Saturday, 14 May 2016

Images In Books - Quality Versus Compression

I'll continue with this short series of posts about creating books. Yesterday I looked at running heads. Today I'll delve into the world of using images in books. This follows a recent question I was asked by another author about the best way to work with illustrations, so is a truncated form of the advice I gave to her.

If you are going to include images in a book then you always start off with the best quality images you can. High resolution, usually at 300 DPI (or, more accurately, 300 PPI). This also means they will be larger files, taking up more disk space. You can downgrade images (shrink them, or lower their quality) later if required; but if you start with a low-quality image you can never put extra detail back in. So start with the best.

That will be great for print, where you need high quality and the file sizes don't really matter. However, you probably need to lower the quality and compress the images for e-books. My print version of They Move Below is around 70mb with a few illustrations; for the e-book I compressed it to 7mb by dropping images to 220dpi and it still looked good. Since Amazon charge for e-book distribution based on file size, if I'd left it at 70mb I'd probably have had to pay THEM for every Kindle copy sold!

"How many illustrations are okay for an e-book? At what size?"

It depends on a lot of factors. For example, a small line drawing with no shading will be smaller (more compressible due to lots of areas of the same colour) than an image where every pixel is different from the next (e.g. a very busy photo). The image size in pixels, the format (png, jpg etc), the level of compression and so on are all relevant.

I always start with the print book and include images at maximum quality, such as uncompressed png files. (NB, if using Word, you have to turn off automatic compression in the advanced settings - see the screenshot halfway down this post). When the book is finished I save that print version, then copy it multiple times, adding the suffixes 96, 150, 220, and default. For the ones with numbers I open the file, right click on an image, choose compress, set the DPI at the appropriate number, tick to apply it to all images, then save the file. Do it on each version. You then have a file with the image DPI listed, and can compare the file size and quality for each, and decide which one to use as the basis for your e-book. You want to trade off a teeny bit of quality (ideally hardly noticeable to the naked eye) for a big filesize saving.

This is an example I did recently. As you can see, compressing to 220 DPI made a massive difference to filesize. Since the quality was good and further savings negligible, I used that as the basis for my e-book version.

This was how I set it up to compare the quality of each version. (This image is now compressed so is just to give you an idea of how to compare - when you do it live on your screen you can then make an informed judgement.) 220DPI was almost as good as the original. 96DPI was quite blurry. I should add that this isn't an illustrated book, but in a short story about social media I felt that I needed fake chat and fake Windows error messages as images to look more real.

The ideal is to minimise work in these processes. It's why I write fiction in a specially-formatted document and use styles to control everything; it is already set up for correct formatting, such as new chapters beginning on odd-numbered pages, numbering beginning on the first page of the novel (not the frontmatter), new chapters and sections having a no-indent paragraph as the first one and so on. I end up with a master doc that can then be copied so I have a print and e-book version, and each only needs a few tweaks. Software strips out the page numbers, running heads etc when converting to epub, so I don't need to do anything with that; for the print I already have the running heads, high DPI images and so on. The main tweaks with print are to add any extra fonts and spacing.

The other principal use of images is on the cover. Print covers in particular are a challenge, since the number of pages in each book varies, which alters spine width: and that is the central part of the output PDF so is hardest to fiddle with. It helps to record all the steps you take and make sure elements are replicated consistently (e.g. the font types and sizes, colophon and ISBN placement). This week I finished a print cover which wraps around, but also has a front that can be saved as a separate file for the e-book. Again it is an efficient way to do things.



Post a Comment