• Ultraviolet Photography

Bird vision (again)

44 replies to this topic

#21 Andy Perrin


  • Members+G
  • 3,995 posts
  • Location: United States

Posted 22 November 2020 - 22:36

Jim - they make robotic panorama shooting heads, like the GigaPan Pro. You may want one?

#22 Cadmium


  • Members+G
  • 4,141 posts

Posted 23 November 2020 - 01:30

Yes, but one can do it hand held, or at least on a tripod. I don't use mine anymore. Mine is the Papywizard setup.

Attached Image: Untitled_HDR2a12.jpg

#23 Andy Perrin


  • Members+G
  • 3,995 posts
  • Location: United States

Posted 23 November 2020 - 03:55

Cadmium- I know you CAN do it by hand, but looking at how much trouble Jim had obtaining even coverage, I just thought having something take them for you would give better results. Why don’t you use yours anymore?

#24 UlfW

    Ulf W

  • Members+G
  • 1,567 posts
  • Location: Sweden, Malmö

Posted 23 November 2020 - 09:46

Remember that the result from these types of stacks can vary quite a lot depending on light situation and differences in Bayer sensor responses.

I cannot see any of the red Cadmium show in post #11 above, when I am using the same filter stack on my Canon 60D.
My Canon sees the world differently than Cadmium's Nikon.

The outcome also depends much on how the raw file is white balanced, specially with these filter stacks.

WB against PTFE or Spectralon is not always (seldom?) a good alternative.
That opens up for a wide spread of different artistic interpretations.

The adjustments needed to reach a real proper WB against PTFE is also quite extreme and sometimes out of reach for some raw-converters.

Edited by UlfW, 23 November 2020 - 09:48.

Ulf Wilhelmson
Curious and trying to see the invisible.

#25 Jim Lloyd


  • Members(+)
  • 328 posts

Posted 23 November 2020 - 12:59

Thanks Andy and Steve - re panorama heads.

These look interesting - I could probably find somewhere in the university to borrow one. Yes it was a bit tricky doing in manually (on a tripod). It's interesting / frustrating that even if one goes to this trouble you end up with an image that you then view (obviously) with human eyes in a similar way to how you would view the original scene - so I don't particularly feel like I am seeing like a bird - who would be (depending on species) aware of what was in front and behind simultaneously. Still, like the colour aspect worth exploring further ...

Also I think I need to look think about the bird's detailed and binocular field of view - which is far more limited, rather than total monoccular field of view. To complicate matters I understand that some birds have two foveas or foveas that are not circular. Probably I need to focus on a specific species rather than talking in generalities.

Thanks Ulf

Yes I agree - this particular combination I find particularly difficult to WB and software dependant. The only reliable way I have found so far is to use a spot WB using View NX2 (on something in the image - like say a branch or other grey object, seems surprisingly forgiving on what is chosen) - then save this as a jpg and do fine adjustments in Lr. I have also tried, Photoshop, Photo ninja and FastRawview. I did try PTFE reference which was OK, but not perfect.

#26 Cadmium


  • Members+G
  • 4,141 posts

Posted 23 November 2020 - 21:59

I use PTFE for UV-only-pass shots, but I don't use PTFE for UG5/U-330 stacks.
I use Photo Ninja white balance using RAW file for almost everything, NX-D works also, but I prefer Photo Ninja these days.
For UG5/U-330 stacks I usually sample the whole screen, sometimes I sample a smaller area, say a gray area, etc..
Ulf is right, different cameras may work differently, even when WB'ing for RAW, I don't use any other cameras except the Nikons.
NX2 (CNX2) works even better than NX-D for me, if the camera is old enough, NX2 does not support newer Nikons. NX2 and NX-D work about the same, but NX-D will not do a full frame sample, it can only do a small sample,
but it works if you find a gray spot. Like I said, I have slowly moved away from using either of those, because Ninja seems to work best for me.
If you are using a Nikon, then I would think you will get similar results as me.
The example I show with the red flower was done using NX2, before I have NX-D or Ninja.
If you like, I can resample the red flower using Ninja...?

