Channel Stack: UV channel or UV monochrome? (And comparison to Filter Stack)

[Alex H]

EDITOR'S NOTE:

 

I split Alex' question from the thread it was posted in so that we could pursue it in more detail. To review, a "channel stack", for our purposes, is basically a 3-layered (R-channel, G-channel, B-channel) difference set where different wavelengths of light are placed in separate channels. There is more than one technique for making a channel stack depending on which editor or tools are used.

 

Here we are concerned with making channel stacks which combine UV and Visible layers. For example, Bee Vision can be simulated with a channel stack made up of UV, green and blue layers where the UV is placed into the Red channel.

 

To set the stage, I had shown a channel stack for a flower foto made up of the UV red channel and the visible blue and visible green channels.

 

***********************************************************************************************************

 

 

Andrea, why do you split UV shot into three channels? Shouldn't it be just treated as equivalent of one channel for composing? E.g., just converted to monochromatic?

[Andrea B.]

I have no answer Alex, except that is how I was taught to do this. But I will try what you say and post here in this thread. "-)

Thank you for a very good question.

[Andrea B.]

Wait a minute.....

 

If you take the UV shot, make it monochromatic and then place it in the blue channel, then that is the same as the blue channel of the UV shot.

They are the same.

 

I'll make an example in my Affinity app later and provide screen shots.

[Alex H]

Wait a minute.....

 

If you take the UV shot, make it monochromatic and then place it in the blue channel, then that is the same as the blue channel of the UV shot.

They are the same.

 

I'll make an example in my Affinity app later and provide screen shots.

 

Not necessarily the same.

The blue channel will be the noisiest in many cases.

Also, try to use something like Bidens or Rudbeckia to split channels. Green channel will have higher contrast and brightness.

Or instead, use a flower that is predominantly "blue" in UV. There will be differences between different channels, at least in contrast and brightness of different areas.

Just food for experimentation.

[Andrea B.]

I am trying Alex' suggestion.

 

Flower: Heliopsis helianthoides 'Tuscan Sun' which has a central bullseye in UV

Equipment: D600-broadband + 105/4.5 UV-Nikkor

 

Visible [f/16 for 1/200" @ ISO-200 with Baader UVIR-Cut and On-board Flash]

 

UV [f/16 for 20" @ ISO-200 with BaaderU and Nikon SB14-mod]

Long exposure was taken for multiple flashes.

As Edited for Display in UV Botanicals

 

UV Raw Composite

Additional White Point Reset in NX2 to brighten the frame.

 

UvBG Filter Stack [f/16 for 20" @ ISO-200 with UG5(1.5mm) + S8612(1.75mm) and Nikon SB14-mod]

As Edited for Display in UV Botanicals

I'm adding this frame so that we have something to which we can compare the channel stack.

Note that, as predicted by what we know about bee visual receptors, the yellow or yellow-orange, UV-absorbing bullseye appears as green because it stimulates only the bee's green receptor. The yellow, UV-reflecting ray tips which stimulate the bee's green and UV receptors appear as yellow.

 

UvBG Filter Stack: Raw Composite

Additional White Point Reset in NX2 to brighten the frame.

This filter combo does pass some red light, but it seems not to affect the final result (preceding foto) too much.

But it's better to use a thicker UG5 or S8612 or both.

 

Continued in next post.

[Andrea B.]

I must post this experiment in two posts because I have so many fotos to upload.

Now I'm going to get the RGB channels and construct a channel stack.

 

Channel Stack using Bvis + Gvis + Ruv

I'm going to make a channel stack which use the visible blue and green channels and which assigns the UV red channel to the red channel.

Question: Which UV version should I use in that red channel - the edited final display or the raw composite?

I will try both UV versions to see what happens.

 

 

Everyone is free to download these channels and play with them. :D

 

 

Visible Channels: Blue, Green, Red

In an expanded browser these fotos will display as two rows of 3 each for a total of 750 pixels in width.

There is little to nothing in the Blue channel, of course, because the flower reflects yellow to yellow-orange.

If brightened a lot the background comes out a bit, but the flower stays dark.

 

 

 

UV Channels from UV Edited Foto: Blue, Green, Red

In an expanded browser these fotos will display as two rows of 3 each for a total of 750 pixels in width.

 

 

UV Channels from Raw Composite Foto: Blue, Green, Red

In an expanded browser these fotos will display as two rows of 3 each for a total of 750 pixels in width.

 

 

Channel Stack #1: Bvis + Gvis + Ruv where the Ruv is from the UV Edit

This is very simillar to the UvBG filter stack bee vision photo shown in the preceding post.

The red shadow error could be cleaned up.

 

 

Channel Stack #2: Bvis + Gvis + Ruv where the Ruv is from the UV RawDigger Raw Composite

The UV raw composite frame is very much lighter than my edited UV shot. So look what happens when used in a stack!

We cannot say this channel stack is "wrong", only that it is different. I suppose if I wanted my UvBG bee-vision channel stack to look similar to my UvBG bee-vision filter stack in the preceding post, then I would reject this version.

The primary cause of the difference (I think) is that the edited version of the UV photo has had a white balance edit applied to its false colours while the raw composite has had no white balance edit. So this gets interesting. We would never want to use a visible raw composite in a stack because a visible raw composite simply does not show us the true colours which we saw when shooting. So can we extend that to the UV case and reject the UV raw composite for use in stacks because it does not show us the "true false" colours that we see in the UV shot after white balancing?

You get to decide!!!

 

Continued in next post.

[Andrea B.]

Alex' suggestion was to use the entire UV frame iin a channel stack rather than any particular channel. So we need a monochrome version of the UV frame.

