The left-right full-image stereograms... but those are less common than the pattern-based "magic eye" stereograms. Those are the reverse of this - in the linked image and the left-right full-image stereograms, they're done by crossing your eyes to a point closer to your eyes than the original image.

The pattern-based "magic eye" stereograms are done by looking through the image to focus on a point deeper into the screen further from your eyes.

The latter I think are less painful because they use the more natural depth-perception distances of your eyes instead of using what feels like more unnatural positions, but that might be my bias because I'm a bit farsighted. Maybe they're just more common because they're visually inscrutable at first and so you get the "reveal" of the 3D contour from a single large image instead of two already-visible small ones.