Simulation of Bee-Colours II

[Nico]

Chalwatzis, N. (2013): Simulation of Bee-Colours II

 

reference: http://www.ultravioletphotography.com/content/index.php/topic/648-simulation-of-bee-colours-ii/

 

In the first article of this series Simulation of Bee-Colours I the underlying idea of visualising “bee-colours” by mapping false-colours to UV, blue and green. This part is mainly going to illustrate the method by using concrete examples.

 

First we will look at the image of two Chrysanthemum segetum flowers (cultivars) (fig. 1).

Both are yellow in the centre while the right one has petals that are white on the outer half. The individual RGB channels shown below illustrate nicely that the white parts of the petals have an even reflection across all three colour channels, while the yellow colour appears when blue is absorbed (not reflected). The visible light image is composed of the three individual channels that are shown as greyscale conversions on the bottom. The UV image (upper left) is not part of this RGB image but shows that the flowers show almost no UV reflection compared to the background.

 

Figure 1: Visible light image of two Chrysanthemum segetum flowers with the individual RGB channels below and the UV-image on the upper left side.

 

The next image shows how the three RGB channels are used in order to achieve the false colour representation of the bee-colours (fig. 2): The red-channel of the visual image is not used. A greyscale representation of the UV image is assigned to the blue channel (false-blue), blue has been moved to the green channel (false-green) and green is represented in the red-channel (false-red). The upper right image shows the resulting false-colour image composed of the false-colours assigned to UV + blue + green, the parts of the spectrum that are visible to bees and other pollinators.

 

Figure 2: False colour representation of bee-colours created by reassigning the RGB channels to green, blue and UV as shown.

 

Since the Chrysanthemum flowers shown here have an even UV reflection / absorption, the flower pattern is not modified. Only the colour appearance is changing between the visual image and the false-colour representation of the bee-colours: Visual white is represented as false-yellow and visual yellow is represented as false-red in the false-colour representation of bee-colours.

 

 

As a second example we will look at the flowers of a Myosotis species (fig. 3). With the exception of the one flower that shows some rose tone, the visible colour of the petals is pretty much the same blue for all the flowers. However, the flowers differ strongly in their UV reflection.

 

Figure 3: Visible light image of several Myosotis sp. flowers with the individual RGB channels below and the UV-image on the upper left side.

 

Fig. 4 shows the false colour representation of the bee-colours. Flowers that are UV-dark are rendered false-green while the flowers that are UV-reflective appear false-cyan. Flowers with a more and less reflective areas show both false-colours.

 

Figure 4: False colour representation of bee-colours created by reassigning the RGB channels to green, blue and UV as shown.

 

 

Image references:

Fig. 1-4: Copyright, Nicolas Chalwatzis, 2013

 

 

Published 29 December 2013

Modified (Figure 1) 30 December 2013

edited and finalized 4 January 2014

[nfoto]

Another aspect, not directly related to the colour encoding scheme(s), is the relative weights to be assigned to the different spectral bands (or equivalently, to the channels in the false-colour composite(s)). We do know the camera is exposed to, and records, spectral bands with highly variable intensity between them so irradiance neither is spectrally "white" on the input side nor recorded proportionally to band intensity on the output side of the lens. There are virtually always necessary to utilise bandpass filters to split up various bands so there are filter transmittance curves to consider too. Finally, the biological efficiency or spectral response of the organisms should be factored into the overall picture as well. Thus we ideally require biological dose-response curves for the spectral ranges under study. Such relationships are established in certain fields, eg. cancer research, but I'm not aware of the similar knowledge in the field of UV vs pollinator studies. However the apparent lack of information might well be on my side as I enter this field mainly as a photographer not a biologist.

 

In the "pure" UV documentary work, when camera/lenses are profiled to make reproducible false-colour records, the UV range recorded (typically something between 340/350 to 390/400 nm) is treated as "UV white" by implication and this approach pans out fairly well in practice when UV bandpass filters covering a broad band are deployed . Factor in the daylight spectral distribution and the filter response, and the typical UV band shelf from in the same range is transmitted to the camera in an energy "flat" manner. However, when portions of the solar spectrum outside the UV-A are included to make the final composite, I perceive a definite lack of insight how to balance these components in a biological meaningful manner. I feel better insights here are a prerequisite before the emphasis is moved further into a question of what colour models should be used for a simulation of these false colours.

[Nico]

I do well understand the concept of the biological dose response that we are lacking. However, one should also consider that the difference between pure UV and visible light is pretty arbitrarily defined by the spectrum visible to humans. We know that the boundary is different for other organisms. So, what we define as UV-photography is just the bandwidth that our (modified) cameras can still record but our eyes cannot detect. I do not see a “natural” or principal difference between a filtration that would transmit UV plus a bit of the adjacent visible spectrum and “pure” UV photography.

[msubees]

Nico,

 

fasinating discussion. so in photoshop, how are the false color arrived? blending two photos? one in UV, and one in VIS (but with blue/green switched in channel mixer?).

[Andrea B.]

Create a new RGB image by

• putting the Visible image green channel into a new red channel.
  
• putting the Visible image blue channel into a new green channel.
  
• putting the UV blue channel into a new blue channel.
  

Combine the new RGB channels into one image.

 

It is lots of fun playing around with these kinds of mappings to model insect vision or to create 'multi-spectral' images.

[Nico]

Zach & Andrea,

 

I'm certainly not a Photoshop guru, but I have figured out at least two different ways. Maybe slightly different from Andrea's suggestion.

However, the results should be pretty identical.

In fact I have prepared a number of screenshots illustrating my prefered method, which is quite simple.

 

Bear with me. I plan to publish it soon.

 

Best, Nico