tehqin's blog

By tehqin, 13 years ago, In English
Going through old round I found that I was particularly lucky to not participate in the round with this problem:

http://www.codeforces.com/contest/71/problem/C

See the picture uses colors red and green to distinguish between the knights. I am red green color blind and cannot tell the difference between the two colors so the picture doesn't add as much to help to explain the problem as others. Did anyone else have this problem? Any ideas for ways to help me if issues like this come up in an actual contest?

My small request is that problem writers that use pictures take into account those of us not fortunate enough to see all colors. =) Thank you!
  • Vote: I like it
  • 0
  • Vote: I do not like it

13 years ago, # |
Rev. 2   Vote: I like it 0 Vote: I do not like it

It reminds me of time when I was owner of black/white SVGA monitor and was trying to play Heroes of Might and Magic I. Knights here are distinguished by the color of their little banners - and sometime it was just impossible. For some other game with similar problem I just written little utility which helps me to distinguish colors.

I think that you too should use some little set of screen tools:
- one which yields little floating hint with the color gradients of the pixel which is under mouse pointer;
- and one which allows for a moment turn off all red or all green components of the whole screen picture.

I suppose that you can find such tools for your specific OS and hardware easier than I or even write them yourself! Short search through web shows me that maybe here are not a lot of such software, but something surely exists, for example: http://www.toledo-bend.com/colorblind/shareware.asp
  • 13 years ago, # ^ |
      Vote: I like it 0 Vote: I do not like it
    Hmm, I think it would be best if I could find a browser plugin for images in the browser to be distorted away from the problem colors. The trouble right now is finding such a plugin or writing one myself. Anyone have experience with writing chrome or firefox plugins?
    • 13 years ago, # ^ |
        Vote: I like it 0 Vote: I do not like it
      It is similar to writing any other javascripts for manipulating with document being rendered by browser. However I do not think you will find a way to simply change all the color mappings of the rendered document as if it were simple image. I suppose you would need rather to convert all elements and classes colors, which would be more tricky. And I doubt that you can in any way process colors of images represented with "img" elements.
13 years ago, # |
  Vote: I like it 0 Vote: I do not like it
Did you try “invert colors” feature that every modern OS have? It looks like it can help.
13 years ago, # |
  Vote: I like it 0 Vote: I do not like it
Original picture in the statement is really awful. Either author assigned colors without thinking or he has an improperly calibrated display. Even on my Mac, which is known to have finely tuned display out of the box, all the colored points in this picture just look equally dirty. I have no perceptional disorders, but I still need to make a significant effort to distinguish colors of the points.

I believe authors should at least check how their illustrations are seen in grayscale mode.
  • 13 years ago, # ^ |
      Vote: I like it 0 Vote: I do not like it
    It is strange. I tried original picture on different noname monitors - color points have distinguishably different hue and brightness! Surely, not nicest colors - but if you "need to make significant effort" and you are sure your eyes are all right - then it is time to think about replacement for your Mac! ;-)
    • 13 years ago, # ^ |
        Vote: I like it 0 Vote: I do not like it
      Both colors have the same brightness of 50 percents, I measured it. I have no idea why do you think the brightnesses are different. Obviously they have different hue just because one is red and another is green, but it's not enough to see this difference. I think you (and all others) should better stop using noname displays that fail to show the same brightness for colors with the same saturation level and just different hues.
      • 13 years ago, # ^ |
          Vote: I like it 0 Vote: I do not like it
        Red points are RGB(127, 0, 0) and green ones are RGB(0, 127, 0). But brightness of different gradients does not look equal to human eye, you know. Check google for "RGB brightness formula". I suppose that commonly green is about twice brighter than red and red is twice brighter than blue.

        You can easily check this fact if you have ability to turn your monitor to grayscale mode and look at three areas with colors #FF0000, #00FF00 and #0000FF...

        This chart of eye color sensitivity may explain this better:
        • 13 years ago, # ^ |
            Vote: I like it 0 Vote: I do not like it
          Oh my God... How to resize the picture? It's source is here.
        • 13 years ago, # ^ |
            Vote: I like it 0 Vote: I do not like it
          I see that green color is brighter than red one only when I turn off the calibration. It means that the purpose of calibration is to eliminate this perceptive effects. Anyway, authors shouldn't rely heavily on color coding in their illustrations.
          • 13 years ago, # ^ |
              Vote: I like it 0 Vote: I do not like it
            I see that green color is brighter than red one only when I turn off the calibration. It means that the purpose of calibration is to eliminate this perceptive effects.

            No, I think brightness is hard to see due to small size of color dots and white background. If "calibration" would really distort normal levels of gradients then almost all photographs would look incorrectly.

            Anyway, authors shouldn't rely heavily on color coding in their illustrations.

            I agree. Even more - illustrations should be avoided if it is possible - since among programmers here are even totally blind people. May be they would like to join CF community in a future!
            • 13 years ago, # ^ |
                Vote: I like it 0 Vote: I do not like it
              The fact is that green dots are nearly twice brighter than red dots when I turn the calibration off. And calibration is not as simple as reducing each RGB component linearly. I watched your CF photo with different calibration profiles — they look definitely different, but non of these views could be called “wrong”. The default profile just makes colors look less luminescent.
              • 13 years ago, # ^ |
                Rev. 3   Vote: I like it 0 Vote: I do not like it

                You are right - "wrong" does not mean that colors are diverted in some extreme way. However I could not check your monitor to understand what it is "calibrating" and, I fear, I could not provide more comprehensible example.

                Hm... I tried to use the same colors (127,0,0) and (0,127,0) on a black background. Try to look at it, may be with reduced "brightness" of monitor or with your eyelids almost shut. I suppose it is easier to find the difference about which I am speaking, when color objects are on black background (since contrast is greater).

                On the other hand the fact of which I am speaking is only of minor importance - you would need to remember it, maybe, only if you will by chance working with programming for some specific photo filters - or when designing some machine color-recognition devices...

                UPD: Sorry, picture disappears from the comment, so I provided link to it.
                UPD: Now I see two pictures. I tried to delete one. May be it is a problem of hosting I use... I am sorry, but seems I could not simply upload picture here.
13 years ago, # |
  Vote: I like it 0 Vote: I do not like it
I also have some problems with distingiushing red and green. It is difficult to me to notice difference between red and green circles on the picture.
13 years ago, # |
  Vote: I like it 0 Vote: I do not like it
The author should substitute red circles with simple "X" marked circles... maybe is better