Originally posted by BruceBanner and then do all that stuff that site recommends such as open up the image you want, 'Duplicate Layer' etc..., or is that simply not necessary, just instead opening up the file you want as your fb cover, and then changing canvas size/image size etc? I know it's silly, but I would appreciate a step by step account of exactly what you did as your version came out significantly better than mine, and I think I did (at the time) do the same stuff as you in terms of 'max quality 100, Optimized/Progressive off' etc... :S The steps; cropping, resizing and adding sharpening specifically might help.
I didn't do it like that. The workflow depends on your software. Here is one way I would do it, just off the top of my head:
a) Open the dng in Photoshop.
b) Do the PP in the CameraRaw dialog. You can do more later, just try to get it as close to right as you can. Don't overdo the contrast and stuff, but don't be too shy
c) Now you have the image open in PS. I always right click Background and then 'Layer from background', and then I duplicate the layer so that if I make a mistake, I have the bottom most layer as the original
d) Duplicate the layer again, and apply Filter, Other, Highpass. A relatively small number, between 0.8-1.9. It should be the number that makes things look 'in focus' but not 'prickly' (number too small) or 'blocky' (number too big)
e) turn the Highpass layer blending mode to Overlay and lower its opacity down to 50 - 70 %. Zoom in to 100% and play with the opacity until it looks good, but not oversharpened.
f) Crop photo to 1702x630 aspect ratio. Grab the crop tool and input those numbers to get the exact aspect ratio. Select the crop that you want, it can be just a part of the photo, as long as the aspect is right.
g) Resize. PS has a dialog for resizing and it offers different rescaling algorithms. It tells you which is best for reduction - choose that one (or automatic) and resize to 1702. If the other number does not align perfectly to 630, you can crop off one or two pixels after resizing
h) Now grab the layer beneath the Highpass one. The one that is a copy of the original background. Select that, and then go Filter, Sharpening, Smart sharpen. Look at the image at 100% when you do this
i) You can do more adjustments now, like finalizing contrast and saturation (levels, curves, use whatever you wish, or all of it. I do small adjustment with a lot of those, so the small adjustments stack up without looking too harsh)
j) If the sky is banding or there is posterization, add Film grain or Noise (monochrome, choose the %. Whatever is the minimum % that makes the banding go away)
k) If the photo looks a bit over the top, as in too sharp or too noisy, you can lower the opacity of that layer a little bit. You have the original Background layer behind as the base
l) If you want to add text or something you can add it now. Text tool, add new layer, choose font, etc. Keep colours harmonic and make sure it is readable. Avoid Comic sans
m) File, Save for web or devices, choose color space (sRGB), disable progressive and optimized, and set compression to highest quality.
n) Send me $100 for this wonderful tutorial.
If you use something like Lightroom as well as PS, I would do
everything in Lightroom, except for the final resizing, cropping, and sharpening. I would export the file as 16bit uncompressed tiff from LR, work on that file in PS, and then delete the tiff when the upload of the final jpeg is successful. You can use RawTherapee as well, though it has a steeper learning curve
I don't do this process on all my photos, or even many of them. I would only do it to share a small thumbnail online, for web purposes. And I would adjust the process per photo. Normally I just use Lightroom or RawTherapee and nothing else. I don't crop unless there is something terrible in the frame. I only resize for web sharing.
Edit: Oh, and Noise Reduction, if you shot at high ISO. NR should be done near the beginning, before the first sharpening. Note that resizing the photo so much will remove a lot of the noise anyway, so don't apply a lot of NR. NR removes some detail, creates artifacts.. Lens CA correction should be done near the beginning, as well. Probably before the other things. LR does it automatically after you set it up. I think the CameraRaw dialog has this option somewhere