Action: Selenium Toning

[photoken]

A while ago, Alan and I were discussing how to create a "selenium" colour tone on a B&W image:
http://www.pl32.com/forum3/viewtopic.ph ... %3A#p38586
and I've finally got around to recording an action for this.

To use the action, first use your favorite method of converting the colour image to B&W. Then, highlight the original image layer:

STaction layers 01.png

and run the action.

You'll get two primary adjustment layers, each having a layer mask with a child Curves adjustment layer for the layer mask.

This primary Curves adjustment layer:

STaction layers 02.png

sets the colour of the tone. I use a very subtle colour adjustment in Lab mode, so this is the adjustment layer you modify to change the hue and strength of the colour toning.

The layer mask for this Curves adjustment layer serves to target the colorization to the dark values of the image, and not affect the whites and brightest values. You'll use its child Curves adjustment layer to precisely adjust the areas that receive the toning.

The primary False Color adjustment layer at the top of the layer stack adds some black to the darkest areas of the image because applying a colour tone to an image often leaves the darkest areas looking somewhat weak. Its child layers work as described above to allow precisely targeting the values getting the darkening effect.

Here's the action:

Selenium Toning.zip

Enjoy!

[photoken]

In the last week-and-a-half, I've enhanced this Selenium Toning action by doing some experimenting with B&W conversions.

I've begun using a Curves adjustment layer using the Lab colour space, setting the "a" and "b" channels to flat lines at y=50.0%. This is a terrific way to convert to B&W because it lets me then alter the curve of the "L" channel (as well as the Contrast, Brightness, and Gamma settings) to adjust the contrast of the B&W image. I really, really like this method and I created a preset to do the "a" and "b" channel work.

So, I've modified the Selenium Toning action to include this method of doing a B&W conversion. The zip file package includes two files:

• B&W conversion Lab.clu (this is the preset for the Curves layer to set the "a" and "b" channels.)
• Selenium Toning 05.actions (this is the new selenium toning action which uses that Curves preset.)

To use this new action, you first need to copy the .CLU file to your Presets directory. Then open a Curves adjustment and merge the .CLU preset into your presets. Add and merge the .ACTIONS file into your Actions as usual.

I've also enhanced the layer names so they make more sense:

st 05 action layers.png

When you run the action, it will pause after the creation of the B&W conversion Curves layer to allow you to modify the "L" channel to adjust the image contrast.

If you'd rather not use that Curves method for the B&W conversion, simply disable the first 4 steps of the action:

st action disabled.png

I've tested this action on a wide variety of images:

• Nature scenes with a "normal" range of values, such as pictures taken in sunlight.
• Nature scenes with a very limited range of values, such as foggy scenes, etc.
• Portraits, both perfectly lit studio works as well as casual snapshots.
• Cityscapes, from midday sun to twilight.
• High-key and low-key images.

and it's never failed to produce an excellent result!

Here's the zip package with those two PL files:

Selenium Toning 05 pkg.zip

Enjoy!

[bkh]

To use this new action, you first need to copy the .CLU file to your Presets directory. Then open a Curves adjustment and merge the .CLU preset into your presets. Add and merge the .ACTIONS file into your Actions as usual.

Just a quick remark: you don't need the Curves preset for the action to work – the action contains the curves and does not read the preset.

Btw., the action needs to be run in 16 bit mode, otherwise it produces a lot of banding in the dark shadows.

Cheers

Burkhard.

[photoken]

Just a quick remark: you don't need the Curves preset for the action to work – the action contains the curves and does not read the preset.

Interesting. Thanks for that. The B&W Conversion preset is a bonus, then.

Btw., the action needs to be run in 16 bit mode, otherwise it produces a lot of banding in the dark shadows.

? I haven't noticed banding when using the action on 8-bit JPGs.

[bkh]

? I haven't noticed banding when using the action on 8-bit JPGs.

Here's a small sample file (a crop of some leaves in the background):

leaves.jpg

With 8 bit processing (leaving the curves adjustment as it was), I get

leaves selenium 8 bit.jpg

while everything is ok with 16 bit processing:

leaves selenium 16 bit.jpg

Cheers

Burkhard.

[photoken]

I'm not noticing any "banding" in the example you posted.