The benefit of this stacking technique is that if the UV is to be placed in a blue channel, then converting the entire RGB UV frame to monochrome for use in the blue channel will give you a less noisy stack given that blue channels are the noisiest.

 

I'm making a bee vision stack, so I'm going to put the monochrome UV into the red channel. But we can still compare this kind of stack to the preceding versions. Usually you will see some differences regardless of where you place the monochrome UV into the layers.

 

Monochrome UV Foto and Corresponding Red Channels

Top left: Desaturated UV foto

Top right: B&W conversion of UV foto

Bottom row: Corresponding red channels

The black & white conversion is brighter and a more contrasty than the desaturated version. No surprise there.

 

 

Stack #3: Bvis + Gvis + R(UvDesatMono)

This bee vision stack is less contrasty than all the other stacks. The desaturated version of the UV foto is just too dull to contribute well to the stack.

 

 

Stack #4: Bvis + Gvis + R(UvB&wMono)

Naturally a B&W conversion of any foto will look better than a simple desaturation. And the improved contrast and/or brightness improves the channel stack.

[Andrea B.]

Final Roundup: Four UvBG Channel Stacks

 

Let's look at all four channel stacks together. Each is labeled. Which one do you like best?

If browser is expanded, you will see a block of 4.

 

 

 

Here the corresponding red channels which were used.

 

 

 

 

And now I'm tired and do not wish to look at anymore R, G or B for awhile !! :D :D ;) :D

[Andrea B.]

My Conclusion:

The properly edited, white-balanced UV frame seems to work best for stacking.

If stuffing an entire UV shot into one channel, then a B&W properly toned conversion seems to work best for stacking.

 

 

 

[Added Later] oh giant flapdoodle....I forgot to do a final comparison of the UvBG filter stack to the UvBG channel stack.

 

 

 

Comparison #1: Filter Stack vs. Channel Stack

Left: UvBG Filter Stack with UG5 (1.5) + S8612 (1.75)

Right: UvBG Channel Stack with Bvis + Gvis + Ruv

 

 

Comparison #2: Filter Stack vs. Channel Stack

Left: UvBG Filter Stack with UG5 (1.5) + S8612 (1.75)

Right: UvBG Channel Stack with Bvis + Gvis + RuvB&Wmono

 

 

 

These comparisons are VERY interesting. The Filter Stack gives us a better photograph, no surprise there.

But it is difficult to say whether the Filter Stack or the Channel Stack is a better simulation of bee vision.

It might depend on what you believe about bee vision after taking into account all its complexity.

 

Two basic outcomes we need to see are that:

• yellow, UV-absorbing areas are rendered as green to represent bee-green, and
  
• yellow, UV-reflecting areas are rendered as yellow to represent bee-UV-green
  

That happens in both the Filter Stack foto and the Channel Stack foto. Good.

 

The central disk of this flower seems to be a bit of a problem. The florets on the disk are the same yellow (or yellow-orange) as the rays. But in the Filter Stack foto, the desired green is less obvious in the disk. I think that maybe the shadows in the disk play a role. The green is there, just not as much of it.

Here is a 100% extract from the disk:

.

 

 

 

There are brightness and saturation differences between the Filter Stack and the Channel Stack fotos. I'd have to think more about that. But it is rather obvious that if you send the saturation of the Filter Stack foto up to the stratosphere and brighten the green, then it looks more like the Channel Stack foto. I kinda like the Filter Stack foto the way it is.

 

Filter Stack Foto with High Sat and Bright Green. Yikes, the blue came along for the ride!

[Alex H]

Hi Andrea, what is "B&W properly toned conversion"?

Alex

[Andrea B.]

An attempt to describe conversion to black & white followed by an effort to ensure the brights are not blown and that the shadows are not blocked and that there is some local contrast. Seems to me like an effort to get the b&W tones in order would make for a better layer in a stack? When I used the simple desaturation of the UV foto in a layer, the result was too flat with less demarcation between the yellow and green.

 

All this is open for discussion, of course.

 

For the record, when I tried using a monochrome version of the Visible foto in the channel stack, everything fell apart. It looked awful.

 

I hope sometime that someone else comes along and tries some of these stacks. What if I did everything all wrong??!!

[Alex H]

The reason I am asking is that in PSE conversion to black&white always involves unequal treatment of RGB channels, while desaturating is equal. There is no problem is one adjusts contrast/levels/curves in desaturated image, of course.

 

I am more and more leaning towards using monochrome camera for UV...

[Andrea B.]

Alex, it would be so cool to see some nicely photographed monochrome UV work. Please go for it!

 

Thank you for asking these questions because it makes me think and gives me ideas for experiments or example posts. "-)

I'm hoping some of the bee vision followers will try some of their own experiments when they see these threads.

 

I will make another stack comparison with 3 monochrome UV fotos used in the stack.

This time I think I'll put the UV into the blue channel instead of the red channel.

UV Monochrome:

1) desaturation with no adjustments

2) desaturation with adjustments (bw points? contrast? local contrast? whatever improves the foto.)

3) some B&W conversion from some editor (vague, I know)

 

For the Heliopsis above there did not seem to be large differences between the channel stacks. But differences could show up depending on UV-signature or where the uV is placed.

[Alex H]

I do not have monochrome camera any more, so these are the only results that I have posted on UVP: http://www.ultravioletphotography.com/content/index.php/topic/758-sony-nex-5n-monochrom-conversion-results/page__view__findpost__p__4512

[Andrea B.]

What happened, Alex? You sold the MonoCam? It did not work out? For a while monochrome cameras were a big topic.

[Alex H]

I debayered a camera for a colleague, so I only had it for short period of time.