I think the three best stacks for UV+Blue+Green+Red (aka bird vision) are these, and the first of which I have tested and shown here.
UG5 (or U-330) 1mm + S8612 2mm
UG5 (or U-330) 1mm + S8612 1.5mm
UG5 (or U-330) 0.75mm + S8612 2mm
(however, I can't find any UG5 0.75mm or any S8612 1.5mm at the moment)

Here are the three best bird vision stacks I can think of.
I will retest the 1 + 2 stack tomorrow if I can, but unless I can find the other thicknesses I can't compare the difference, but they should have stronger red.
Attached Image: Bird_Vision_3_Stacks.jpg

Here is the difference between UG5 and U-330, which I don't detect in actual photos.
There is a big difference in price, UG5 being more expensive. U-330 having the data drop out, which is common for some of Hoya's data sheets.
Attached Image: U330_1mm_2mm_UG5_1mm_2mm.jpg

Edited by Cadmium, 25 November 2020 - 16:56.

#27 Cadmium


  • Members+G
  • 4,141 posts

Posted 24 November 2020 - 01:49

By the way, Andrea might have a UG5 or U-330 1mm thick filter. She has an S8612 1.5mm, 1mm, and even 0.75mm.
She might be too busy to be testing things though.

Edited by Cadmium, 24 November 2020 - 02:00.

#28 Cadmium


  • Members+G
  • 4,141 posts

Posted 24 November 2020 - 06:06

Took me a while, but I found the original NEF (RAW) file.
This time I used Photo Ninja to white balance. It was a little over exposed, so I adjusted that as well in Photo Ninja.
Looks about the same as the previous old version.
This is UG5 1mm + S8612 2mm.

Attached Image: UG5_1mm_stack_redo_1280w.jpg

#29 Jim Lloyd


  • Members(+)
  • 328 posts

Posted 24 November 2020 - 11:49

Thanks for all the information Steve

My theoretical concern was that added the red might be at the expense of seeing contrasts with the UV, but your image above appears to dispel that concern.

#30 Cadmium


  • Members+G
  • 4,141 posts

Posted 24 November 2020 - 19:27

I will try to do a new test on something red today if possible.

#31 Cadmium


  • Members+G
  • 4,141 posts

Posted 25 November 2020 - 02:14

These didn't work out exactly like I had anticipated.
Conditions were not good, overcast skies, no sunshine.
With the very red plastic drill pit case I expected less red with the first stack and more red with the last stack, more like the older example I have posted.
I did find an S8612 1.5mm thick to include in the comparisons.

Attached Image: U-330_Visual_Comp1.jpg

Attached Image: U-330_1p5mm_x_S8612_2mml_Comp2.jpg

Attached Image: U-330_1mm_x_S8612_2mm_Comp3.jpg

Attached Image: U-330_1mm_x_S8612_1p5mm_Comp4.jpg

#32 Cadmium


  • Members+G
  • 4,141 posts

Posted 25 November 2020 - 06:51

OK, here we go. Here is a really simple good example from I did this past summer, shows the difference between 1mm + 2mm stacks using both UG5 and U-330 with a red flower and rudbeckias side by side.
Much better example with better light.

#33 Jim Lloyd


  • Members(+)
  • 328 posts

Posted 05 December 2020 - 22:07

Attached Image: thin-8.jpg

Now received U-330 1mm and S8612 1mm. [EDIT - correction these are with S8612 1.5 mm]
These on D3200 full spectrum camera Nikon E 50 mm
Afternoon sun, but low UV this time of year.
Hand held iso 1600, f/5.6 around 1/30 sec.
White balance View NX2.
Channel swap Ps
Adjustment in Lr

Sunny periods are infrequent and short lived at the moment. Will do proper tests later.

Attached Image: thin-2.jpg

Attached Image: thin-3.jpg

Attached Image: thin-4.jpg

Attached Image: thin-5.jpg

Attached Image: thin-6.jpg

Attached Image: thin-7.jpg

Edited by Jim Lloyd, 06 December 2020 - 09:53.

#34 Cadmium


  • Members+G
  • 4,141 posts

Posted 05 December 2020 - 23:33

Jim, Are you doing red/blue channel swap?

#35 Jim Lloyd


  • Members(+)
  • 328 posts

Posted 06 December 2020 - 07:39

Steve, it was a triple swap - I need to go back to check but it was one of

R to G, G to B, B to R
R to B, G to R, B to G

The second one is a rotation which (sort of) makes sense for the bug vision - assuming that the R channel is responding to UV, then it can be thought of as shifting from the bug view to human

Not sure if there is any logic for the other version

Not sure if it makes sense in this case where there is visible red present

Of course colour vision is a lot more complex than just response in three channels in humans - so trying to think of what it is like for birds with 4 channels is nigh on impossible

So I am thinking of producing images using a little science and a lot of intuition

A fairly modest aim of (hopefully) producing images that defamiliarise our view of the world

Thinking of these printed very large in a gallery

Or maybe projected in a slide show ?

