44 replies to this topic

[Jim Lloyd]

Hello again.

Just to remind people, I am doing an art PhD project about bird vision. I made some progress with very generous help from people on this site a while back and then I got sidelined into other avenues (mainly a project on translating birdsong and then writing poetry). I had become rather disillusioned that trying to represent bird vision was too difficult. But I'm going to give it another go.

Birds have 4 colour receptors evenly spaced in UV,R,G,B or violet,R,G,B - there is a lot of individual species difference, but basically they follow that pattern. UV peak sensitivity is typically 370 nm.

I have only read a little, but I think that they way these channels are combined and interpreted by the bird is not fully understood.

I did think about obtaining aligned UV and visible images and doing some processing to merge into a single RGB image, but I am not sure if this will really be better than using a suitable filter.

So for now I am working with U-330 1.5 mm plus S8612 2mm. Nikon D3200 multispectral. I have several 35mm, 28mm, nikkor enlarger lens and nikon E series lenses.

(I also have UG2A 2 mm which I might try out, but I think cuts out too much of the longer wavelength visible - also have BG 40 2 mm as an alternative IR block, but I seem to remember this might cut too much red - but something to try ...)

I have decided that I will just have to use some artistic imagination. - But I think I should at least start out with a reasonably white balanced image. I have found this a bit tricky in the past, but have opted for now to use Nikkon view NX2 to do a spot white balance against something grey in the image and then do finer auto adjustments in Lightroom, Then play around with boosting colour sharpness contrast etc,

Second thing I have decided is to try to get a similar field of view to bird - this of course varies with species, but tend to fall into 3 broad groups - most commonly around 240 degrees horizontal field of vision (this is both eye, tends to be largely mon-occular with a narrow binocular portion) and more limited vertically (need to check the figure). I have had a look on this site at fisheye lenses. I like some of the images, but generally it seems complex to mount the filters, plus its not really giving the right field of view. So I think I will use image stitching of multiple images to do this.

Thirdly I will aim to take images representative of particular bird's habitats.

Today was my first attempt and I will post other images here as this develops - I would be interested in any thought you have.

So to start just a couple. A cloudy day in late November afternoon (Northern England), might not be the best time !.

These are handheld ISO 1600 Photax 28mm lens f/5.6 1/15. Both are two images stitched together in photoshop (just to check the principle)

I am encouraged to do more - obvious next step is with tripod and getting the angles for FOV correct.

I am not really sure why I am getting these colours - not sure why the scene in the distance is redish. I don't remember that last time I tried.

Anyway I am going to press ahead. Aim is to print these large - about 1.5m (5 ft) tall ( I mean eventually when I have images I like ...)

Edited by Jim Lloyd, 20 November 2020 - 23:27.

[Stefano]

Interesting. I have some thoughts:

You said you are using U-330 (1.5 mm) + S8612 (2 mm). This should be the typical BUG stack Steve uses and recommends. This stack removes red, to have red too you need to lower the thickness of the U-330/UG5 filter: https://www.ultravio...dpost__p__38012

So, the images above do not have red, since you used this filter stack instead of stacking a UV and a visible image like I did in my IRG attempts, right?

You also said that you assigned colors to your images using a bit of imagination, and that makes sense, as I don't think you can make people perceive 4 primary colors, since we only have three types of cones in our retinas (some people may be tetrachromats, but that's a different story). Our red and green cones are already quite similar, so we are already somewhat limited.

As of your reddish color in the distance, you can see that it is haze. That makes me think it is UV or blue/violet, as the shortest wavelengths appear as haze and as the dominant color in the sky.

Edited by Stefano, 21 November 2020 - 00:57.

[Cadmium]

Yeah, the typical U-330 1.5mm + S8612 2mm is UV+Blue+Green, some times called 'bee vision'.
Side note: I do like that tree shot by the way.
For some red mixed in, you need to use U-330 (or UG5) 1mm.
Also, there is the new UG2A 0.7mm thick, very expensive though, twice the price of the U-330 1mm.
Here is a comparison graph below.
I have not used the 0.7mm yet. It is an Ulf special, his idea.
I am kind of interested to find out how the UG2A 0.7mm and the U-330 1mm compare in actual photos.
You 'bee' the judge.

