from Kolari.

 

Apparently there has been an issue with a particular production batch of the 52mm filter size. This batch has a '' flawed spectral transmission profile '', which seems to match with what my results are showing.

 

The problem is therefore indeed a flawed filter, specific to this batch. This is why the results are different to the proper performance as observed by Andrea.

 

Kolari has offered to replace the filter when a new batch comes available. This batch is to have a revised coating which should fix the issues.

 

So, the solution to the mystery is a bit boring ;)

 

----

 

When I get the new filter I will make a separate topic to show it off, hopefully with some nice proper looking images. Its a shame I will have to wait to get shooting in UV, but I still have my IR filter to play around with,

 

Thank you for all the help in this post and helping exclude all other options besides the filter.

 

Kolari Vision's response so far has been very polite. I will update this topic if there are any developments that would warrant mentioning (but I suspect that wont be the case)

[Andy Perrin]

Thanks! Mystery solved. I am happy to hear Kolari are doing the right thing. My impression of that company is good.