Blog on UV and other aspects of microscopy and imaging

Here are links to articles for your reading pleasure. You may also submit brief reviews or discuss the contents of the articles.

Moderators: Pau, rjlittlefield, ChrisR, Chris S.

jmc
Posts: 96
Joined: Fri Apr 17, 2020 3:14 am

Blog on UV and other aspects of microscopy and imaging

Post by jmc »

On my site I keep a blog going covering various aspects of my work on UV microscopy, my microscope refurbishment, and other bits including my research in camera spectral sensitivity. Here's the link in case people fancy a read;

https://jmcscientificconsulting.com/blog/
Jonathan Crowther

Lou Jost
Posts: 4580
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Re: Blog on UV and other aspects of microscopy and imaging

Post by Lou Jost »

This is a very nice resource; I learned a lot from your articles on silica slides and on Luminars for UV!

jmc
Posts: 96
Joined: Fri Apr 17, 2020 3:14 am

Re: Blog on UV and other aspects of microscopy and imaging

Post by jmc »

Thanks Lou. As and when I write anything new in the future I'll put a post on this thread and give the details of what the article is about.
Jonathan Crowther

Lou Jost
Posts: 4580
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Re: Blog on UV and other aspects of microscopy and imaging

Post by Lou Jost »

That will be much appreciated by many of us.

jmc
Posts: 96
Joined: Fri Apr 17, 2020 3:14 am

Re: Blog on UV and other aspects of microscopy and imaging

Post by jmc »

No problem Lou.

Today's piece, looking at leakage of UV through stock camera sensor filters, its effects and how it can be rectified.

https://jmcscientificconsulting.com/uv- ... r-filters/
Jonathan Crowther

Lou Jost
Posts: 4580
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Re: Blog on UV and other aspects of microscopy and imaging

Post by Lou Jost »

A very interesting topic! May I make a suggestion? The x-axis of your graph doesn't have enough values marked, in particular no tick for 400nm, which is a critical number.

Lou Jost
Posts: 4580
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Re: Blog on UV and other aspects of microscopy and imaging

Post by Lou Jost »

I'm curious if you have found problems getting all the fluroescence colors right in an image. It's clear that adjusting the colors by white balancing can't get all the colors correct, because the fluorescence colors are not produced by reflected light. In my photos, adjusting the color balance to make the red fluorescence come out correctly, makes the blue fluorescence incorrect. I've been thinking maybe to use Adobe Profile Editor?

jmc
Posts: 96
Joined: Fri Apr 17, 2020 3:14 am

Re: Blog on UV and other aspects of microscopy and imaging

Post by jmc »

Lou,
Thanks for the note about the graph. You are right, I shall fix that in a minute.
I am lazy when it comes to white balancing my fluorescence images. I set the camera to daylight balance and just leave it at that, using the jpegs made by the camera.
Jonathan
Jonathan Crowther

rjlittlefield
Site Admin
Posts: 21028
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Re: Blog on UV and other aspects of microscopy and imaging

Post by rjlittlefield »

Lou Jost wrote:
Fri Aug 28, 2020 8:00 am
I'm curious if you have found problems getting all the fluroescence colors right in an image. It's clear that adjusting the colors by white balancing can't get all the colors correct, because the fluorescence colors are not produced by reflected light. In my photos, adjusting the color balance to make the red fluorescence come out correctly, makes the blue fluorescence incorrect. I've been thinking maybe to use Adobe Profile Editor?
I am curious, what exactly would it mean to get the fluorescence colors "correct""?

Clearly the captured & redisplayed image cannot reproduce the spectrum of the fluorescence. And as you note, there is no natural "white" reference because the fluorescence is emitted, not reflected.

The best I can figure out is that maybe you mean to reproduce the subjective appearance, given whatever your eyes happen to be using as an implicit reference when you do the viewing.

Is that what you mean, or something else?

--Rik

Lou Jost
Posts: 4580
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Re: Blog on UV and other aspects of microscopy and imaging

Post by Lou Jost »

Rik, I am struggling to capture what I see when doing fluorescence photography, and still struggling with how to think about it.
Clearly the captured & redisplayed image cannot reproduce the spectrum of the fluorescence.
Why do you say this? That would be my goal, to display the colors accurately, at least those within the gamut of the monitor.

