Optimize Images - Best Workflow?

As RodWard said re-sizing to the correct size in Photoshop (or equivalent) and importing will save on size. I generally mock out images or placeholders when story boarding so I have an idea of the sizes that I need. If you hover over an object or image in captivate it will tell you what size your object is in pixels (just a quick tip), and then you can replicate in Photoshop.

If you already have a project built, but want to swap out images. I would save your new image over the existing one (so it has the same name), and then use the update setting in the library to update it (note, that you may need to press the reset to original size button if you have made your new image a different size). This way you do not break any of your design in a responsive project.

The advantage of using Jpeg is that you can compress the image to reduce file size, but as Rod has mentioned there are instances when you should use png.

One thing to note, in responsive project CP outputs all images as pngs which cancels out some of your efforts to compress Jpeg.

You need to preformat your images so that they are the exact size you need them to be on the Captivate stage.  For most images JPG will be your most economical format, but for some types of images where the subject is a computer interface screen capture and the UI has lots of flat colour with text, then your better format will be either GIF or PNG.