Divide a picture into four pieces with a horizontal and vertical cut. Invert the colors of two pieces diagonally across from each other, then put the pieces back together.
This puts crosshairs at the point of intersection without overlaying any obscuring reticles. There's a bit of ambiguity because the intersection defines a point between pixels.
Color inversion doesn't work so well if the region is gray. Investigate alternative reversible color transforms (e.g., rotation of hue, though that won't work either for gray). We may need some AI or human advice to decide when to use which, because it depends on other colors and textures in the neighborhood. The orthogonal cuts could also be at an angle, e.g., X.
Previously vaguely similar in the theme of drawing lines without drawing lines.
Probably no lossless transformation works if it's all pixels of random colors.
No comments :
Post a Comment