Jump to content
UltravioletPhotography

UV and visible difference and bird vision


Jim Lloyd

Recommended Posts

I was attracted to UV photography through a somewhat romantic notion of gaining insight into how birds see the world given that they have 4 colour receptors including UV.

 

I wonder whether they might have a sense of the difference in their view between UV and visible. See below visible, UV and photoshop difference between the two. A bit garish maybe for some, but I think potentially interesting ...

 

post-175-0-09608800-1517614154.jpg

 

post-175-0-60692900-1517614171.jpg

 

post-175-0-56966500-1517614200.jpg

Link to comment

Well, Jim, I don't know......diffing of UV and Vis shots has been looked at since the git-go, but nothing much has come of it except some pretty colours. This perhaps sounds dismissive but I do not intend it that way because it is always possible that you can present it in a meaningful way which has not yet been seen.

 

One objection to the diff you have shown is the following. In the visible foto, the non-mossy part of the tree trunk on the right is reflective, relatively bright. Looks like maybe a pale green. In the UV foto the same part of the right-hand tree trunk is UV-reflective, almost white. So the bird with UV visual receptors and green visual receptors is going to see that non-mossy area on the right tree trunk as reflective, light-coloured, bright. But the difference foto does not show this at all. Instead, it shows us that the net result of diffing two bright areas is a dark area. (That should be no surprise for anyone who has played with Difference layers.)

 


 

Just to get started learning about how to best represent tetrachromatic Bird Vision, you might enjoy looking first at a slightly simpler case such as trichromatic (UV, Blue, Green) bee vision. We have quite a number of posts about that. Generally speaking they do not involve photoshop "diffs" but rather what I call "channel stacks". We have a tutorial about that here: Simulation of Bee-Colours I. (And there are two more tutorials in that series).

 

We have trichromatic filters which pass UV, blue and green light which we use for modeling Bee Vision. Here is a post about that which also mentions briefly Avian vision: The Bee, the Chicken, the Human and the Camera Look at a Sunflower.

 

Klaus Schmitt has experimented with some filters that he felt represented tetrachromatic Butterfly vision. Look on his blogsite for posts about that. http://photographyof...d.blogspot.com/

 

The problem in any model of insect or avian vision is representing the "imaginary" colours. Here is one of my more fanciful attempts at that. I used a kind of mottled patterning to represent the imaginary colours. http://www.ultraviol...-here-first-jk/

Link to comment

Hi Andrea,

 

Many thanks for your thoughts and those links. I have just had a quick skim through at the moment and they certainly look interesting and worth careful study.

 

I think your idea of using a separate parameter (such as texture) as an extra dimension was inspired! - seems almost obvious now, but only after I had been told!

 

What I was thinking was that it will always be impossible to know what seeing in 4-colour is like, so I wondered if presenting the difference between what we see and what the bird sees might be informative. I know this is somewhat fanciful since each species is different and there are also differences in the R G B cones (I know that is an inaccurate description). Anyway in the example above then the tree trunk appears dark because that is saying that the bird doesn't have any extra information about that than we do - whereas the moss is very bright to indicate that the bird knows more about it than we do.

 

I am trying to say something about colour correlation in the scene. In most scenes we can understand most of what we are looking at by looking at any one of the R G or B channels as they are reasonably correlated - something that appears white the RGB channels are completely correlated so that we could know all we needed to know about it from any one of the channels - if this object is also white in UV, then then having the "UV vision" doesn't add anything .

 

In medical image analysis (and many other fields) when we want to reduce the dimensionality of image data we can use principle component (PC) analysis. The first PC is the weighted sum of the variables that has the greatest variance. So in this case one would end up with PC1 = aR + bG +cB +dUV . PC2 and PC3 could also be calculated and mapped back into RGB.

 

I guess someone might have done this?

 

But my gut feeling is that what you end up with might not be that interesting in that it will probably look fairly similar to the visible image anyway.

