I’ve tweaked the code to try and improve the final image (more info below), and here’s the latest batch…
The new version of the code breaks each images down into a series of blocks of pixels, each being 2 pixels wide and 3 high. Each block is then replaced with a book cover whose colour roughly matches that block of pixels…
…this improves the skin tone colours, as well as smoothing the boundaries of sudden colour changes.
You can see the difference with these two versions of the picture of Helene — the old code is on the left, and the new code is on the right:
Quite a few people have emailed me images to convert, so it might take a little while to do them all!