Contrast masking and Gaussian blur

(Don) asked

I’m currently busy contrast masking in order to boost the (very!) limited dynamic range of my film scanner. This involves scanning the same slide twice, once for highlights and once for shadows. The two scans are then stored in layers and the bottom layer is used as the layer mask for the top layer, resulting in an image composed of the best parts of both.

During this procedure a certain amount of Gaussian Blur should be applied to the layer mask.

Now then, there are basically two distinct cases here: one, when an image mostly comes through (highlights scan supplies the highlights, shadows scan supplies the shadows), and two, the "bit in the middle" where both layers contribute.

The problem is, case one (an image comes through) wants minimum, read *no*, Gaussian blur while case two (the mix) wants *lots* of Gaussian blur.

Choosing the "happy medium" of Gaussian blur just makes them both unhappy… Case one develops "auras" while case two becomes washed out.

I could edit the layer mask manually but that’s far too time consuming and finicky, and I’m trying to streamline and automate the procedure as much as possible.

Any ideas on how to do that?

Check out the Optipix plugins from ReindeerGraphics.com, which includes a "blend exposure" routine intended for exactly what you are trying to do. It uses a more advanced algorithm that automatically does the blending, does not use a manually derived mask, and does not use any Gaussian blur.

Don wrote:

I’m currently busy contrast masking in order to boost the (very!) limited dynamic range of my film scanner. This involves scanning the same slide twice, once for highlights and once for shadows. The two scans are then stored in layers and the bottom layer is used as the layer mask for the top layer, resulting in an image composed of the best parts of both.

During this procedure a certain amount of Gaussian Blur should be applied to the layer mask.

Now then, there are basically two distinct cases here: one, when an image mostly comes through (highlights scan supplies the highlights, shadows scan supplies the shadows), and two, the "bit in the middle" where both layers contribute.

The problem is, case one (an image comes through) wants minimum, read *no*, Gaussian blur while case two (the mix) wants *lots* of Gaussian blur.

Choosing the "happy medium" of Gaussian blur just makes them both unhappy… Case one develops "auras" while case two becomes washed out.

I could edit the layer mask manually but that’s far too time consuming and finicky, and I’m trying to streamline and automate the procedure as much as possible.

Any ideas on how to do that?

Don.

I use "The Claw". This is the Ctrl-Alt-~ command. That will progressive select the highlights. I select the highlights on the bright layer and delete. Then I select the highlights on the dark layer and invert the selection and delete the darkest on the dark layer.

This doesn’t always leave the mid-tone quite right. So, after selecting with "The Claw" use Levels on the Quick Mask to adjust how much of the highlights you want to keep or not.

Then again, I usually shoot three shots. That’s because me camera has that nice function. So, it’s a tad easier with a nice mid range shot that I layer the dark and light over it. Then I play with the opacity to get what I want.

It’s simple and quick.

Clyde

On 10 Apr 2004 11:25:09 GMT, (DrJohnRuss) wrote:

(Don) asked

I’m currently busy contrast masking in order to boost the (very!) limited dynamic range of my film scanner. This involves scanning the same slide twice, once for highlights and once for shadows. The two scans are then stored in layers and the bottom layer is used as the layer mask for the top layer, resulting in an image composed of the best parts of both.

….

Check out the Optipix plugins from ReindeerGraphics.com, which includes a "blend exposure" routine intended for exactly what you are trying to do. It uses a more advanced algorithm that automatically does the blending, does not use a manually derived mask, and does not use any Gaussian blur.

Thanks very much, I’ll have a look.

Don.

On Sat, 10 Apr 2004 14:17:51 GMT, Clyde
wrote:

I’m currently busy contrast masking in order to boost the (very!) limited dynamic range of my film scanner. This involves scanning the same slide twice, once for highlights and once for shadows. The two scans are then stored in layers and the bottom layer is used as the layer mask for the top layer, resulting in an image composed of the best parts of both.

I use "The Claw". This is the Ctrl-Alt-~ command. That will progressive select the highlights. I select the highlights on the bright layer and delete. Then I select the highlights on the dark layer and invert the selection and delete the darkest on the dark layer.

I never heard of Ctrl-Alt-~! Can’t find the shortcut in the docs either. It appears to do a threshold and then selects the highlights, with some feathering and/or opacity thrown in – all in one big swoop?

So, what is this Ctrl-Alt-~ magic exactly? How do you get at it from the menus?

This doesn’t always leave the mid-tone quite right. So, after selecting with "The Claw" use Levels on the Quick Mask to adjust how much of the highlights you want to keep or not.

I "invented" this whole thing by using threshold but, as you say, this leaves the midrange iffy. As well as the transition areas! The main problem was the color mismatch as my highlights and shadows scans are quite wide apart (2 to 3 Analog Gain clicks).

I later learned not only has this already been invented but it’s called "digital contrast masking" and it’s much more elegant than my crude attempts. That appeared to cure all the problems until I came across some real hard cases as described earlier…

Then again, I usually shoot three shots. That’s because me camera has that nice function. So, it’s a tad easier with a nice mid range shot that I layer the dark and light over it. Then I play with the opacity to get what I want.

My camera has framing too (although I haven’t used it yet) but right now I’m scanning slides. In theory, I could do three scans and have indeed been playing with the idea of doing just that. It’s a bit of a hit-and-miss, though, because the scans don’t always match so I have to keep scanning until they do.

Anyway, thanks for the tips!

Don.