Roger Clark
Banned
Simon,
Here is my attempt. Processing went like this (many steps took longer to write down than do):
raw: exposure=0
recovery=0
fill light =0
blacks=0
brightness +50
contrast +25
clarity +7
vibrance +10
saturation +13
save as image a, 16-bit tif.
cs5:
select background with magic wand tool
refine edge: smooth 12, feather 8.2, contrast 9, shift edge +10
darken with curves
select bird
refine edge: smooth 12, feather 13, contrast 9, shift edge -35
curves: increase brightness
select bird again, smaller parts of main body
refine edge: smooth 12, feather 13, contrast 9, shift edge -35
curves: increase brightness
save image (new name) (image b)
ImagesPlus (software package, version 3.0):
Richardson-Lucy image deconvolution: 5x5 12 iterations; save image
open image b again:
Richardson-Lucy image deconvolution: 5x5 25 iterations; save image
back to CS5:
Stack the 3 images in layers: image b (top), 5x5 12 iterations, 5x5 25 iterations (bottom)
erase layers so bird is 5x5 12 iterations, stump is 5x5 25 iterations
select background of top layer: gaussian blur radius=1.8 (noise reduction)
unsharp mask 5x5 25 iteration layer (stump): radius 1.5 pixels, 81%, threshold 6
unsharp mask 5x5 12 iteration layer (bird): radius 1.5 pixels, 81%, threshold 6
save image (d)
flatten image
clean up a few places of ringing (around bill and legs) with blur and clone tools
final tweaks:
select bird body: curves for some contrast
select stump: curves for contrast
convert to sRGB, 8-bit
convert to LAB, sharpen luminance channel radius=0.7, amount 120, thershold 6
convert back the RGB
save a tif
save for web
In the attached image, I pushed the deconvolution steps more than I normally would to illustrate the effect of true sharpening to bring out feather detail on the bird and fine textural detail on the post. The result is a little more noise than I would normally accept in my own images, but I think the idea is illustrated. One can trade noise for detail and where one sets the bar is up to the individual.
In the posted image, there is some residual ringing that I would clean up (e.g. with clone or blur tool), but also would be less with less aggressive processing.
Roger
Here is my attempt. Processing went like this (many steps took longer to write down than do):
raw: exposure=0
recovery=0
fill light =0
blacks=0
brightness +50
contrast +25
clarity +7
vibrance +10
saturation +13
save as image a, 16-bit tif.
cs5:
select background with magic wand tool
refine edge: smooth 12, feather 8.2, contrast 9, shift edge +10
darken with curves
select bird
refine edge: smooth 12, feather 13, contrast 9, shift edge -35
curves: increase brightness
select bird again, smaller parts of main body
refine edge: smooth 12, feather 13, contrast 9, shift edge -35
curves: increase brightness
save image (new name) (image b)
ImagesPlus (software package, version 3.0):
Richardson-Lucy image deconvolution: 5x5 12 iterations; save image
open image b again:
Richardson-Lucy image deconvolution: 5x5 25 iterations; save image
back to CS5:
Stack the 3 images in layers: image b (top), 5x5 12 iterations, 5x5 25 iterations (bottom)
erase layers so bird is 5x5 12 iterations, stump is 5x5 25 iterations
select background of top layer: gaussian blur radius=1.8 (noise reduction)
unsharp mask 5x5 25 iteration layer (stump): radius 1.5 pixels, 81%, threshold 6
unsharp mask 5x5 12 iteration layer (bird): radius 1.5 pixels, 81%, threshold 6
save image (d)
flatten image
clean up a few places of ringing (around bill and legs) with blur and clone tools
final tweaks:
select bird body: curves for some contrast
select stump: curves for contrast
convert to sRGB, 8-bit
convert to LAB, sharpen luminance channel radius=0.7, amount 120, thershold 6
convert back the RGB
save a tif
save for web
In the attached image, I pushed the deconvolution steps more than I normally would to illustrate the effect of true sharpening to bring out feather detail on the bird and fine textural detail on the post. The result is a little more noise than I would normally accept in my own images, but I think the idea is illustrated. One can trade noise for detail and where one sets the bar is up to the individual.
In the posted image, there is some residual ringing that I would clean up (e.g. with clone or blur tool), but also would be less with less aggressive processing.
Roger
Attachments
Last edited: