UVR Batesian Mimicry of Social Wasp (Vespula sp.) by Hoverfly (Episyrphus balteatus)

[Andrea B.]

Andrew, I'm being curious. You reference this paper: The key mimetic features of hoverflies through avian eyes. 

Are you remembering that these UVR photos only show one channel, so-to-speak, of avian vision?

The wasp and the hoverfly may have some similarities in these one channel UVR photographs.

But how do they look to the bird?

(I don't know, of course!!)

 

So this got me to thinking how we would emulate bird vision in our photos for those birds having tetrachromatic vision. Anything UV-absorbing would be the same color to the bird as it would to us. Anything UV-reflecting would need some special method of showing UV-red, UV-green, and UV-blue. Interesting.

[JCDowdy]

4 hours ago, Andrea B. said:

So this got me to thinking how we would emulate bird vision in our photos for those birds having tetrachromatic vision?

 

Interesting approach, might be possible?

https://www.quora.com/Can-we-simulate-tetrachromacy/answer/Kilian-Lachner

[Andrea B.]

That was really interesting! I have some different color cellophane filters. Just put a different color on each eye seems to be the suggestion??

That tetrachromancy does not involve reflected UV though.

 

 

 

[Andrea B.]

Here's my take on Avian vision for those birds having UV, Red, Green, Blue sensitivity. As I mentioned above, any UV absorbing area would appear to the bird the same as it would appear to us. Any UV reflective area would involve a combined reflection of both the UV and a visible color. So in PS, I used a purple exclusion scatter brush to paint over all the UV reflecting areas to give such regions an unusual look.

 

This is just for fun. And I've used this bouquet photo before to try to come up with an emulation of Bee Vision here.

 

Visible Bouquet

 

 

Reflected Ultraviolet Bouquet (raw composite, no white balance)

 

 

 

Avian Vision ??? 

Note UV-reflective tips of sunflowers and the lack of the usual dark central bulls-eye. That is only seen in "one channel" UVR photos.

This was made quickly, for fun. So, for example, shadows are not properly dealt with. It's tricky to separate the dark shadows from UV-absorbing areas. Like, those lilies are UV reflective everywhere, so the bottom lily needs the inner portions also scattered with purple exclusion.

 

 

[Yves W]

Thank you Andrea, OK for the avian vision, the warning colors R+N, J+N... play their roles

Your work from 2016, given by the link, is REMARKABLE.

Ulf's photos are magnificent, I try but it's not as good

 

The insect vision of flowers and cryptic spiders fascinates me.

[Andrew Dayer]

 

Wow - thanks for taking the time, Andrea!

Andrea said: "Just an observation from my own experience. It is sooooo difficult to make a UV photo of a flying insect!!"

 

Actually, I wasn't trying for the hoverflies! I was just looking for ideas for use in a project (actually part of my masters thesis) and they just got in the way. Serendipitously, I suppose, the image (hoverfly v's wasp) were noticed and  - starting this week - I've started a doctoral program around the difference in patterns in in UV v's vis. Shame its not funded!

Andrea said: "I am always cautious about making any judgements based on false color though. Be very careful there."

 

I feel its pretty worthless without calibration - so many variables; I'll start trying to replicate the normalisation processes used in recent papers. Buying a range of Specralon samples is not going to be cheap but if I can find a 'master set' and calibrate some barium sulphate / Black x.0 that may be OK. Another topic for another day

Andrea said: "The raw composite ... shows the typical raw color of a BaaderU photo."

 

Is 'raw composite' synonymous with demosaicing in this context?

Andrea said: Typically there should not be an overall yellow cast....
 

I've seen this yellow cast whenever I've tried to WB images largely light by flash with ZWB1 on it. I wondered, thinking of lens sparticle tests, if I had cut out most of the longer wavelength blue tones and hence ~350nm. I've never noticed much green maybe due to the Baader U and camera sensor.

Andrea said "There is some kind of flare or light leak. There are usually not violet/blue-ish shadows..."

Many of my UVR images have something of the sort; just assumed it is 'normal' consequence of white balancing an image made with a sensor getting signal beyond its design parameters and then using 'a one WB fits all parts of the image'. Seems to be mostly underexposed areas (so includes shadows), more prevalent in images illumated with flash and a particular feature of using the ZWB1's on the flash (yellow highlights / blue shadows). Doesn't seem to be flare; nothing looks troubling to my eyes if the raw file is converted to monochrome (although that is another can of worms).

How can I test?

The set up was camera (viewfinder covered) / lens / Baader M52 - M54 - 'M48' adapters / Baader U. Flash was on a stand & behind the lens and to the left; sunlight from the behind and to the right and at about 55o elevation.

