Often, drawing tools on computer screens suck, especially on the tiny computer screens we’re starting to use more often (“cell phones”, “tablets”). But computers are coming with better and better cameras, printing on paper remains high-quality and relatively cheap, and there are a wide variety of tools available for drawing and otherwise making images on paper — pens, markers, pencils, erasers, rulers, compasses, paintbrushes, paint, ink, and so on. Furthermore, it’s easier and easier for a computer to recognize barcodes and things like that on photographed or scanned paper, which allows it to precisely coregister things drawn on preprinted paper, both spatially and in colorspace.
There’s a web site called Fontifier where you print out a template with a box for each letter, draw letters in all the boxes, then scan in the template and upload it to their web site. The website converts the uploaded scan into a TrueType font and sells it to you.
Scantron machines optically detect the positions of pencil marks on preprinted paper slips, used to encode the answers to multiple-choice exams, marking each one correct or incorrect.
Presumably human-readable bureaucratic forms are commonly processed by OCR nowadays.
So, what new possibilities does this modality of interaction offer us?
Most obviously, you could print out a document that needs illustrations with blank pre-sized boxes for the illustrations, illustrate it by hand, then scan or photograph the document to automatically add the illustrations to the document. Barcodes on the paper would orient the illustration-cropping software, and if the printer is colorimetrically calibrated, color calibration patches on the paper can correct for unknown lighting color, camera white balance, and camera focal plane sensitivity.
Additionally, though, if you can print out an image in a single color channel, such as magenta, then you can use it as a reference for another image you draw on top of it in black — whether merely marking points or areas of interest on the underlying image with “X” or “O” or similar marks, marking arbitrary contours, or even drawing an arbitrary overlay image. If grayscale isn’t needed, you can even do this on a monochrome printer by printing the reference image in light gray, then thresholding it away to leave only the human-made black marks.
For example, you could draw accent marks for a font on a preprinted letter glyph to provide reference, places of interest on a map (perhaps with their names or other data such as hours of operation), coloring or shading on an outline drawing, connections on a network diagram, data points on a scatterplot with a preprinted reference graticule.