UV color experiment/comparison

[Stefano]

UV color is a complex and controversial topic. For example, Andrea will always remind you that UV false color is not strictly related to wavelength, as it depends on many factors (lighting spectrum, lens transmission, filter transmission, sensor response, white balance...), although we always see the same colors in our UV photos: blues, lavenders/purples, yellows, and sometimes green. Red is not a color that we would expect.

 

A different way of thinking at color outside the visible spectrum in general is to make a TriColour/trichrome/tri-band image, which often produces more natural-looking colors (for example, the sky is still blue) and also there is a wavelength-color relationship. UVP member Bernard Foot has experimented with the technique some years ago, and I have already tried it before.

 

Other people (notably UVP member OlDoinyo) like to render white-balanced UV photos in BGR (swapping the red and blue channels), which also produces blue skies and a different color palette.

 

Since I have a color camera, the images I take when making a TriColour image have colors, which I normally get rid of to make the channels. If I stack those images instead, I can simulate the raw image taken by a camera with an approximately flat response between about 310 nm and 400 nm, and with sunlight having a uniform spectrum too. This never happens in real life, even with a UV-dedicated lens. The interesting part is comparing the resulting colors with those of a normal UV photo.

 

The equipment I used is the usual one: full-spectrum Canon EOS M, SvBony 0.5x focal reducer lens and the following filters:

 

TriColour:

Blue channel: double 310 nm Chinese bandpass filter + ZWB1 (2*2 mm) (the ZWB1's are not necessary, but I used them anyway);

Green channel: BrightLine 340/26 filter + ZWB1 (2*2 mm);

Red channel: BrightLine 387/11 filter + Chinese BG39 (2 mm);

 

Standard UV:

ZWB2(2 mm) + Chinese BG39 (2 mm);

 

Visible reference:

Chinese BG39 (2 mm);

 

The technique used to make the TriColour images is also the usual one, described here. The major difference is that I took multiple 310 nm exposures this time and stacked them taking the darkest pixels (5 in both cases). As for the raw color stacks, I set the brightness of each image to be about the same by eye and stacked them. Also, following Andy's advice last time, I raised the brightness of my images and the contrast in the TriColour stacks (also because the contrast in the original channels was removed in PhotoNinja during the processing). The visible and UV references are white-balanced in-camera, the raw stacks were white-balanced in PhotoNinja.

 

I used both UV-lavender and UV-yellow subjects.

 

For the lavender, I picked three items with varying degrees of lavender: a magnifying glass on the left (transparent at 387 nm, mostly transparent at 340 nm and opaque at 310 nm), almost colorless; a white LED lightbulb in the middle, and a plastic lens on the right (mostly transparent at 387 nm but opaque at 340 nm and below, which shows a strong blue-purple color).

 

Visible reference:

 

Standard UV:

 

White-balanced raw UV stack:

 

As you can see, the color palette didn't change much, but since here the shorter UV wavelengths contribute much more to the image, the magnifying glass is noticeably darker. In general, objects with a pale lavender color got a color boost.

 

UV TriColour:

 

Here the color palette is obviously richer, with the color giving a good indication of the transmitted/reflected wavelengths.

 

Standard UV, BGR:

 

Compared to the TriColour rendition above, only the plastic lens on the right looks similar, while the color deviates more for items with a flatter UV response.

 

For comparison, here's the raw stack, in BGR version:

 

...and now for the yellows. Here I used a 3 mm thick ZWB1 filter on the left, and a 2 mm thick ZWB2 filter on the right.

 

Visible reference:

 

Standard UV:

 

Here the colors look similar, with the ZWB1 filter being slightly greener, as expected. Also the paper tissue I used apparently contains UV-absorbing fibers.

 

White-balanced raw UV stack:

 

Here things get weird. The ZWB1 filter got orange, which is a bit different from its normal color. Also, and this was expected, the difference in color (hence transmission) between the filters is more evident now.

 

UV TriColour:

 

Standard UV, BGR:

 

UV stack, BGR:

 

Raw or .tif files are available.

[Andy Perrin]

Very nice! I think this procedure is working well for you now. You demonstrably bring out a larger color gamut than the usual blue-green-yellow here. Now I think you should apply it to more kinds of objects. 

[Stefano]

Thanks Andy. This technique requires time both in the image-taking phase (I need to take out all my filters, change them without moving the camera, and find the focus at 310 nm by trial-and-error), and also it requires time to align the images, so I can't do it often (in total, I made just 14 TriColour images), but the end result is indeed very nice. Flowers in a windless day could be an idea.

[Andrea B.]

Nice work, Stefano. 

 

Try a TriColour photo of an egg.

[Stefano]

I will have to wait for a sunny Saturday/Sunday again, otherwise the 310 nm channel is too noisy.

 

Intact or open?

[Andy Perrin]

20 hours ago, Stefano said:

Intact or open?

One of each in same photo if possible!

[dabateman]

See if you can image a white egg. I looked at a brown egg, and they should be different. 

We just only seem to get brown eggs.

[Andrea B.]

I was thinking of an egg in its shell, not cracked open. 

 

I'm afraid if I took an egg outdoors this afternoon to photograph it, it would cook in its shell in this hot sun. 

 

I once made a really nice UV photo of some eggs in a woven basket which made a nice shadow pattern if light was at a certain angle. IIRC, the eggs were moderately UV-dark.

 

Egg shell fluoresces dark purple:  https://www.ultravioletphotography.com/content/index.php?/topic/4672-egg/

 

[Stefano]

I have brown eggs only, and seeing David's brown eggs series I think they should appear brownish in TriColour, as bright yellows and oranges only appear if there is a significant difference in reflectivity between the bands.

 

I think I should be able to try a hard-boiled egg like David did.