 

So I am thinking that I need a technique that highlights uncorrelated aspects of the UV image compared to the visible. Not sure how to do that, but a simple subtraction might be a start?

 

I will keep exploring, but interested to know if this makes sense to anyone ! ;) :huh:

Link to comment

Anyway in the example above then the tree trunk appears dark because that is saying that the bird doesn't have any extra information about that than we do - whereas the moss is very bright to indicate that the bird knows more about it than we do.

 

Not quite, Jim. You have it kind of reversed. The bird does have extra information about the bright area on the trunk. The bird sees that bright area on the trunk in a colour we can only imagine. In the literature insect or bird colours involving UV are often designated with a UV prefix. So the bird sees what we humans might call a "pale green" area as a "UVgreen" area because both the UV and the green receptors in the bird are being stimulated by reflected UV and reflected green wavelengths.

 

In contrast, both the bird and the human see that green moss as just plain green because in both bird and human only reflected green wavelengths are stimulating the green receptors. The UV is absorbed by the moss, so the bird receives no reflected UV wavelengths from the moss. The bird has no more information about the green moss than we humans do.

 


Human colour vision: red, green, blue, yellow, cyan, and magenta ( a "brain" colour).

Looking at human colours first for a better later analogy, here's a shorthand way to designate those colours. But mind you, this is very, very oversimplified. Human vision is vastly more complex than just these simple light combinations. Still, it gives us a handle for dealing with animal vision.

A plus sign designates reflection.

A minus sign designates absorption.

  • +R-G-B = red, one receptor stimulated (therein lies one huge oversimplification, but carry on...)
  • -R+G-B = green, 1
  • -R-G+B = blue, 1
  • +R+G-B = yellow, two receptors stimulated
  • +R-G+B = magenta, 2
  • -R+G+B = cyan, 2
  • We can also add white and black, in terms of receptor stimulation.
    • +R+G+B = white, ALL receptors stimulated
    • -R-G-B = black, NO receptors stimulated


Bee colour vision: blue, green, cyan, UV, UVblue, UVgreen

Again vastly, hugely oversimplified. Insect vision is every bit as complex as human vision.

Plus and minus signs as before.

Please note that use of the word "imaginary" here is relative to human vision. Kindly excuse this anthropomorphism.

  • +UV-B-G = the colour UV, which for humans is imaginary, 1 receptor stimulated
  • -UV+B-G = blue, 1
  • -UV-B+G = green, 1
  • +UV+B-G = the colour UVblue, again imaginary, 2 receptors stimulated
  • +UV-B+G = the colour UVgreen, imaginary, 2
  • -UV+B+G = cyan, 2
  • And as before we will add black and white.
    • +UV+B+G = white, ALL receptors stimulated
    • -UV-B-G = black, NO receptors stimulated

Here is where I am duty bound to note that while the bee does not detect red wavelengths, that does not mean that the bee cannot detect a red flower. Many references for bee vision can be found by using the search tag "Insect Vision" or by browsing in our Reference section.

 


Avian colour vision.

Now, as is commonly written in mathematics books, "I leave as an exercise for the interested reader" to write out the designations for Avian colour vision primaries and secondaries. How many imaginary colours would there be? How would you best emulate Avian colour vision.

 

(I think not all birds see UV? Only some? Anyone want to look that up and report back? Thanks.)

 

I HAVE TO NOTE HERE that there are some scientists who dispute the UV receptor in Bees or Birds or other animals. They think that there is simply an "extension" of some or all of the R/G/B receptors into the UV region.

 

Interestingly, I was just reading that there are some scientists who dispute the photonic interaction of light with the eye and postulate a wavelength interaction instead. Look for a physics guy named Dr. Huth. And look into what Edwin Land thought.

 

 

 

Corrections always welcomed when I get going like this......

