Masking with tone(p. 88)?

[dpnew]

I'm having trouble understanding the meaning of what's going on in the steps for setting up the blue midtones of an image for adjustment. I just had a bit of a revelation when I was examining the names of the layers in figure 3.33, and now I realize that the steps separate the blue channel into two layers: the midtones and the highlights/shadows.

So, now I guess the part I don't understand boils down to how "Group with" works. My understanding is that when you group one layer with another, only where there is content in the lower layer will content in the upper layer be visible. That seems to fit what is happening for the Blue Midtones group of layers on p.88, but when I apply that concept to the Blue Highlight/Shadows group, I can't make sense of it. It seems to me that since there is content everywhere in the lower layer-- the Blue Highlight/Shadows layer--all the black pixels in the Blue Mask layer should be visible.

But then, why is the Multiply mode necessary for the Blue Mask layer? I understand Multiply will turn every midtone color in the lower layer to black because (0,0,0) multiplied with any other pixel color is (0,0,0), but why doesn't "Group with" and the Normal blend mode accomplish the same thing? I know it doesn't from experimentation, but I don't know why, and that's why I guess I don't understand "Group with".

Oh, and one more thing. I find that when I use Curves to create the masks, I can't map, say, 74% input to 0% output while simultaneously mapping 75% input to 100% output: Curves will only let me use the input points 73% and 75% or 74% and 76% but not 74% and 75%. I've played around with curves a bit to make several masks based on tone, and once in about 5 tries I'm able to use consecutive input points, but usually it won't work.

Thanks

[Richard_Lynch]

I'm having trouble understanding the meaning of what's going on in the steps for setting up the blue midtones of an image for adjustment. I just had a bit of a revelation when I was examining the names of the layers in figure 3.33, and now I realize that the steps separate the blue channel into two layers: the midtones and the highlights/shadows.

Right.

So, now I guess the part I don't understand boils down to how "Group with" works. My understanding is that when you group one layer with another, only where there is content in the lower layer will content in the upper layer be visible. That seems to fit what is happening for the Blue Midtones group of layers on p.88, but when I apply that concept to the Blue Highlight/Shadows group, I can't make sense of it. It seems to me that since there is content everywhere in the lower layer-- the Blue Highlight/Shadows layer--all the black pixels in the Blue Mask layer should be visible.

They are...but because of modes and layering you just can't see them as black. What you should see (with the Blue Midtones layer off) is the lack of blue (or yellowing).

But then, why is the Multiply mode necessary for the Blue Mask layer? I understand Multiply will turn every midtone color in the lower layer to black because (0,0,0) multiplied with any other pixel color is (0,0,0), but why doesn't "Group with" and the Normal blend mode accomplish the same thing? I know it doesn't from experimentation, but I don't know why, and that's why I guess I don't understand "Group with".

You are grouping white above. If you multiply white with anything, it is the 'anything'. Quite honestly, the whole addition of the Layer 1 was not necessary. I am sure it was a late addition (due to someone's objections) and just complicates the matter. If you just name the Layer 1 layer Blue Mask and skip the fill and merge, the whole thing works the same way (and that ends up being what is pictured).

Oh, and one more thing. I find that when I use Curves to create the masks, I can't map, say, 74% input to 0% output while simultaneously mapping 75% input to 100% output: Curves will only let me use the input points 73% and 75% or 74% and 76% but not 74% and 75%. I've played around with curves a bit to make several masks based on tone, and once in about 5 tries I'm able to use consecutive input points, but usually it won't work.

the inconsistentcy in the behavior is probably why there is something about it in the errata and both seem to be right -- on occassion.

Ok, I'm working my way through this: Normal won't work because even though the layer is called Blue Mask, it really isn't a true mask since it was merged with a white layer, so Normal would cause the Blue Mask layer to completely cover the layer beneath it.

Yes, without the mode change the grouped layer would cover the base of the clipping group entirely. About it being a mask: I think it is. Masking takes on many different forms, and is an odd case where different names are used for the same thing (most people like to go the other way and blur terms, like PPI and DPI for example). part of the concept behind playing the little game with the Layer 1 layer and the white fill would be to make the layer LOOK like a mask (i.e., like the layer masks are traditionally with Photoshop and as they are in Elements if you use the layer mask free tool). To me, and Websters, a mask is something that covers something else up. Whether the mask is hiding the solid part, the white part the transparent part or whatever, it is forming a mask...and mask it is, accurately.

Some errata: figure 3.33(p.88) shows the Blue Mask layer with a transparency checkerboard for the background, but since that layer was merged with a white layer in step 6(p.88), it will not have any transparency, so the picture is incorrect--that layer should have white in it wherever the mask isn't present, not transparency.

If you do as I suggest above and ignore the filled layer, the result comes out exactly as pictured and produces exactly the same results either way. Granted, the image could better match the text--but either method works. I probably updated one and forgot to update the other (that is, I fixed the image and not the technique). In this case i'd actually rather change back the text than go for the easy fix and change the image.

I hope all this has helped clarify. To review: Yes, there is a mismatch between the image and the text, and while it might be fixed, it is minor. Grouping works based on solidity, but mode also plays a factor. using this technique, you can effectively and selectively mask color in any combination.

[engineerbob]

When I do this excercise, I have two images (i.e., windows) with the same Background. For the 1st I use Blend Mask to create the midtones mask. For the 2nd I separate the RGB channels with preview. But when I copy the mask from the 1st image to the 2nd image, the mask does not align properly. Even though I use Select All, Copy, Paste, the copied portion is only big enough to include the "painted" pixels, not the whole layer image. So after the paste, this smaller portion is centered in the layer making it misaligned with the other layers. I can of course realign the mask image using the arrow keys. But what am I missing to get the full image space copied? (I have copied the mask layer, leaving behind the Background copy above it, as well as merging the Background copy with it before copying. It doesn't seem to make any difference.)

Also, if I follow the book's directions, the Blue Highlight/Shadow group is just above the Blue channel group. In the book, it is shown at the top of the layer stack. I think it gives a nicer looking result when the Blue Highlight/Shadow group is just above the Blue channel grouping than when it is at the top of the layer stack. It is less contrating with smoother transitions. What do you thinnk?