I used ACR and adjusted the Exposure and Brightness to amounts that seemed to best satisfy the histogram and display. I adjusted Contrast waay down because the scene contrast was just soo high to begin with. Shadows were also set to 0 because there was still plenty of black point at 0.
I also used ACR to interpolate the image to 25MP, to give me the best possible starting number of pixels.
Finally I brought the image into PhotoShop as a 16 bit image, again to maximize the starting tonal information.
Inside of PS I used Levels to re-distribute the tones to a more natural (to my eye) distribution. I used a mid-tone sharpening Action to enhance sharpness and local contrast. I further tweaked the contrast using Curves, emphasizing mostly the shadow areas. I used 2 does of a custom Action designed to increase sharpening in a very subtle fashion and improve overall contrast, especially edge contrast.
I used 2 doses of Neat Image to reduce noise, once for overall appearance and a second time for despeckling. I also built a new layer from a color-selection mask just to concentrate on the blue uniforms and I applied a 3rd dose or NR just to that layer.
I added Saturation back to the reds in the base layer and then added Sat over in the blue layer. Flatten Image finished the process.
I resampled the image back to native size and used the Bicubic Sharper option.
If I were doing any quantity of images I would, of course, build some custom actions to speed the process dramatically.
I have communicated with Ann before this post and she suggested that the blue uniforms should be more navy blue in color, so that would be a correction I would account for if I were doing a more serious job of the image.
The final image looks too bright to me, but I guess it's that you have properly calibrated badass monitor, while I am now typing at some random notebook.
What surprised me was that you haven't used Neat Image as a first thing at all. I would think that having the image in raw and having the camera 'noise' profile should give the best results. But this is just my theory, nor I do have registered Neat Image, nor I had original raw.
I also used ACR to interpolate the image to 25MP, to give me the best possible starting number of pixels.
Is there some magic telling that 25MP is the best number of pixels ?
To have some fun I tried to play with the image myself using gimp. It was quite simple.
1) Colors/Levels
Auto settings gave me decent results, I just had to lighten midtones (below the histogram, drag the middle arrow to left). To enhance color balance I also had to adjust blue and red channel separately.
2) Filters/Enhance/GREYCstoration
This plugin is not shipped with gimp, but is also opensource (google it). It's using de-noising algorithm similar to Neat Image or Noise Ninja (granted compared to them it's just a toy).
Comments
I'll see what I can do in this regard after work today.
ann
My Galleries My Photography BLOG
Ramblings About Me
Ann sent me a RAW image to process and here is the before and after:
Here are the links to the full sized images:
Before:
http://ziggy53.smugmug.com/photos/411943753_Ls9qw-O.jpg
After:
http://ziggy53.smugmug.com/photos/411943069_guq6Q-O.jpg
The process:
White balance was just under 4K.
I used ACR and adjusted the Exposure and Brightness to amounts that seemed to best satisfy the histogram and display. I adjusted Contrast waay down because the scene contrast was just soo high to begin with. Shadows were also set to 0 because there was still plenty of black point at 0.
I also used ACR to interpolate the image to 25MP, to give me the best possible starting number of pixels.
Finally I brought the image into PhotoShop as a 16 bit image, again to maximize the starting tonal information.
Inside of PS I used Levels to re-distribute the tones to a more natural (to my eye) distribution. I used a mid-tone sharpening Action to enhance sharpness and local contrast. I further tweaked the contrast using Curves, emphasizing mostly the shadow areas. I used 2 does of a custom Action designed to increase sharpening in a very subtle fashion and improve overall contrast, especially edge contrast.
I used 2 doses of Neat Image to reduce noise, once for overall appearance and a second time for despeckling. I also built a new layer from a color-selection mask just to concentrate on the blue uniforms and I applied a 3rd dose or NR just to that layer.
I added Saturation back to the reds in the base layer and then added Sat over in the blue layer. Flatten Image finished the process.
I resampled the image back to native size and used the Bicubic Sharper option.
If I were doing any quantity of images I would, of course, build some custom actions to speed the process dramatically.
I have communicated with Ann before this post and she suggested that the blue uniforms should be more navy blue in color, so that would be a correction I would account for if I were doing a more serious job of the image.
Moderator of the Cameras and Accessories forums
My Galleries My Photography BLOG
Ramblings About Me
thank you for your detailed explanation
The final image looks too bright to me, but I guess it's that you have properly calibrated badass monitor, while I am now typing at some random notebook.
What surprised me was that you haven't used Neat Image as a first thing at all. I would think that having the image in raw and having the camera 'noise' profile should give the best results. But this is just my theory, nor I do have registered Neat Image, nor I had original raw.
Is there some magic telling that 25MP is the best number of pixels ?
To have some fun I tried to play with the image myself using gimp. It was quite simple.
1) Colors/Levels
Auto settings gave me decent results, I just had to lighten midtones (below the histogram, drag the middle arrow to left). To enhance color balance I also had to adjust blue and red channel separately.
2) Filters/Enhance/GREYCstoration
This plugin is not shipped with gimp, but is also opensource (google it). It's using de-noising algorithm similar to Neat Image or Noise Ninja (granted compared to them it's just a toy).
3) Resize image to 800xwhatever the height is
4) GREYCstoration again
Full image at
http://neuron.smugmug.com/photos/412446847_bpED7-O.jpg
Small image:
Cheers
--
neuron
Pictures | Website | Blog | Twitter | Contact