Read up on this stuff yourselves! Don't take my word for it!! B)

Link to comment

In medical image analysis (and many other fields) when we want to reduce the dimensionality of image data we can use principle component (PC) analysis. The first PC is the weighted sum of the variables that has the greatest variance. So in this case one would end up with PC1 = aR + bG +cB +dUV . PC2 and PC3 could also be calculated and mapped back into RGB.

 

I guess someone might have done this?

 

Yes, me. :)

I've tried both PCA and ICA.

See here for example,

http://www.ultraviol...dpost__p__14760

 

a technique that highlights uncorrelated aspects of the UV image compared to the visible

That would be ICA.

Link to comment

I think technically, to create some kind of bird vision, one would probably need to reduce the UV to monochrome, and combine it with visual.

There are filter stacks that will combine UV with visual, but the UV is colored, so that changes the result.

The typical UG5 UV+Blue+Green, AKA Bee Vision, uses UG5 (or U-330) 1.5mm + S8612 2mm.

If you want Bird Vision, then I recommend UG5 (or U-330) 1mm + S8612 1.5mm, which will give you a UV+Blue+Green+Red.

That stack would probably be the best for combining all of that range in one filter/shot.

 

Here is an example of Visual, Bird Vision, and Bee Vision.

This example used UG5 1mm + S8612 2mm (Bird), and UG5 2mm + S8612 2mm (Bee), these are not the exact thickness formulas I would use, but show the idea.

post-87-0-15646000-1517687865.jpg

Link to comment

Here is a simple example. But I'm going to stick to UV/B/G trichromatic. Perhaps some other time I'll tackle an example for UV/R/B/G.

 

Visible flower.

potentilla_vis.jpg

 

UV flower.

potentilla_uv.jpg

 

Difference of Visible & UV. Quite dramatic, but not sure it's telling us anything about Vis & Uv?

potentilla_uvVisDiff.jpg

 

Flower under UV+Blue+Green filter (UG5-1.5+S8612-1.5).

Remember, there is a difference (not the PS type of Diff) between how the Camera + Filter Stack sees the yellow flower and how the Bee sees it. If (and that is a BIG IF) we assume that the Bee's green receptor "extends" enough to be tickled by reflected yellow wavelengths, then the Bee is going to see the UV-bright portions of this flower as an imaginary reflected UVGreen and is going to see the UV-dark center of this flower as reflected green.

 

Now, would the reflected green in the flower center be dark green? Suppose the yellow wavelength produces a very low stimulation of the extended green receptor. Would we represent that as a less saturated green or as a dark green? Suppose the yellow wavelength produces an all-or-nothing stimulation of the extended green receptor. Would that mean we have a bright, saturated reflected green? It could be that the yellow wavelength tickling the extended green receptor is interpreted as some kind of actual yellow. And finally, maybe the bee cannot see yellow at all, so the center of this flower is just gray to the bee.

Does everyone begin to see the problems in modeling alternate vision??

potentilla_ug5-15_s8612-15.jpg

 

Modification of preceding to green colours.

But that dark center is still a problem.

potentilla_beeVee1.jpg

 

Andrea's three fanciful takes on Bee vision's green center and imaginary UVgreen petals.

The first version places an "imaginary" UVgreen on the outer petals.

potentilla_beeVee2.jpg

 

Here again is the imaginary UVgreen but this time with a reflected green center which is not dark.

potentilla_beeVee3.jpg

 

The last interpretation assumes the bee can't see yellow at all so the outer petals have an imaginary UVgreen colour and the center is bee-gray. :D

BTW, it is said in the literature that the bee does not see green foliage backgrounds as green but rather as a neutral tone.

potentilla_beeVee4.jpg

 

 

Black & white UV as Luminosity layer over Visible frame.

The same problems as before. Neither Bee nor Bird sees the flower like this.

potentilla_uvLumOverVis.jpg

Link to comment

In the next life I will study Quantum Biology.

