For as long as I can remember I’ve had trouble digitizing traditional media, and have thus been a bit reluctant to do a lot of colored pencil or watercolor work.
At the same time, I’ve always scanned on the family all-in-one printer/scanner, the latest of which is an HP OfficeJet Pro 8600, which is a pretty vanilla low to mid range unit. These AiOs have always been fine – or maybe even good – for scanning pencil and pen drawings because they’ve tended to wash out a lot of color. I don’t know if this attribute is specific to HP, to cheap combo devices, or to something else, but it helps when all you want to keep is dark lines.
2016 was one of my worst years for digital coloring pretty much since I started. Looking at my archives, the previous lows were five in 2010 and 2011, but last year I managed four. And two of them were tests, and the other two were so-so. 2017 seems to be on more or less the same track, but at the same time PaintsChainer has at least temporarily re-invigorated my interest in painting.
I once saw an April Fools or some other joke that suggested that some future version of Photoshop could paint your pictures for you using fancy algorithms or whatnot. Well, that day has actually arrived, though it’s a bit more primitive than I’d have liked.
J pointed me to PaintsChainer some weeks ago, and from the moment I saw it I was hooked. PaintsChainer does exactly what I just described: you upload a lineart and it paints it for you. Machine Learning, Whatever, don’t care how. The outputs are quite amazing even if you don’t touch any of its knobs and dials, but they have some flaws and for now I think it’s difficult to correct them.