Edited by Cadmium, 21 November 2020 - 01:21.

[Cadmium]

How about this stack? You could also use S8612 1mm for this stack, increases the red slightly but crosses the 700nm line up to about 730nm (second graph set).

Edited by Cadmium, 21 November 2020 - 01:54.

[Stefano]

Steve, that's interesting. You have up to orange and a touch of red. 600 nm is orange, I have an LED emitting there. Red starts at around 620.

[Cadmium]

Stefano, Anything above the 1E-03 line will show up. Not sure what you are referring to.

Edited by Cadmium, 21 November 2020 - 02:02.

[Stefano]

You reach almost zero at 600 nm (orange), and you have red up to 700 nm+, not much, just a bit. It is a secondary bump. With the typical BUG stack, you don't reach as far, only about ~565 nm at OD 3. 565 nm is lime green, yellow is at about 580, so no orange.

[Cadmium]

...but, zero isn't zero, it would shows up, more than the 1mm + 2mm does.

Edited by Cadmium, 21 November 2020 - 03:26.

[Jim Lloyd]

I've missed this forum! - You guys are so knowledgeable and helpful! - On the other hand I am rather rusty on all this.

Anyway - of course you are right - my combination is "bee vision" - I had forgotten that - I was thinking there was some red in there.

Just did a few tests to satisfy myself on that with a few berries and some gorse flowers. I know from previous tests that these berries do not reflect UV (apart from specular reflection from the shiny surface). Same lens and camera as before. Placed in window sill - so sunlight through glass. First image visible (just BG40 2 mm filter). Then the "bee vision". Third image is with Optomax 35mm lens which has slightly better UV transmission - and with brighter sunshine.

Despite my error - I am now wondering whether this filter arrangement might actually work for my purpose. Yes I am missing red, but then something has to give as birds have 4 channels and we have 3. So the UV can use the red channel and the G and B stay the same? I wonder if I did add in the red whether the UV contrasts would be less noticable? Of course the red/green contrast must be important for birds that forage on berries.

Anyway I need to give this al some more thought...

Overall the aim is to produce images that are aesthetically pleasing (what ever that means) - whilst retaining some "truth" (whatever that means!) - with the effect that the viewer has some feeling of glimpsing into another world

PS - I also tried U330 1.5 mm with BG 40 2mm - I was surprised that the result was rather too much like the visible to be that interesting - so not included it here.

Edited by Jim Lloyd, 21 November 2020 - 14:43.

[Jim Lloyd]

Thinking more on using the R channel for UV then led me to think that channels can be swapped to give examples shown below:

- Maybe I could choose one - I think the r-b is most logical, but others are attractive.

Or maybe in a gallery situation one could display all of these - making a visual comment about the difficulty on imagining how others see?

Probably I prefer the original - best not to produce anything that looks too much like IR / aerochrome - or some type of known film / look. Possibly the RB swap or RGB swap2 works?

RB swap


RG swap


RGB swap 1


RGB swap 2

Edited by Jim Lloyd, 21 November 2020 - 19:02.

[Cadmium]

Yes, the standard UV+Blue+Green stack (the so called 'bee vision') is UG5 (U-330) 1.5mm (or 2mm) + S8612 2mm.
To get red mixed in you need to drop the UG5 (U-330) thickness down to 1mm thick + S8612 2mm.
I will show a comparison of that below, showing how some visual red shows up when the UG5 (U-330) is thinner.
Also, here is a graph showing all the combinations I can think of that might be good to compare for doing 'bird vision'.
I hope this helps determine what might be the best stack to choose.

Turn red on and off:

[Cadmium]

Myself, the UG5 (U-330) 1mm + S8612 1.5mm looks like it would be the best pick for stronger red and cutting off under 700nm. Purple dotted line. Bird Vision!
However, the 1mm + 2mm stack works also.

