Comment This looks pretty good with the notable exception of the color correction. There is a widespread habit among photochrom "restorations" to remove the yellow tone in the erroneous belief that the photochroms yellowed with age, while in fact the turquoise-to-yellow or turquoise-to-orange gradation of the sky was a characteristic feature of the DPCo (and P.Z.) photochroms. As such EV is clearly not enhanced by the digital color correction. ~ trialsanderrors (talk) 16:17, 7 September 2010 (UTC)[reply]
While I certainly have seen problems with that - c.f the recent Charge of the Light Brigade delist - LoC scans are not colour-corrected, so it's necessary to adjust accordingly, though conservatively. As such, my only colour change was to set the white point to the brightest point in the image, and the black point to the darkest. Since it is known for a certainty that the LoC images are almost never true to the original colours, if we accept this image had any white at all, that should be a safe presumption. -Adam.
If you get the opportunity, try to look at the photochroms in the book by Sabine Arqué (ISBN 2212542704) for comparison. I know of four different collections of photochroms: the LOC, the Zürich central library, Photoglob's own archives and the private collection of Marc Walter. There are a few overlaps in the published images between those collections, and from comparing those I've now concluded that the LOC color scheme doesn't need any correction – if anything their ppmsc/ppmsca scans are too light. The color scheme was intentionally set to yellow hues to create the impression of sunset, even if from the shadows it is clear that the original photo was taken during mid-day. ~ trialsanderrors (talk) 14:33, 13 September 2010 (UTC)[reply]
When using what is literally the only information we know for certain - the histograms - we get an image with perfectly naturalistic colours, appropriate to the time of day. I can't see any reason to engage in speculative yellowing which would require literally every ink in the image to have a yellow tint to it, including the paper put behind the image before scanning it. Adam Cuerden(talk) 23:52, 13 September 2010 (UTC)[reply]
I'm not talking about speculative yellowing, I'm talking about not engaging in speculative color correction. ~ trialsanderrors (talk) 14:20, 14 September 2010 (UTC)[reply]
That might be reasonable - IF almost every LoC scan with a colour box didn't show clear evidence of a colour cast, usually yellow. The LoC tends to scan off-colour, usually biasing yellow, and that's well-known. LoC scans are known to almost always have the wrong colours, due to being uncalibrated, so using the information available int he histogram is a basic and necessary correction. Adam Cuerden(talk) 14:39, 15 September 2010 (UTC)[reply]
