Here's a really simple one, but I've found it useful when checking to see how well images tile.
It offsets your image by an inputted value.
I'm not sure if it would be considered bad practice to use getTileSize() within init() to limit the fields, but it does seem to work. If you increase the size of the image while using Offset you do need to add a new instance to be able to offset by higher values, but nothing actually breaks. If it bothers you just replace the two instances of getTileSize() within init() with 64.
Edit:
Here's another simple one: Simple Color.
It takes RGB values and colourizes the image, which can be faster than using the existing colourize node when you just want shades of one particular colour. As the result tends to be a bit smooth for pixel art, there's a quantize feature which acts like a palette knife, cutting down the number of colours used. If you want, you can set all three RGB values to their maximum and just output a quantized heightmap.
Since taking this screenshot I changed the spelling to the American spelling to match the existing nodes.