Out of time I'm going to post this now and come back on the points later.

 

[ulf]

3 hours ago, Andrew Dayer said:

 

Wow - thanks for taking the time, Andrea!

Andrea said: "Just an observation from my own experience. It is sooooo difficult to make a UV photo of a flying insect!!"

 

Actually, I wasn't trying for the hoverflies! I was just looking for ideas for use in a project (actually part of my masters thesis) and they just got in the way. Serendipitously, I suppose, the image (hoverfly v's wasp) were noticed and  - starting this week - I've started a doctoral program around the difference in patterns in in UV v's vis. Shame its not funded!

Andrea said: "I am always cautious about making any judgements based on false color though. Be very careful there."

 

I feel its pretty worthless without calibration - so many variables; I'll start trying to replicate the normalisation processes used in recent papers. Buying a range of Specralon samples is not going to be cheap but if I can find a 'master set' and calibrate some barium sulphate / Black x.0 that may be OK. Another topic for another day

Andrea said: "The raw composite ... shows the typical raw color of a BaaderU photo."

 

Is 'raw composite' synonymous with demosaicing in this context?

Andrea said: Typically there should not be an overall yellow cast....
 

I've seen this yellow cast whenever I've tried to WB images largely light by flash with ZWB1 on it. I wondered, thinking of lens sparticle tests, if I had cut out most of the longer wavelength blue tones and hence ~350nm. I've never noticed much green maybe due to the Baader U and camera sensor.

Andrea said "There is some kind of flare or light leak. There are usually not violet/blue-ish shadows..."

Many of my UVR images have something of the sort; just assumed it is 'normal' consequence of white balancing an image made with a sensor getting signal beyond its design parameters and then using 'a one WB fits all parts of the image'. Seems to be mostly underexposed areas (so includes shadows), more prevalent in images illumated with flash and a particular feature of using the ZWB1's on the flash (yellow highlights / blue shadows). Doesn't seem to be flare; nothing looks troubling to my eyes if the raw file is converted to monochrome (although that is another can of worms).

How can I test?

The set up was camera (viewfinder covered) / lens / Baader M52 - M54 - 'M48' adapters / Baader U. Flash was on a stand & behind the lens and to the left; sunlight from the behind and to the right and at about 55o elevation.

Out of time I'm going to post this now and come back on the points later.

 

Spectralon is just a very pure expensive form of PTFE that is sintered to make it matte.

 

You will not need a series of those if you are just aiming for WB of UV-photos.

A good alternative is to get a sheet of virgin PTFE and make it matted by sanding it with 200-300 grit water sand paper under flowing water.

The sanding should be done in all different directions to get a randomly matted surface.

 

As for a series of references with different reflectivity that is nice for measurement purposes, but for photo you can just take a calibration image with a suitable exposure to avoid overexposure of the PTFE and then transfer the WB-setting to the actual image.

 

A second best alternative WB target is a nature-anodised (non tinted) aluminium piece with a matted appearance. Aluminium and aluminium oxide have a rather good flat enough reflectivity into the UV-A range.

---

About (yellow highlights / blue shadows), I think this is due to the mixed illumination of filtered flash and steady background illumination that better reach into the shadowed areas where the flash do not fully reach. Such an image is impossible to fully WB as there are two different light sources, with different spectral contetnt.

 

Filtering the flash is only meaningful if you want to take UVIVF-images.

That is normally done in a rather dark environment

 

Just to think about it:

It can be compared with filtering the sunlight with a big ZWB1!

That is naturally not possible, but I guess you are using the flash to avoid long exposure times and blur.

[Andrew Dayer]

3 minutes ago, ulf said:

Spectralon is just a very pure expensive form of PTFE that is sintered to make it matte.

 

You will not need a series of those if you are just aiming for WB of UV-photos.

A good alternative is to get a sheet of virgin PTFE and make it matted by sanding it with 200-300 grit water sand paper under flowing water.

The sanding should be done in all different directions to get a randomly matted surface.

 

As for a series of references with different reflectivity that is nice for measurement purposes, but for photo you can just take a calibration image with a suitable exposure to avoid overexposure of the PTFE and then transfer the WB-setting to the actual image.

 

A second best alternative WB target is a nature-anodised (non tinted) aluminium piece with a matted appearance. Aluminium and aluminium oxide have a rather good flat enough reflectivity into the UV-A range.

---

About (yellow highlights / blue shadows), I think this is due to the mixed illumination of filtered flash and steady background illumination that better reach into the shadowed areas where the flash do not fully reach. Such an image is impossible to fully WB as there are two different light sources, with different spectral contetnt.

 

Filtering the flash is only meaningful if you want to take UVIVF-images.