 

[Rant]

Andrea does wish that those who are providing Bee Vision photographs to museums and exhibits would learn more about Insect and Avian Vision in order to explain WHY they think their models are useful. The bees and the bee scientists are all rolling their eyes at some of the stuff "out there". I don't know all that much about it but Im rolling my eyes at some stuff "out there" also. Granted, models are only models. But shouldn't they be somewhat representative of the reality? We know here on UVP that we are only experimenting and playing around and that what we produce by way of Alternate vision is flawed. And we know some (not all) of the reasons why.

[/Rant]

 

This is certainly NOT directed at any member here.

Link to comment

The new Star Trek has a strong Quantum Biology theme to it, used for space jumping.

 

Your rant is true. Alex has perhaps the best approach to Bee and Bird, which uses two shots, or in his case, two simultaneous shots.

I have seen some UV+Blue+Green images in museum/TV videos, that are UG5'ish, so I have to agree.

The main difference to me is that the UV-A has colors within its band, so instead of UV -> Blue, Blue -> Green, Green -> Red (or whatever other arrangement, percentages, and shifting),

you end up with RGB + RGB, the Blue and Yellow and Gray of the UV mixing with the Blue, Green, and Red of the visual.

Alex explained to me once a possible method of subtracting the different in order to use the single stacked filter method to simulate the two filter composite method.

It might be something ask him about, and it might be a very good method to use.

I still like the stacked filter method, regardless, easy and fun and really ads something to the mix.

Now try one for the Mantis Shrimp... :D

Link to comment
Be sure, I do LOVE my UG5 and U-300 filter stacks. I use them every summer. But I cannot see myself selling one of those photos to a museum and saying it is "bee vision". The UV+B+G result must be modified to better match the reality (insofar as we can possibly know it, which is not too far). Too many of these photos show the UV-absorbing areas as too dark. For example, UV-absorbing blue-reflecting areas are not dark to the animal with UV and blue receptors. To these animals reflected UV is a color not a tone. And, after all, the camera is trying to model human vision. We cannot expect it to produce a good version of bee vision just because we turn away the red/IR wavelengths.
Link to comment

While we are at it, take a look at this idea, what do you think.

---

To subtract UV from UG5 stack shot you would need to subtract each channel individually.

 

Split UG5 shot in R, G and B channels.

Split UV shot in R, G and B channels.

Subtract each channel individually.

Recombine resulting R, G and B into one image.

 

It may also be important to take the UV shot by stacking a Baader U on top of UG5 stack.

---

Link to comment

What means "subtract" here. A photoshop difference layer?

 

Diff of UV and UG5-Stack isn't going to provide anything useful?

 

Diffing is arithmetic modulo 256 so the Diff of two photos is the same as the Diffs of each channel recombined.

Link to comment

An idea Alex was discussing with me once.

Idea being to separate UV as one monochrome color channel, from the visual Blue and Green.

For more control over the mix.

Link to comment
doesn't solve the problem of UV-absorbing areas looking dark when the bee/bird does not see it that way
Link to comment

The idea was to separate the UV from the blue and green, so that blue and green would be near pure visual blue and green, and UV would be near pure UV.

From there those 3 channels can be used as desired.

 

Maybe Alex can explain it better than me.

Link to comment

Andrea, I don't know. Just trying to come up with a way of separating the UV from the visual.

Maybe message Alex. I could be way off, it has been years since I discussed it with him.

I tried the idea with several shots back then, results were not exactly my favorite, so I didn't do much more with the idea.

I can't find those shots at the moment.

 

Here is a Sparticle of a UG5 stack. Just food for thought, not exactly;y related to the above.

post-87-0-45236100-1517716511.jpg

 

Here is a UG5 shot of Rudbeckia.

post-87-0-63368800-1517716532.jpg

Link to comment

I really appreciate having such ready access to so much accumulated wisdom on this subject. I am in the process of trying to get funding to study this topic from an art/science perspective (I don't really like the term art/science or any of the alternatives like SciArt - but its about whether art practice can make a unique contribution that advances the subject beyond what is available to science alone). So this is all really helpful. Maybe most encouraging of all is that it appears to be an unresolved issue and therefore there is scope for research (and hopefully funding!)

