Some of my post-processed images contain identical layers, the only differences being their blend modes, blend opacities, and the possible adjustment layers targeted to them. This got me thinking if I could save some disk space using the "smart objects"-feature. (since I always save my recent files with full layer information without flattening the image, for possible future uses)
If I've understood correctly: when you create a smart object out of a normal layer and duplicate it, the duplicates aren't independent since they change together with each edit made to one of them. (excluding layer blending options and adjustment layers, of course) Apparently they can also be transformed losslessly multiple times, but let's not get into that for now.
So if I create smart object out of a layer and then duplicate it, say, 5 times, the memory usage should be only 1/5 compared to duplicating regular non-smart layers - is this correct? After all, all five change together when edited, which would indicate that there's only one copy in the memory and the other four are instances of that copy.
When such file is saved on disk though, the file size doesn't decrease - instead, converting identical layers into instanced smart objects increases the file size.
I was hoping to shave off about 100-150 MB of my 600 MB (still editable) image files, but apparently that doesn't seem to be possible with the "smart objects"-feature. Is that correct or is there something I should still try?
Another thing that crossed my mind is that perhaps the LZW / layer compression options of TIF-files are already capable of optimizing the file size by finding layer similarities and attempting to merge them into such instances, in which case my file was already optimized for what I was attempting to do manually. But I could be wrong, since I don't know exactly what the compression options are capable of. I'll do some reading about those meanwhile.
Hmm, any thoughts?