Early morning here - waking thoughts ....

#36 Cadmium


  • Members+G
  • 4,141 posts

Posted 06 December 2020 - 07:53

Well, OK. :smile:
Personally, at least as a starting point for myself, I would like to see some nu-swapped images using the new filters stacks, just to see how they look, just optimally white balanced, say a full frame NX2 'marquee', say,
so I can get my footing.
You have U-330 1mm, right? and S8612 1mm? 1.5mm? 2mm?
I would like to see, again, just my personal interest, 1mm + 2mm, 1mm + any other
, basically all combinations compared, white balanced the same way optimally, making sure you have something red in the frame to judge the red by...

#37 Jim Lloyd


  • Members(+)
  • 328 posts

Posted 06 December 2020 - 08:04

Yes - I will do all that once we have some decent reliable sun

#38 Jim Lloyd


  • Members(+)
  • 328 posts

Posted 06 December 2020 - 08:08

Yes - I will do all that once we have some decent reliable sun

Starting off here with my art hat on ...

#39 Jim Lloyd


  • Members(+)
  • 328 posts

Posted 06 December 2020 - 10:05

Hi Steve

Sorry I made an error above (correction added) - the new images are with U-330 1 mm plus S8612 1.5 mm (not 1 mm as I originally stated). I also have U-330 1.5 mm and S8612 2 mm.

I agree with you that in the right conditions my inclination is to make the tests you suggest. My quick informal test did indicate to me that I am seeing more red as expected. I think it very unlikely to learn much as a surprise with more thorough tests as you have already nicely demonstrated this and the charts would strongly suggest the red coming through more with the thinner filters. I guess what is of particular interest is whether the additional red transmission is at the expense of any UV specific features. So yes I will do them some time.

My particular issues doing this as an arts practice based project is how to present images to people who know very little about the scientific theory behind the images. Of course one approach is to try to explain that to them in advance (or along side the images in a gallery). The real problem I face is around the "truthfulness" of the images for people who don't know the background to how they are produced. It seems to me there has to be some clues in the images themselves - but I don't know how. Anyway these are the sort of questions I grapple with as part of my research ...

Somehow I think I need to frame the images as "images produced by a system that responds to a wider range of wavelengths than the human eye can detect - a range that is accessible to various animals including birds" - rather than "what a bird sees" - but obviously I need to have a much snappier way of saying that!

One could maybe say the ambition is downgraded from "reproducing bird vision" to "using what we know of bird vision to produce images that make us see the familiar world anew".


Steve, how do you do "full frame marquee" for WB in NX2 - I seem to remember having this same issue before - I can only seen how to do a point correction (or up to 5x5 square)

Edited by Jim Lloyd, 06 December 2020 - 10:18.

#40 Cadmium


  • Members+G
  • 4,141 posts

Posted 06 December 2020 - 23:10

Jim, Yes the U-330 1.5mm + S8612 2mm is the standard UV+Blue+Green stack, so called 'bee vision'.
U-330 1mm + S9612 2mm is the 'bird vision', UV+Blue+Green+Red stack that I have used.
I have not tried U-330 1mm + S8612 1.5mm or S8612 1mm stacks before the fast tests in overcast skies I tried above, and because the weather was dark and I had no rudbeckia to shoot... I don't consider those test to be very conclusive.

OK, in NX2, load RAW file (NEF).
right hand menu: Camera Settings > White Balance > Get Gray Point > Marquee > Start, then click/hold mouse butting wile dragging mouse from top left to right bottom corner, then release mouse button, that will select the full frame and use the full frame sample for the white balance, same can be done for any smaller area of the photo, to sample a smaller area.
See photo below for example. Works for UV or IR or any shots suing RAW.

Attached Image: CNX2_Tree_Marquee.jpg

Furthermore, I load the resulting tiff from the NX2 into Photoshop, and I often do Auto-Levels, or Auto Contrast, I try those to see how they look, and one or the other usually 'cleans them up' to me eye,
but if they don't look better then I leave the photo alone. Sometimes auto levels doesn't look better but bakes it look worse, so then I try just auto contrast, which can be better...
Usually one of those two clean up the photo, improve the contrast slightly...
Then I might sharpen the image, depending.

Edited by Cadmium, 09 December 2020 - 06:54.