I still don't see any "banding" here in a JPG enlarged to 800%:

st 800.png

[bkh]

I'm not noticing any "banding" in the example you posted.

I still don't see any "banding" here in a JPG enlarged to 800%:
st 800.png

It's usually hard to see banding at such a high magnification. However, it looks as if a lot of shadow detail is lost in your image. You should see the difference if you switch to 16 bit processing (by changing the bg image to 16 bits in image mode).

Cheers

Burkhard.

[photoken]

However, it looks as if a lot of shadow detail is lost in your image.

Not really. That area of the image is very dark in the original, too:

st orig 800.png

I'd expect to see "banding" across large smooth areas of subtle tones (such as the sky), but no "banding" is apparent....

[bkh]

However, it looks as if a lot of shadow detail is lost in your image.

Not really. That area of the image is very dark in the original, too:
st orig 800.png

I do see the same loss of detail in the darks as in my sample image. E.g., the horizontal branches disappear when processing in 8 bit. (This isn't a problem with your action, but rather a general rule: 8 bit isn't enough if you work in Lab.)

If you can't see the differences, then maybe your monitor isn't properly calibrated?

Anyway, here's the sample images, but with a histogram correction layer applied, to emphasise the differences. Left half: 16 bit, right half: 8 bit

leaves 8 vs 16 bit.jpg

selenium toning 8 vs 16 bit.jpg

Cheers

Burkhard.

[photoken]

(This isn't a problem with your action, but rather a general rule: 8 bit isn't enough if you work in Lab.)

The good news is that the action does give the control that I want over the conversion and toning, whether it's in 16-bit or 8-bit.

As a side note, for those who want to use another/additional technique to convert to B&W (such as using the Gray Mixer or the Hue Editor), applying the action with the B&W conversion steps works great. You'll still have the nice ability to adjust the contrast of the B&W image using the "L" channel.

[Herbert123]

I can confirm the banding in 8bit.

Ken, I believe you work on a laptop screen? Most laptop screens are technically quite limited compared to good stand-alone desktop screens. Do you use a Spyder or Huey (or something similar) to hardware calibrate? I calibrate my three screens with a Spyder Pro. Best investment I ever made.

[photoken]

Ken, I believe you work on a laptop screen?

Yes, that's correct.

So, the bottom line is to either test 8-bit for one's devices and see whether the "banding" is noticeable or work in 16-bit from the beginning and save the work in a notJPG format....

[photoken]

Do you use a Spyder or Huey (or something similar) to hardware calibrate?

No. The cost of those hardware calibration tools is too prohibitive for me.

But, you got me thinking about this, so out of curiosity I went through the Lagom LCD Test suite on the Web. The result is that my monitor "passed" all but one of the tests. In other words, I can see all the distinct 32 colour test bands in all the colour channels; I can see all the dozens of test squares (from almost black #1 to the whitest); I can see all the resolution test lines; and all the middle gray values are spot on. The only thing missing is distinguishing between the two very brightest "checkerboard" white intensity test images. Oh, and in the gamma test, my "gamma blending" occurs at less than their recommended value of 2.2.

I'm satisfied.

[bkh]

So, the bottom line is to either test 8-bit for one's devices and see whether the "banding" is noticeable or work in 16-bit from the beginning and save the work in a notJPG format....

You can start with an 8 bit image and save the final result in 8 bit JPEG, no problem (all the examples above are 8 bit JPEG). You only need 16 bits for the processing steps (mainly, RGB <-> Lab conversions).

Burkhard.

[photoken]

You can start with an 8 bit image and save the final result in 8 bit JPEG, no problem (all the examples above are 8 bit JPEG). You only need 16 bits for the processing steps (mainly, RGB <-> Lab conversions).

That's interesting advice and should be considered by anyone wanting to do this type of processing. I can appreciate the value of providing 16-bit "headroom" for the processing steps, but I think there will always be some quality issues when a JPG conversion is in there somewhere.

I've always been a proponent of maintaining a fully 16-bit workflow, so I'll add my two cents: Start with a RAW image from your camera, do all the image editing on the 16-bit image, and save your work in a lossless 16-bit image format. Avoid converting to JPG unless as a last step when necessary for scenarios such as posting on the Web.