Edited by Cadmium, 22 November 2020 - 02:27.

[Stefano]

Steve, I know it is a tedious and time-consuming task, but if you already have all the ingredients (S8612 and U-330 at all thicknesses), can you try at least some combinations to see what comes out? There are so many stacks we haven't tried, they may yield interesting results. If you want and when you have time, of course.

[Cadmium]

It's cold out there, rainy, windy... I have already tried all those except the U-330 1mm + S8612 1mm and 1.5mm, and the UG2A 0.7mm + S8612 1.5 (etc.).
The UG2A stacks will not work good for bird vision stacks, and I don't think the U-330 1mm + S8612 1mm stack is good, has too much 700nm+.
Like I said, the two I would use are the U-330 1mm + S8612 2mm or 1.5mm, either of those will work to get red like the flower I show about.
I have already tested the U-330 1mm + S8612 2mm, that is shown in the photo.
Sometime I may try the U-330 1mm + S8612 1.5mm. I don't even know if I have any S8612 1.5mm thick right now, I don't see any here, but I might have one.
I use to have a UG5 0.75mm thick version, but I looked around, don't know where it is. Theoretically that would work nicely also, but anything under 1mm is really hard to make, fragile, and expensive.
Quick answer, you will need the U-330 1mm first, get that, use the S8612 2mm that you already have,
and then sometime get the S8612 1.5mm thick version, but I don't think you need that, but you WILL need the U-330 1mm.

NOTE: By the way, a U-330 1.5mm + S8612 2mm stack will not see red either.
My example above shows 2mm + 2mm, I have tried both, neither 1.5mm + 2mm or 2mm + 2mm will get any red, but 1mm + 2mm will.

Edited by Cadmium, 22 November 2020 - 05:43.

[Stefano]

Thanks Steve for the information.

[Jim Lloyd]

D3200 camera, Photax 35 mm lens, iso 800, 1/5 sec, f/5.6 U-330 1.5mm, S8612 2mm.
Late November, Northern England, 2 pm, bright sun, cold and windy

Only managed to a get a couple of poor quality shots as it is very cold and windy - but I can see the red berries - also gorse flowers showing as yellow and green.

I'll try and get some better versions ...

[Jim Lloyd]

The other part of what I am trying to do is to get the field of view more bird like.

This is my first attempt. Camera set up as above. About 50 separate images stitched together in photoshop after WB each one individually using View NX2. Then RB channel swap and then some enhancements in Lr.

Uuuummm ? definitely work in progress ... !

I will need to think about this ...

[Cadmium]

Stefano, I can tell from the graphs what will happen, given I have tested the U-330 1mm + S8612 2mm, and the result is shown.
The difference on the graph when using S8612 1.5 instead is noticeable, and not extending about 700nm, and will produce more red.
However, I don't have any S89612 1.5 right now, even if I could find a warm coat. However, I am sure it would work nicely given the graph.

Jim, Nice composite pan... Here is an example of the kind of landscape I get with the U-330 (UG5) 1.5mm + S8612 2mm stack:

[Jim Lloyd]

Regarding the amount of red needed ...

I think it would be good for me to try the thinner U-330 to get more red, but I am not certain in the end whether it is the right way to think about "bird vision" - the combinations discussed above essentially combined UV and red into the R channel of the camera. I don't know if this is how the cone responses in a bird would be combined, but my hunch is probably not.

But as I say I think I should do the practical experiment and also read up more on this as part of my research.

Many thanks Steve for all the information.

[posted simultaneously with Steve]

Edited by Jim Lloyd, 22 November 2020 - 21:02.

[Jim Lloyd]

Thanks again Steve,

Nice landscapes - they broadly look similar to what I have been getting - (I have been playing with pushing saturation etc for artistic effect). Although I did get more red than expected in the images above with the berries (the outdoors ones).

I think it has been discussed somewhere else on the forum - but is UG5 interchangeable with U-330?

[Steve, I sent you a personal message, I seem to remember in the past there might have been some technical issue in you getting them via the forum?]