When I look at the fluorescence, everything else is dark, so there is no incident light, no color temperature, and no correlation between the fluorescence colors, as there would be in ordinary reflected-light photography.

I would like my monitor to use an absolute colorimetrically correct rendering for the image, just like we can try for absolute colorimetric rendering profiles for printers.
...reproduce the subjective appearance, given whatever your eyes happen to be using as an implicit reference when you do the viewing
In the dark there is no implicit reference, just the fluorescence colors, which I suppose are providing some kind of reference. But as I said, my photos (as displayed on a monitor in a dark room) do not reproduce what I see in the fluorescent scene. Especially, I can't simultaneously get the right reds and the right blues.

I am thinking of photographing a prism-rainbow, displaying it on my screen, and calibrating my screen and camera so that a monitor-displayed photo of the monitor rainbow matches the monitor rainbow itself. When that condition is met, the monitor will be displaying the actual colors seen by my camera. I would know the true colors of the rainbow (their wavelengths) by their position in the rainbow (perhaps calibrating witha few known laser lines.

rjlittlefield
Site Admin
Posts: 21028
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Re: Blog on UV and other aspects of microscopy and imaging

Post by rjlittlefield »

Lou Jost wrote:
Fri Aug 28, 2020 12:50 pm
Rik, I am struggling to capture what I see when doing fluorescence photography, and still struggling with how to think about it.
Yeah, me too, at least the aspect of how to think about it. I've always considered the other part way too difficult to even attempt.
rjlittlefield wrote:Clearly the captured & redisplayed image cannot reproduce the spectrum of the fluorescence.
Why do you say this? That would be my goal, to display the colors accurately, at least those within the gamut of the monitor.
What I meant by "spectrum" is the intensity-versus-wavelength profile that would come out of a spectrometer. The spectrum of a point in your fluorescent samples will consists of a sum of more or less sharp spikes, one or more per type of material that is fluorescing. The spectrum of a pixel on your monitor will be a sum of rather broad peaks, one or more per RGB channel of the image. The fluorescent spectrum and the image output spectrum are never going to be the same. At best, they might produce the same visual appearance under certain viewing conditions. (Aside: there's a nice article at https://scienceblogs.com/principles/201 ... uter-color about how different monitor spectra can produce similar appearance.)

Anyway, your example of the real prism-rainbow versus its image on the monitor, and your words about "colorimetric rendering", make your intention clear.

So then, I think the problem that you're facing is a severe form of metamerism caused by the spikiness of the fluorescent spectra. The standard camera profiles will have been tuned to reproduce commonly photographed materials, which tend to have smoother spectra. What you're looking to do is map each spike, as seen by the camera, into whatever ratio of monitor RGB will produce the appearance of the spike as seen by a human. Or if fluorescent colors overlap, then you want to map a combination of spikes, as seen by the camera, into whatever ratio of monitor RGB will produce the appearance of the combination of spikes as seen by a human. It's not at all clear to me that this is a linear problem, so I can imagine that you'll need to map some combinations separately from their components. I think that's why the ColorChecker targets have lots of intermediate colors, not just gray and a few primaries and secondaries.

Reading a bit about the Adobe Profile Editor, it does seem like that's the sort of tool you need to work the problem.

Considering my doubts that this is a linear problem, I think I'd be more inclined to work directly with trying to create a profile for the fluorescent images, rather than trying to match the prism rainbow.
The rainbow will essentially be a lot of independent spikes, most of which you do not care about, and with no combinations, which you do. So, it feels to me like a different problem.

I will be very interested to hear your results, either positive or negative, or totally unexpected.

