1. if you activate it network-wide, then it must be configured network-wide. Otherwise, you can activate it on specific blogs, and configure each blog differently.
2. it does NO resizing, use something like Imsanity if you want to cap uploads to a certain dimension. It does optimize on upload, and allow you to bulk optimize all previously uploaded images.
3. The plugin will tell you if anything is missing.
4. it is self-contained, although there is a ‘cloud’ service like smush.it that you can pay for to offload the processing if you like (or need to). The EWWW IO Cloud does NOT store images EVER, they are deleted immediately after processing, which is different than smush.it
5. not really
6. right, those are just galleries that the plugin supports
7. the images are compressed on upload, and never modified again (it would slow things down too much to optimize on-demand every time, yikes). If you change a setting later and want to re-optimize, there is a checkbox to force the plugin to re-optimize images. It DOES integrate into the wp_image_editor class, so any new resizes or editing of the image will result in an optimized image.
Lastly, you should ensure you have a backup of your images before doing any bulk operations. While the plugin is well tested (after 275,000 downloads), there is always the possibility that your server has some unique configuration or some sort of images that could still behave in unexpected ways.
Ok, so the way it optimized is by removing the unnecessary junk right?
That is just one portion of optimization, if you are referring to the stripping of metadata, and is actually disabled by default because some people need to keep their metadata.
Each image type is optimized differently, but by default, no pixels are altered. There is an optional lossy mode for PNGs, but I recommend backing up your images before you attempt it in case you are unhappy with the results. There are also options to attempt conversion to different image formats if you think particular images might be more efficient in a different format. So here is a bit more technical version of what happens during normal optimization:
JPG – images are re-encoded using Huffman encoding, and also tested to see if a progressive or non-progressive image will yield the best filesize.
PNG – this one is tricky, because there are so many variables that go into PNG compression. Essentially what the tools do is look for some basic palette reduction removing unused colors, and a couple other techniques to re-arrange the pixels more efficiently. Then, they perform tests to find the best mix of compression variables. If you want to understand more about that, there is lots of reading material on the pngout and optipng homepages.
GIF – mostly, it only helps animated GIFs, and does so by removing pixels that are repeated between animation frames. If you have non-animated GIFs, it is often more efficient to convert them to PNGs. The PNG format is almost always more efficient, and of course the plugin will only save the converted PNG if it is actually smaller.
In all of the optimization routines, the optimized version will overwrite the original (only if it is smaller though). However, in the conversion routines, the originals are preserved by default, and the converted versions have a numeral appended to the filename.
