http://forums.dpreview.com/forums/read. ... e=33717039
еще пару постов идет обсуждение этого.
Iliah Borg писал(а):
Dear Tom,
Please allow me to address the sharpening a little.
Sharpening for weak channels that look blotchy is best applied on a separate duplicate layer, with pixel-level sharpening, that is setting radius to 0,3..0.4, threshold to 0 and amount to high like 350, with a reducing the opacity of the layer to get appropriate results. Different amounts of sharpening is usually applied to shadows and midtones, while no sharpening at all is applied to highlights.
One of the things to keep in mind is that demosaicing propagates artefacts from underexposed channels to the channels that are exposed properly. To avoid this, it makes sense to use a simplified form of demosaicing, that is binning. Next, we slightly sharpen the channels with pixel-level, overall (100/1/10), and de-fog sharpening (10/20/0) sharpening (of course, using separate layers, separate sharpening settings for shadows where we want to split the blotches and midtones, where we want mostly to apply overall sharpening and de-fogging); and upsize them using bicubic sharper, setting dpi to any even number higher then 400 and setting the size to the print size, combined with immediate downsizing using same bicubic sharper, keeping the size but setting dpi to the value we need for printing. To add to this, using a split of channels and operating in floating point helps further.
Of course, everything packs nicely into a Photoshop action.
Please allow me to address the sharpening a little.
Sharpening for weak channels that look blotchy is best applied on a separate duplicate layer, with pixel-level sharpening, that is setting radius to 0,3..0.4, threshold to 0 and amount to high like 350, with a reducing the opacity of the layer to get appropriate results. Different amounts of sharpening is usually applied to shadows and midtones, while no sharpening at all is applied to highlights.
One of the things to keep in mind is that demosaicing propagates artefacts from underexposed channels to the channels that are exposed properly. To avoid this, it makes sense to use a simplified form of demosaicing, that is binning. Next, we slightly sharpen the channels with pixel-level, overall (100/1/10), and de-fog sharpening (10/20/0) sharpening (of course, using separate layers, separate sharpening settings for shadows where we want to split the blotches and midtones, where we want mostly to apply overall sharpening and de-fogging); and upsize them using bicubic sharper, setting dpi to any even number higher then 400 and setting the size to the print size, combined with immediate downsizing using same bicubic sharper, keeping the size but setting dpi to the value we need for printing. To add to this, using a split of channels and operating in floating point helps further.
Of course, everything packs nicely into a Photoshop action.
еще пару постов идет обсуждение этого.
Via est Vita.