Sizing the graphics to precisely as large as they need to be to print at the scale and resolution required is the only good way to keep your PDF file size down. Scrivener is just going to assume you know what you want, and so is the macOS PDF generator.
Personally I’d say your PDF could do with some optimisation. Do you own Adobe Acrobat Pro? If so, Distiller has tools for compressing images further, and of course Acrobat itself has plenty of tools for optimising graphics.
As a form of reference on where I’m coming from in saying that: the Scrivener user manual has close to 300 figures in the PDF. Like your case, many of them are quite small, only showing a cropped area of the interface—but the source files used to generate them do clock in at around 200mb of PNG 24-bit losslessly compressed graphics. So on the surface I’d say we’re both working with a similarly sized data set.
However, if you open the user manual PDF and examine its size, you’ll see the distribution copy is only around 16mb. Prior to typesetting the PDF, I run a batch conversion from PNG to mid-high quality JPEG on the source folder. The PDF is instructed to use JPG instead of PNG graphics.
The specifics of how I do that probably aren’t applicable to how you work. I use LaTeX to typeset the PDF, which refers to images stored on the disk and thus it is a simple matter to have “.png” changed to “.jpg” automatically when compiling. The closest equivelent to that is using linked images in Scrivener instead of embedding them—something I would highly recommend to anyone working with a lot of graphics. Keeping your files on the disk means being able to do stuff like batch compression, keeping separate sets for different purposes and as well keeping your draft folder clean of large embedded files. For example you could:
- Duplicate your graphics folder and name it so you know it’s the source material.
- Batch compress/optimise the original folder.
- Reload the project and compile with the correctly sized and compressed images.
And you’re probably done. Need black and white? Similar solution. Need ebook low-res size? Same. You can keep however many sets of folders you want, and rotate them out by name before loading the project.
For now though, if the files are all embedded, the only way to fix them is to right-click on each image one by one, save it as an image, fix it and replace it. Consider however that in doing so you will need to be placing each image in Scrivener again manually—so it would be a good opportunity to convert to a linked image workflow (§15.7.4, Linked Images, in the user manual).
You do describe having dragged the image from the binder—under default settings the above still applies as that merely embeds the graphic as a new copy. However do check Behaviors: Dragging & Dropping, Link to images dragged from binder into editor. If that has been enabled this entire time, then you can edit the images from the Binder. Not quite as simple as having them on the disk—for purposes of batch processing—but a lot easier than the one-by-one fixit approach!
If you’re concerned there may be a mix, that’s no problem, you’d use the Draft material as a starting point. When double-clicking on an image it will display different functions depending on whether it is embedded, linked to the binder or linked to the disk. Both of the latter conditions have a button that brings you to the original file where it can be easily edited.