ur opinion, I solicit it

[zwol]

It is an unfortunate mathematical fact that you cannot draw a dotted line that is one pixel wide, an even number of pixels long, and has dots at either end. This being the case, which of the following possible alternatives to that unachievable ideal looks least bad? Including None of the above; I have a better idea which I will explain, of course.

You might want to magnify the image to see exactly what is going on; however, the aesthetics at this size are what matters.

ETA: I should have explained that both of the boxes in each row are generated by the same algorithm; the difference is that in the left column, the vertical sides are an odd number of pixels long, whereas in the right column, all four sides are an even number of pixels long.

Comments

[elsmi.livejournal.com]

Is anti-aliasing nonsense an option?

I don't like the first row and am fairly indifferent to the rest. Maybe the second from the top of the left and the bottom right ones are somewhat better, I'm not sure.

[xorphus.livejournal.com]

Agreed on antialiasing, if possible. I prefer big dots on the clockwise end, which gives a sense of positive forward motion.

[zwol.livejournal.com]

Re antialiasing, see my response to elsmi. I am leaning toward big dots on the clockwise end myself.

[zwol.livejournal.com]

I tried anti-aliasing nonsense; it produces a big gray blur in the middle of the line. If you don't like the first row, you'll hate that.

[falsedrow.livejournal.com]

I like the last one; the absence of a big dot is not as noticeable as the presence of one.

[mrissa.livejournal.com]

Yes, this.

[brooksmoses]

Huh. Oddly enough, for me it's slightly the opposite. (They're both pretty noticable -- especially on the case where there are two of them; it's rather less so on the case with just one. Edit: Of course, there are always two of them -- I missed the right-bottom one on the left-hand box, which perhaps argues strongly against my point.)

Speaking of which, the right-hand box on row 2 is missing its right-bottom dot, isn't it?

Also, an option you didn't try is the double white dot in the middle of the line. (I would expect this would also be quite obvious and no good, though.)

[zwol.livejournal.com]

Could you clarify which you mean by "the case where there are two of them" versus "just one"? I don't think any of the boxes has just one artifact.

The right-hand box on row 2 is not missing its right-bottom dot, but I'm not sure I can explain why. Perhaps it would do to say that all of these variations come from adjusting the start and end points of the lines, except the first row, where we are drawing two lines per side?

Double white dot in the middle of the line is not an option because if you actually draw it, you discover that it either has to be slightly off-center or four pixels wide. I decided that was just too ugly.

[zwol.livejournal.com]

Reply to edit: So you find it more noticeable when the two skipped corners are diagonally opposite? That's interesting, I find it more noticeable when they're on adjacent corners. Of course, I drew the damn thing.

[brooksmoses]

Or I find it more noticable when they're on the right-hand column rather than the left; it's not exactly that easy to say. (Maybe there was a strategic dust speck on my monitor? It's pretty dusty....) It's more noticable now that I know it's there, certainly.

As for whether the right-hand box is missing its right-bottom dot, I think that's because of our different understandings of the algorithm. If it's defined by shifting the lines, then the white corner dot is correct. However, if it's defined literally as "put a big two-pixel dot at the right or bottom end of the line to make it come out even", the corner dot should be black.

[zwol.livejournal.com]

Well, I was specifically looking for patterns that I could, in the end, implement by shifting the lines, you see.

Also I'm pretty sure that a three-pixel L-shaped thing in the lower-right hand corner would be worse than what's there now. But it looks like people prefer #3 or #4 anyway, so it's probably moot.

[fadethecat.livejournal.com]

The third row is my favorite, but near equal with the second, as least distracting. Or maybe the second and then the third... They're fairly near equal. The fourth isn't bad, but is a little odd, and the first is distinctly weird.

[lutin.livejournal.com]

corner gaps.

[elisaana.livejournal.com]

I'd say third row is best, followed by the first row. They both feature symmetry that reinforces the rectangular-ness of the image, whereas omitting the dots makes me think it's a parallelogram or an insubstantial border.

The clockwise big dots trump the middle big dots because the effect is less noticeable. The middle big dots could be someone's favorite effect (making it frustrating if they don't appear for odd numbers of pixels) or least favorite effect (causing frustration in the opposite circumstance).

[zwol.livejournal.com]

Thanks for the detailed answer!

Indeed, the middle big dots would not appear for odd numbers of pixels. I should probably have included a sample all-four-sides-odd box for reference.

Unfortunately, there seems to be a jury split between row 3 and row 4 developing...

[queenpam.livejournal.com]

anything but the first one. the rest are about even.

[kirizal.livejournal.com]

#4 looks best to me, though neither 2 nor 3 offend me.

[vvvexation.livejournal.com]

The last two are more noticeable to me; the lack of symmetry feels weird. Corner gaps at the bottom might be slightly better than big dots at the bottom, though--and also I think they'd be better than big dots in the middle.

[packbat.livejournal.com]

I say either big dots at the corners or gaps at the corners.

[puppygalore.livejournal.com]

I say 3, followed by 4.

[weinberg.wordpress.com (from livejournal.com)]

Just to be a contrarian, I like the first one because it's symmetrical - all the other ones make me feel off balance.
But I am outweighed by the voters.

The third row

[(Anonymous)]

Rotational symmetry is still symmetry.

Re: The third row

[weinberg.wordpress.com (from livejournal.com)]

But it makes me dizzy! It may meet the definition of symmetry, but not the sensation of balance.