Can anyone please explain Smart Object file sizes?

[critter_monster]

Hi, one of my students had a question I promised I would find an answer for.

What are the "ground rules" for how a Smart Object affecs document size?

In this specific case the student wanted to know if a given .psd file, in which he is using a Smart Object with some Layer Styles, will be bigger or smaller if the Styles are applied within the SO itself vs. applied to the SO layer in the .psd file.

I did a little experimentation and found quite radical differences in file size, but they depended a lot on which Styles I used. Sometimes version 1 was a larger file size, sometimes version 2. So rather than try and discover every possible combination for myself, I am asking here: just what is file size a function of, when working with Smart Objects? Thanks !

[c.pfaffenbichler]

I think you don’t provide enough information to assess the actual situation. 

Please post meaningful screenshots including the pertinent Panels. 

• The Smart Object has its own size; this adds to the size of the containing document (edit: unless in the case of Linked Smart Objects naturally). So for example a flattened SO has much less effect than a SO with additional Layers or no Background Layer, because the preview has to be included. 

• Each instance of the SO in the containing document is a Layer and as such adds to the containing file’s size. That includes the effect applied transformations have. 

[critter_monster]

Yes, the question itself is vague and I don't have any specific file to show for it. The student wanted to know the difference in file size if he had a Smart Object in a .psD document, opened the Smart Object (as .psb) and then saved it with layer styles applied to it in the .psb --- versus having the same Smart Object, in the same .psd document, and simply applying those same layer styles to the Smart Object layer itself (not opening the Smart Object as a .psb).

I said I didn't know, I'd find out! But I don't have his specific files so all I could do was experiment, and ask this forum.

My experiment yielded a lot of different possible results, depending on what the contents of the document were... and to compound the problem we both have Version 21.1.2, which as I understand from this forum has some fixed-in-V22 issues with cropping the canvas of the Smart Object .psb. Good times.

Thank you for helping me, if you have any more ideas I'd love to hear them but I can also just keep my ear to the ground here and see what I can learn about the behavior of Smart Objects and their file sizes.

[c.pfaffenbichler]

An example: An RGB 8bit image 2048px x 2048px. 

• »flatNoSo.psd« is juat a flat, noisy image, so it does not compress well. 

• »flatSoWithLayer.psd«: The image is a plain Layer in a Smart Object. Roughly four times larger than the flattened image – the image in the SO, the preview of the SO, the instance, the preview of the containing document. 

• »flatSoFlat.psd«: The image is a flattened Smart Object. Roughly three times larger than the flattened image – the image in the SO, the instance, the preview of the containing document. 

• »flatSoFlat50Perc.psd«: The instance of the flattened SO is scaled 50%. Less than twice the size of the flat image – the image in the SO, the smaller instance, the preview of the containing document. 

• »flat2SoFlat50Perc.psd«: 2 instances of the flattened SO are scaled 50%. Roughly twice the size of the flat image – the image in the SO, two times the smaller instance, the preview of the containing document.  

[c.pfaffenbichler]

Another issue: A (non-Vector) Smart Object, when created on Photoshop by converting Layers to a Smart Object, is a psb. 

However one can embed other file formats like tif or jpg by placing them (File > Place Embedded). 

And lossy compression like jpg can decrease filesize for some images much stronger than non-lossy compression. 

But the instance/s of such a SO are not affected by that, they are compressed according to the containing document’s compression. 

[c.pfaffenbichler]

Well, what about the Smart Object instance/s?

Are there more than one?

Have they been transformed? 

What are the Layer Styles in question (in particular: do they incorporate patterns in any way)? 

[critter_monster]

So far, the example my student had was just one Smart Object instance. I'm not sure if he transformed it at all, and as far as I could see (we were on Zoom) he didn't use any patterns in the Layer Styles applied to the Smart Object.

[c.pfaffenbichler]

And what are the image’s and the Smart Object’s bit mode, bit depth, pixel dimensions and the resulting sizes?