File talk:DPI and PPI.png

There's something very wrong with your printer, or at least your printer driver's sRGB model if you're sending it that image then getting that result on the page. It looks almost as if it's been converted for display on a CGA monitor, rather than being printed out in CMYK. I'll see if I can produce a more realistic copy myself, though I'm not entirely sure how to upload it.

For a start, flat blue should NOT come out as a mixture of "magenta and black". No, that's what dark purple would look like. ISO standard for making primary blue, with suitable inks, is 50% magenta, 50% cyan, with those figures reducing evenly towards zero as you adjust towards a lighter or darker tone, being joined either by nothing (lighter), or a compensatory amount of black (darker). In fact, given that ink printing is additive, and each "pixel" can have any combination of the four inks present, this too is false - it should be 100% M, 100% C, and then maybe up to 50% K for a darker tone before you have to start backing off the M and C.

I understand that you're maybe trying to represent more the idea that printers effectively make images from nothing more than a great many very finely dithered low-colour pixels, and will probably have to introduce some inaccuracy in my own version to convey that (maybe flattening to 8 colours (or 5...) with error diffusion rather than 15~16), but choosing plainly inaccurate colours doesn't seem the right way to go about it.

(Similarly, the aqua green colour in the central dot really can't be represented as "cyan + white". 75% Cyan + 25% yellow (representing a real-world 90+30), maybe...? And quite how you ended up with some of the still vaguely greenish surrounding pixels having magenta elements to them when it's not a 3-colour (ie CMY only) printer, I wouldn't even want to try and guess...) 193.63.174.211 (talk) 09:35, 19 February 2014 (UTC)

... Update: I presume you used a K/W/C/M/Y palette and regular Floyd-Steinberg dithering in GIMP? Its handling of error diffused colour reduction with nonstandard (or indeed, standard) palettes seems... suboptimal, to put it kindly. Not only do I get a very nearly identical result to yours when trying to reduce it that way (which I would have rejected out of hand, if I'd done it for something else independently), but it's not very good for other customised palettes. I'll see what I can do, however.

This does at least mean I can now see that yours DOES have some yellow in the middle, despite what the caption text says.. 193.63.174.211 (talk) 09:51, 19 February 2014 (UTC)