I got this picture I generated myself and I don't understand how come we can see two diagonal lines when, I think, there is no actual lines.

Squarish rainbow

Zooming in:

Zooming in

Center: Just another 2px wide square…


After thinking about it, I think it might come from the fact that, if you zoom really closely, it is only a succession of square, each time 2px smaller and of a different color.

Although there is no actual line directly built, we can see the corners of each square clearly and, together, our brain understand this as two diagonal lines. … I might just have answered my own question but I would appreciate your opinion/confirmation on this.

Related: My question on meta-SE about where to ask this question.

  • 2
    $\begingroup$ That is so cool! I'm not sure what's going on any more... $\endgroup$ – anongoodnurse Jul 12 '15 at 3:07
  • $\begingroup$ What is supposed to be in the center? Also what format are the pictures in and is there any compression? $\endgroup$ – StrongBad Jul 12 '15 at 15:09
  • $\begingroup$ @StrongBad Added a picture of the center, just a 2px wide square. I generated it using PhotoImage in tkinter in Python, assigned each pixel one by one, creating the bigger square first then smaller and smaller (not very efficient, but pretty to watch it being built) then saved it as a png with PhotoImage.write, I don't know the compression or anything else, I left it by default. $\endgroup$ – jeromej Jul 12 '15 at 17:05
  • 1
    $\begingroup$ My guess is it is a rendering artifact. The corner pixels have a different number of same color pixel neighbors from edge pixels, so they are in fact a different color when draw on the screen. $\endgroup$ – StrongBad Jul 12 '15 at 17:25
  • 1
    $\begingroup$ @StrongBad I checked, it isn't. When I zoomed on the rendered picture (from a screenshot for instance) it looks exactly the same as when zoomed on the actual picture. Nice guess though, I didn't think about that. $\endgroup$ – jeromej Jul 13 '15 at 4:24

I can't zoom in on the picture, but what I see is bleeding from one color into the next along the diagonals (most evident in the green to blue and purple to red transitions). Are you saying that they are not there when you zoom in? I would think that because there is a gradient in the color transition, the gradient has to be emphasized in an angled meeting, more pronounced the sharper the angle (e.g.if it were a triangle, the angle would be 120° as opposed to 90°). It's not an optical illusion.

In short, you see two transecting lines on the diagonal because they are strongly suggested. The mind sees this accurately and reads the suggestion.

It is kind of like asking, why, in the following picture, do I read all the dotted lines as lines, where only two of them are actually solid lines?

enter image description here

Or, if you will, why do I see curves here?

enter image description here

Because the color differences in the offset squares are really there, suggesting a familiar curve, though in this question, it would be fair to ask Why do the curves appear smooth? In fact, the curves are not smooth.

If you do want to see interesting optical phenomenons,

  • stare at the center of your image for 30 seconds, then scroll down to the blank area next to my diagrams. You will see squares of reciprocal colors not in your original (like a bright yellow.) Also interesting is that as the image fades, blinking will "refresh" it.

  • after staring directly at the center for a minute or two, switch the focus very slightly to one side then back; you'll be pleasantly surprised by bursts of vibrant color parallel to each other each time you change.

Edited to add: This may be an optical illusion, but if it is, I'm unsure of what type it is. Looking at the zoomed image, clearly the impression is that there is some bleeding, but that's not the reality. Sorry I can't help more on this one. But it does get much more intriguing!

  • 2
    $\begingroup$ Hi, thanks! I provided a zoom in in my question as suggested. $\endgroup$ – jeromej Jul 12 '15 at 2:59
  • $\begingroup$ +1 for the answer. Please feel free to add some stuff from my chat if you think it fits your answer :) $\endgroup$ – AliceD Jul 12 '15 at 3:59
  • $\begingroup$ @AliceD (Your chat? I know the reply wasn't directed to me but it left me curious!) $\endgroup$ – jeromej Jul 12 '15 at 6:27
  • 1
    $\begingroup$ @JeromeJ - I was chatting with anongoodnurse about the bleeding of green in blue as well as the parallel lines within the colored bars. The parallel lines create a physical pattern and perpendicular edges in the corners. these two factor are enough to create an edge. It is not an illusion in my opinion. $\endgroup$ – AliceD Jul 13 '15 at 12:09

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.