Navigation

    • Login
        No matches found
    • Search
    • Recent
    • Back to Lospec.com
    1. Home
    2. b236
    b236

    b236

    @b236

    6
    Posts
    1
    Profile views
    0
    Followers
    0
    Following
    Joined Last Online

    • Profile
    • More
      • Following
      • Followers
      • Topics
      • Posts
      • Groups
    b236 Follow

    Best posts made by b236

    This user hasn't posted anything yet.

    Latest posts made by b236

    RE: Censor - a perceptual palette analyser

    @quickmarble amazing! thank you!

    posted in Palettes •
    RE: Censor - a perceptual palette analyser

    @quickmarble thanks for the test! i wouldn't rely on von luschan's work as it is quite old, but yeah, the skin tone line is just an approximation (i believe it is cca 15 degrees, but i don't know for sure) so there's always going to be some spread:
    https://homepages.inf.ed.ac.uk/rbf/CVonline/LOCAL_COPIES/GONG1/cvOnline-skinColourAnalysis.html
    or
    https://larryjordan.com/articles/color-correction-make-people-look-normal/
    anyway, it was just an idea, i don't think it is crucial to be included.

    posted in Palettes •
    RE: Censor - a perceptual palette analyser

    i see, thank you very much!

    posted in Palettes •
    RE: Censor - a perceptual palette analyser

    btw. i just got confused by spectral distribution graph, could you please explain it to me? i assumed that 4100 - 6650 range is in Kelvins, but that doesn't seem right.

    posted in Palettes •
    RE: Censor - a perceptual palette analyser

    hi! sorry for the late reply.

    i understand. i was considering it only from practical perspective, but you make a valid point. and after all, if there are going to be marked primaries and contour line (great idea, btw.) it will show both chroma as distance from center to outline and saturation as distance from center to contour.
    i don't think you have to be concerned about visibility for bigger palettes, because palettes like this: https://raw.githubusercontent.com/Quickmarble/censor/master/examples/aurora.png basically show everything by themselves. but i guess primaries can be marked also like this:
    example-4.gif

    skin tone line: there's a line (often marked as -I) in vectorscopes which shows approximate location of skin tones. of course, it works only for normal light conditions, but generally it is very useful for colour grading. i couldn't find any technical specification for that, so i can post only how it looks like:
    https://images.squarespace-cdn.com/content/v1/55aeab62e4b0485b4e8ab594/1503436597539-E8CGC52JPWCVO06QJK16/038+The+Skin+Tone+line-04.jpg

    https://images.squarespace-cdn.com/content/v1/55aeab62e4b0485b4e8ab594/1503418215638-SKAWGH3T9YDDTOFDG9CS/038+Skin+Tone+line-PMF-Squarespace-Thumbnail-1920x1080.jpg?format=1500w

    alternative UI with greys would be great option.

    PS: you're right about visibility of darker colours, your solution sounds better.

    posted in Palettes •
    RE: Censor - a perceptual palette analyser

    hi, nice job! i'd like to make few suggestions regarding vectorscope (polar hue-chroma diagram) and non-palette elements:

    1. i believe vectorscope would be more practical if it was hue-saturation. because a) digital vectorscopes work like that, b) dawnbringer's palette analyzer works like that, c) usually all painting softwares provide users with control over saturation, but not chroma, so it is rather alien concept for most. as a result it is confusing to see fully saturated yellows to be close to the center of the diagram, while one would expect them to be on the outside edge of the circle. right now it makes proper assesments about the palette balance harder.

    2. it shouldn't be rotated and mirrored compared to other diagrams (notice the location of primaries in polar hue-lightness diagrams).

    3. the skin tone line would be nice.

    4. vectorscope should show location of primaries (RYGCBM). it's hard to estimate where the colours are located in relation to primaries.
      and of course, if you don't want to change vectorscope to hue-saturation, then this is even more important feature. in that case it would actually be better to show not just primaries, but all main colours (R, O, Y, G, C, A, B, M) if not all hues separated by 30 degrees. it seems like there are areas of the disk where no colours will ever be present and users should be able to see that clearly.

    5. i recommend the background and elements which aren't part of palette itself to be presented in neutral greys - black, 50% grey and white only. having one of the palette colours to be dominant colour significantly affects the overall look of the palette and perception of its colours. for instance, one of your examples with green background creates the feeling the palette is green heavy, while the opposite is actually true.

    i made a quick example here:
    example-3.gif

    PS.
    also, it would be neat if you could colour the lines in spectral distribution and temparature diagrams with corresponding palette entries.

    posted in Palettes •