some explanation please

[Herbert123]

123,

Thanks for the explanation.

FWIW, I had no problem using 16-bit TIF as an interchange format between PL and CorelPhotoPaint (part of my CorelDraw Graphics Suite X5, which is a couple of versions old by now). When the image layer came back into PL, I zoomed in to the soft edges of the erased areas and did not notice any black or white artifacts.

Corel considers PhotoPaint their "Professional" image editor and it has many features/capabilities not available in PSP....

It really depends on the application how it treats the alpha channel of a TIFF file. As I said before, in compositors such as After Effects, Nuke, Fusion, Blender, and so on, it is left up to the user to decide how to handle pre-multiplication of the involved matte(s).

The trouble is that applications like photoshop and PSP take that decision out of the hands of the user.

[brucet]

Just another head scratcher to add to the mix. Simple example. When I create a file I first of all duplicate the layer. Then create a new layer. (All transparency layers). The new layer goes between the other two layers. I flood fill it with pink. Pink allows me to better see what I have to erase from the top layer. If I'm cutting out a car I end up with just the car and transparent sections showing the pink layer below. (A screen capture would be nice but I talking two different computers and it's not so easy to do). If I then send the top layer, just the car, to PSP as a 16bit tiff to do a few things and send it back again that's where the issues arise. Now if I go to my pink layer and flood fill with another colour, say white, it will seem ok. But if I zoom in to 100% or more I can see various artifacts and colour variations in the white. These are not actually in the while layer. These artifacts are in the transparent sections of the top layer. If I have only used a low opacity on the car windows then all heck can break loose.
So it seems the issues are arising from PSP's handling of the transparent sections.

Scratch, scratch!!

Last night I had one quick try using PNG's. So far that seems to work.

regards

[photoken]

OK, I don't think duplicating the image layer is the best way to work, but since it's your preferred method of working, here's what I suggest:

Instead of erasing the image areas where you want the bottom layer to show through, leverage the work you did to create the lasso by using that lasso to create a layer mask for the top image:

atlantis landing layers.jpg

By changing the Brightness value in the Hue/Saturation layer, you can easily tweak how much of the bottom layer will "show through". The Gaussian Blur layer will soften the edges of the layer mask.

The advantage for your preferred method of working is that whether you send the top image layer or the bottom image layer to PSP, what you will be sending will be a simple, opaque image, which PSP should be able to handle. No worries about transparency, no worries about alpha channels, no worries about soft-edged erased areas, no worries about missed spots in the erasure. When the edited image(s) come back from PSP, the PL layer masks will automatically be applied for the "show through" effect:

Atlantis landing.jpg

[brucet]

Ken the example I gave was a very simplistic one. I work this way after years of work flow experimentation. Or pigheadedness
The duplicate often comes into play when I want to 'borrow' a selection from the original to put back into another layer.

regards

[photoken]

The duplicate often comes into play when I want to 'borrow' a selection from the original to put back into another layer.

That's cool. The advantages of not having a transparent "hole" in the images sent back and forth between programs still apply, as well as not having to worry about "undoing" the erasing.