That is normally done in a rather dark environment

 

Just to think about it:

It can be compared with filtering the sunlight with a big ZWB1!

That is naturally not possible, but I guess you are using the flash to avoid long exposure times and blur.

 

Thanks Ulf

I think I will need to have good quality calibration since I will be normalising (against what - sunlight / organism cone catpture??) and with vis images to obtain data. Likely will use some combiantion of quantiative toolboxes as MICA and the bespoke package in last weeks paper:

 

WB - yellow / blue. I think - and hope - you are correct. I've looked for examples with as different setup as possible. This image was for an article on 'low cost UV' last year, with my trusty Nikon D50 and a Galaxy 135mm f3.5 in natural sunlight. 3 second exposure hence some blur of the petals. Still had to use the Baader U; this was the day I found that my Tangsino ZWB3 was opaque. Background has a blue tone to me.

 

 

You're correct about the filter on the flash - I had been trying UVIVF and just grabbed the wrong unit for the hoverfly pics. Later images (at start of the thread) are with an unfiltered FS Vivitar 285HV.

 

[ulf]

Yes a good quality calibration target is often needed, but the Spectralon is way more advanced and expensive than what is needed!

IMHO the need for Spectralon is valid only if you do advanced laboratory measurements and need a traceability the same way as advanced measurement instruments are sent for calibration on a regular basis. 

 

The background of your nice dandelion above might be a tad bluish too, to my eyes just now, but that can be due to that I am in a warmer light environment for the moment.

When getting close to neutral grey or white the slightest deviation is visible, but also easy to bias due to the current light environment.

 

You let the illumination Sun / flash... hit the target and take a picture with the filters to be used on the camera.

When I take UV-related pictures I often go through many (8-15?) different filters so there are many images with some PTFE beside those almost identical where I removed the target.

Lately such pictures was taken indoors with flash illumination only.

That indoor macro studio was packed away due to renovation and I have not yet set it up again.

Lately I have been more into IR.

[Andrea B.]

Wow - thanks for taking the time, Andrea!

 

I LOVE trying to make bee or bird vision images. Deciding how to represent reflected UV mixed with reflected R/G/B wavelengths is interesting & fun.  I think a lot of people think they are showing insect or avian vision by simply making a UVR photo. They forget that is just one "channel" of the animal's vision.
 

Is 'raw composite' synonymous with demosaicing in this context?

 

High level explanation:  Raw Digger's raw composite photo has been first demosaiced. Then some "gamma" curves are applied to translate the recording into the lights/darks of human vision. But no white balance has been applied. So the false colors you are seeing in a raw composite is what the camera actually recorded in its R,G,B channels. Note that I choose to present the two green channels as one averaged green channel to avoid the inverted appearance that occurs with a 4-channel image.

 

Here are some suggestions about white balancing false color files. In this write-up there are links to examples of the various materials which can be used. PTFE is the cheapest. LINK

 

Here is a WB example using PTFE. That is the white disk on the right. On the left is a visible white balance set. In the middle is the WB area on the Color Checker Passport. 

Gear:  Nikon D200 conversion + Noflexar 35/3.5 + BaaderU UV-Pass Filter + Sunlight.

 

First the raw composite to show what false colors were recorded. This raw comp was given a bit of contrast and some detail enhancement because the output from Raw Digger is rather flat - which is as it should be, but doesn't make for fun viewing as a photo.

 

 

Next the photo which was white balanced on the PTFE in Photo Ninja. A bit of light/dark adjustment and some detail enhancement were also made. This photo has the "standardized" look which most Bayer-filtered cameras show after WB. Some newer cameras have slightly different appearance after WB against PTFE or Spectralon. My Panasonic S1R, for example.

Please note that visible white balance cards are not useable for UV white balance because they do not retain the same reflectivity under UV light that they have in ordinary sunlight.

 

[Andrea B.]

Remind me again what camera you are using?

 

Test for light leaks through viewfinder, ports, LCD windows, etc.

• Get set up to take a typical UV photo.
• Make the photo.
• Now cover everything but the end of the lens with a nice thick, dark towel.
• Make the photo.
• Are there any obvious differences? If so, then something is light-leaking.
• Sometimes light leaks can be subtle. Put the two photos into a converter and run an Auto-Levels on each. Any obvious differences?

[[Do we want Auto-Levels there? Or to push contrast?? I don't have an example at hand to play with.]]

 

Test for internal IR shutter monitor contamination.

• Close the viewfinder.
• Put the lens cap on.
• Set the camera to a 30" exposure time.
• Go into a dark room.
• Put the camera/lens under a nice thick, dark towel.
• Push the shutter.
• Look at the 30" photo. It should be black.