So I have been trying to learn to be as non destructive as possible... one problem Ive recently had is that when I try to merge layers, I will lose the mask that was on one of the layers. How do I prevent this? :/
I try to merge layers, I will lose the mask that was on one of the layers.
Instead of merging you could try converting the batch to a Smart Object. Bloats file size but the Layer Masks would still be available in the SO.
Trevor, why not to a channel? Isn't it lighter? (but a layer might be faster to get to, according on how you set up your workspace)
That's fair comment Pierre, and I have Channels and Paths panels permanently available. But can I create an Alpha mask without having to click on 'Save selection as Channel'? Is there an existing keyboard shortcut?
I'm a CTRL+J person also, but I'm now using more and more the right-click everywhere possible, and in this case save selection. (refine edge is there too, save as path)
I feel that the contextual menu are underused, and they are so fast to get to. (no need to lift the eyes from the image)
Pierre, I've just got to add a comment here. Since you mentioned right clicking, I have been trying it whenever I get a chance, and discovered a real wealth of useful shortcuts. Thanks for the heads up. You never stop learning with this app.
One small note: Every selection specifically is 8 bits/channel no matter what the document depth. I say this because selection masks are always provided to plug-ins in 8 bits/channel format.
So my point is that if you're editing deep (e.g., 16 bits/channel) image data, and whatever workflow you're using uses a selection as part of it, you will have quantized the data to 8 bits at some point.
You can prove this to yourself if you're skeptical... Make a radial pure grayscale gradient in a 16 bits/channel document. Then control-click on one of the color channels to make it a selection, and Save the selection in a channel. Now Copy the data from that channel and paste it over the original gradient as a new RGB layer. Set it to Difference blending and greatly enhance the brightness of the differences...
Noel Carboni wrote:
One small note: Every selection specifically is 8 bits/channel no matter what the document depth.
While "Selections" (the active dancing ants) are 8 bit, layer masks and channels are not limited to 8 bit when in a 16 bit image. The layer masks and channel retain their 16 bit editing. It's only when you make an active selection from a mask or channel that the selection is tunred into 8 bit. So, if you need to modify (feather or adjust) a selection in a 16 bit image, do so at the layer maks/channel stage before turning them into selections.
It's been argued (by Chris Cox) that actual selections really only need 8 bit precision...if you can prove that's not true, I'm sure the engineers would rethink that. I've never been able to prove it...
I just thought I'd mention it since some people do use the selection process as a stepping stone in their workflow to make things happen in their images, and the quantization might matter if the data is subsequently stressed a good bit. As an example, such stressing might occur, for example, when digging very dim data out of astroimages.
if you can prove that's not true, I'm sure the engineers would rethink that. I've never been able to prove it...
Nor have I ever had any problem with being able to select things at roughly 0.4% intervals.
But at this point I wouldn't bet that Adobe could be convinced to embark on changing it for any reason, even if one showed a way that it was turning their images into hamburger. The selection mask strikes me as being at the very heart and core of Photoshop - i.e., ingrained in the very oldest code - and stuff like that no doubt has accumulated a lot of implicit assumptions (e.g., byte addressing shortcuts) all throughout those 4 to 5 million lines of code you mentioned.