 

Anyway, possibly risking making a fool of myself (and risking being accused to trying to teach my grandmother to suck eggs!) I was thinking again above the arguments for/against the difference approach ...

 

In relation to my original mossy tree trunks images, Andrea says

 

"In contrast, both the bird and the human see that green moss as just plain green because in both bird and human only reflected green wavelengths are stimulating the green receptors. The UV is absorbed by the moss, so the bird receives no reflected UV wavelengths from the moss. The bird has no more information about the green moss than we humans do"

 

Surely the bird does have more information - precisely that there is no UV reflection.

 

I was trying a thought experiment ...

 

Imagine that I only have red and green receptors and my friend only has one receptor that spans the same red-green range.

 

If we lived in a yellow world (i.e. where all objects reflected red and green in the same ratio - and assuming it is illuminated by white light) - we would get along fine and have no disagreements about the world. But if into this world is now introduced a red object (i.e. one that reflects red light and absorbs green). Then it is the lack of green reflection as much as the positive red reflection that gives me a different sensation now to my friend.

 

So if I wanted to try to show something to my friend of my inner world I could show a map of differences between my red and green - in this very simple case (a yellow world with a red ball) - the map would be all black except for the ball which I could show in yellow. It would say - look, most of the scene is the same for you and me, but for this bit I have a different sensation. And this difference is due to my awareness of a lack of something - about which my friend has no conception.

 

Obviously I am using this as a way to think about the bird and UV. - The bird might have a sense of the moss which relates to the fact that it reflects visible but not UV.

 

Another sort of similar idea I had was about why we get sunburn. Obviously through overexposure to UV ( B), but why do we let that happen? Basically I would say this is because we can't sense UV, but we make assumptions about it strength based on brightness and warmth - and this guides our behaviour. So we often get burnt (well I do!) in the spring time, when the sun might not be that bright and it's not too hot, but UV levels are high. We extrapolate from one part of the EM spectrum to another - which most of the time is fine as they are correlated, but we get into trouble where they are not. Could we think of the visual equivalent in a similar way?

 

It may well be this will be shown to be nonsense (and hence I will look even more silly for persisting ...), but one learns from ones mistakes ...

Link to comment

As an illustration of above here is the Jasmine with UV , Visible and then difference between monochrome versions of each.

 

post-175-0-81070700-1517738889.jpg

 

post-175-0-81070700-1517738889.jpg

 

post-175-0-07976400-1517738934.jpg

 

Sorry I don't understand why these images appear in this order, but I think you can tell which is which

post-175-0-35718800-1517738906.jpg

Link to comment

Jim, about image placement: After you upload your files, look to the right of each file listing. You will see "Add to Post|Delete". Place your cursor in the post where you want the image to appear, then scroll down and click "Add to Post". If you do not Add images to Post by clicking that, then images will appear under the "Attached" header - perhaps not in the order you would like.

 


 

OK, now let me think over what you have written. :)

Link to comment

Thanks Andrea

 

I am also thinking that UV sensitivity might only be a small part of the overall difference between bird vision and human vision - I need to think also of visual acuity, field of view, motion of the bird etc.

 

Most birds have much wider field of view than ours - this is my first quick attempt at a UV panorama that might be a way to show this aspect.

 

D3200 (modified), UG1 2mm + BG40 2mm, Soligor 28 mm lens f/8, asa 400, exp 2 sec, WB in Lightroom with calibration file, stitching in Lightroom from 6 vertically oriented images. Just a quick try out, there is some variation in WB across the scene that I need to work on - but I think looks like something useful to explore ...

 

post-175-0-11970200-1517762769.jpg

Link to comment

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
×
×
  • Create New...