--Rik

Lou Jost
Posts: 4580
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Re: Blog on UV and other aspects of microscopy and imaging

Post by Lou Jost »

Rik, thanks for your thoughts. I think that there ought to be less metamerism in fluorescent subjects than in pigmented objects under reflected light. The few emission spectra that I have seen are fairly broad, given a single excitation wavelength, and are fairly pure colors. I have yet to see gray fluorescence, for example. In contrast, in regular photography the incident light has a wide range of colors, all interacting differently with each (generally impure) pigment. The fluorescence color situation has one less dimension than the regular one, because the incident light is monochrome.

But this is all hard to wrap my head around.

rjlittlefield
Site Admin
Posts: 21028
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Re: Blog on UV and other aspects of microscopy and imaging

Post by rjlittlefield »

Lou Jost wrote:
Fri Aug 28, 2020 8:44 pm
But this is all hard to wrap my head around.
Mine too. That apparently brief post of mine was roughly the third complete rewrite, separated by several hours of rolling the problem around in my head while I did various simple errands.

I'm guessing the big simplification in your problem is that the number of different fluorescing materials is small. I don't see that the spectrum of the illumination matters. Either way -- fluorescing or reflecting -- each different material is associated with some particular spectrum that goes to the camera, and you're trying to map that spectrum to a correct image color.

To be clear, I'm using the word "metamerism" to mean that the camera sees two spectra as having a different relationship to each other than the human eye does, due to differences in the spectral sensitivity curves of the camera and the eye.

That difference is a lot bigger than we often imagine when we casually speak of R, G, and B as being the primary colors for human vision. Compare for example the graphs of sensitivity for human eye L, M, S cell types, at https://en.wikipedia.org/wiki/Spectral_sensitivity, with the graphs of sensitivity for a Nikon D1X camera R, G, B photosites, at https://nae-lab.org/~rei/research/cs/zhao/files/nikon_d1x.jpg (linked from https://nae-lab.org/~rei/research/cs/zhao/database.html .). Note especially that the sensitive bands of wavelengths for L and M in the eye overlap far more than R and G in the camera. It's kind of miraculous that the cameras work as well as they do!

--Rik

Lou Jost
Posts: 4580
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Re: Blog on UV and other aspects of microscopy and imaging

Post by Lou Jost »

I think metamerism is a bigger problem with reflected light because the relationship between the objects' mix of colors depends on the light. A given magenta object can appear red or blue, according to the camera, depending on the light spectrum. With single-wavelength excitation, fluorescent objects always have the same color. It would be as if we only cared about a single standardized universal visible light source in reflected light photography. That simplifies the problem, removing a layer of variability.

rjlittlefield
Site Admin
Posts: 21028
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Re: Blog on UV and other aspects of microscopy and imaging

Post by rjlittlefield »

Lou, I'm not sure that what you heard is what I intended to say. Let me try again.

The aspect of metamerism that I'm talking about is not associated with changes in illumination.

It is the more fundamental problem that for some spectra, the camera will see colors in a different relationship to each other than our eyes do.

An an example, take a look at http://www.photomacrography.net/forum/viewtopic.php?t=37664 , in particular this section:
Interesting aspect #3 is that this subject is extraordinarily subject to color shift due to metamerism. The camera sees light that comes from the inside of the pupa as much more red than it appears to my human eye. To produce a rendition that is anything close to what I saw by eye, I had to desaturate the colors across the board by a Photoshop Hue/Saturation layer with Saturation of -43. Here's a comparison of the images as adjusted and as shot, with the ridiculous red cast to the pupa even though the background is neutral gray as it should be.
The image that appears there illustrates how striking the difference was.

I believe that you're having the same problem with fluorescent colors that I had with the reddish-brown of the pupa. Even though the camera was properly white-balanced for the spectrum of a gray card, and would have been properly saturated for most common colors, it saw the pupa as red while I saw it as brown.

In some cases, but not all, problems like this can be resolved by color profiles that incorporate 3D lookup tables. ("Turn this RGB into that RGB.") With the pupa, I could solve the much-too-red problem by mapping an intense color to a more muted one. But if the scene had also included a different material that both the camera and I saw as intense red, then there would have been no solution to the problem because making the pupa look right would have unrealistically muted the color of the other material.

I have no idea how your fluorescent images will behave in this respect.

--Rik

Post Reply Previous topicNext topic