Lens contrast & post-processing

[thompsonkirk]

Advertisement (gone after registration)

Sean Reid has written a helpful new section on lens contrast for his website section on lens testing. The gist of it is that older/lower contrast lenses, because of greater flare, compress the dynamic range of digital files. He's said that before, but he explains it more thoroughly, managing to get hard stuff into non-technical language. (So much easier than reading Erwin Puts!)

 

He mentions post-processing in just a couple of paragraphs, but IMO the connection is worth keeping in mind because a lower contrast lens & its more compressed files mesh so well with Photoshop's tools.

 

1. You can always reset your black point & white point, so that a flatter or more compressed file is easy to turn into a full 'Zone System' spectrum from dark to light (which applies to the luminance range of color images, not just BW).

 

2. The Curves function allows you to bring out so much shadow detail from a more compressed file - as he mentions, both globally & locally. It's as if we have innumerable paper graqdes or endless resources for what was called split-filter printing in the days of the darkroom.

 

My point is that if/when you read Sean's text about lens contrast, think at the same time about PS Curves!

 

Kirk

[Marquinius]

Kirk,

 

It is really astounding what I'm learning through this forum. I think I'll have to prioritize, but there certainly is a lot to read and think about.

 

Good tip, about curves vs Sean's text!

 

Marco

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[thompsonkirk]

Hi, moderator -

 

I didn't mean to post this twice! This one is enough. Please delete the other one.

 

Thx,

K

[roguewave]

Nicely explained Kirk. It would be awesome to repost an example in both color & B&W to visually demonstrate this notion. It's integral to how I do much of my work. I can add some dialog as well, if you're game. Nice work & very timely topic.

[thompsonkirk]

I've been thinking about this, Ben, but don't know what to post because I've sold my newer lenses & kept pretty old ones - so I can't take 2 shots & compare them.

 

Would it be useful just to post a shot with an older lens in pretty strong light but with quite a bit of shadow detail?

 

K

[thompsonkirk]

OK, Ben & oithers,

 

Here's an effort to illustrate - but I don't know whether or not it'll show up on a monitor. It makes a big difference in a print, but not necessarily on a screen.

 

I've used an image that I posted recently, taken with a 28mm Summicron, which is generally thought to be less contrasty than a 28mm Elmarit. I'm sorry I don't have an Elmarit shot to compare - this is just an illustration re: improving shadow detail with a Curve, if/when there's some detail in the file. It doesn't prove anything about what Sean was saying, it just illustrates teasing out shadow detail.

 

First image: The RAW file was processed conventionally with ACR/LR. I made a BW conversion layer in PS & set the black & white points. This gives you what you see in the top image.

 

Second image: Here's a version with a mask that selects the darkest areas & applies an appropriate PS Curve.

 

Third image: A screen capture of what the Layers box looks like

 

Fourth image: The Mask: White areas indicate the Curve will be applied 100%; gray areas 50%

 

Fifth Image: The Curve itself, with some tweaking of the shadows (lower left)

 

Maybe these differences will be visible on your screen:

 

a. The doorway to the left should change from dead black all over in Image 1 (open doors inside of open doors!) to more gradation of tones toward the right;

b. The supports for the upper deck should be more visible - in Zone System terms they should move from III to IV.

c. The two wooden shutters should stop looking like holes into blackness & should look more like wood.

 

And if you can't see any of this, we've proved that the lens contrast issue & shadow recovery makes a lot less difference on the web than in a print.

 

I hope all 5 illustrations look OK!

 

Kirk

 

PS, I should have mentioned that I could have improved the shadow detail in RAW conversion by using the Fill slider in ACR/LR, or the Shadows slider in C1. But that would affect contrast in all the dark areas of the image, not just the ones that I wanted to help.

[tdavis]

Advertisement (gone after registration)

Only 3 came through, Kirk. I really appreciate your sharing of this exercise in subtle manipulation.

Thanks,

______

Tim

Opps, on a closer look, I now see 5 - you're right - not much of a change on the screen.

[roguewave]

Kirk, excellent presentation. My experiments demonstrate that there are various methods to achieve this effect. First, following Sean's article, I usually flatten my image to get more detail. The sharper the lens, the flatter I go. I love to have a chance to see EVERYTHING, before I make choices. Try experimenting with localized Brightness/Contrast on these same areas (doorway, supports & shutters).

 

If you have mastered the curves, it's hard to not use them, but I find I get better localized detail by starting flatter & building my own contrast & texture with layers, which bring out only the details I want to show & hide the others. Another trick is to go deep & bring in the heavy hardware available in the LAB space. The Linear Dodge (add) preset is like a light miner, it finds any light that exists in the DNG. Experiment & see what you think. Go to edges & work back to the middle until you get a sense of where the gradient & burnout is, then paint the light in & return to RGB.

 

This whole methodology rests on good exposure values from superb lenses. Post Processing isn't magic & it can't reveal what's not there. Far too often, I see great imagery that's been butchered with too much contrast, just to get the image "done".

 

Usually the OOF areas are the best diagnostic indicator of the health of the image. It's the rich array of details, the woven light patterns & shadow range that reveal if the image has the stuff. Getting those things right is what supports my foreground. It's alway good to have an interesting, if not arresting, foreground.

[thompsonkirk]

OK, let me see if I could follow that:

 

1. I made a duplicate, & changed mode to LAB;

2. Undeer channels, I selected the Lightness layer;

3. I added a Levels layer to the above & checked the Linear Dodge blending mode;

4. I Cmd[Ctrl]-Clicked to select Highlights;

5. Under Select I inverted;

6. I got the FIRST IMAGE shown below, which looks 'bleached,' then Flattened the image, & changed it back to RGB:

7. I Quick-Masked a selection of the shadow areas, & made a Layer Via Copy of this;

8. I moved that layer back onto the Original & got the SECOND IMAGE shown below - which can be compared to the 2 versions in the previous thread.

 

Whew. Or that's how I think I did it.

 

You mentioned painting from the duplicate, rather than making a mask & a Layer Via Copy, but mainly that's a difference of habits? I guess if you paint, you can see a build-up of shadow detail as you go along?

 

Am I (or others) catching on?

 

Kirk

[roguewave]

Kirk, that's one way, but not MY way. First, you need to find the amount of detail you want to retrieve from the shadows/dark areas. If you go to LAB, try to use single digits in the presets, that will prevent burning out the whole image. I don't channel surf here, it doesn't help at all. By entering into LAB you get to sample the different ways to separate Luminescence from contrast and there's the kernel of retrieving the lost details. My images don't look like those. You should be able to get MORE DETAIL everywhere, not less. Good Luck hunting.