color adjustment
Even if you don't have the original image to compare to, adjusting the colors slightly on an image with obvious color problems will improve it greatly. If you still need help after reading this, members on the IRC channel (irc://irc.rizon.net/moe-imouto ) may be able to help you.
Basically, color should be adjusted via tonecurves, and then saturation a bit. Never use brightness, contrast, etc. because they ignore certain color conditions.
You can use your scanner's driver to adjust colors, but this tutorial will use Smart Curve because it can show color histograms accurately and visually. It's an 8bf plugin for Photoshop, and also works on Xnview or Irfanview, etc.
Ideally, color adjustment should be done twice, when scanning and after filtering. But we'll use the raw image resized with bspline (and slightly blurred to simulate a filtered image) here.
#1 on the following image is a 16bit/channel raw scan without any color matching profile.
As computer paintings are drawn on white canvas, I adjust white at the brightest color (R,G,B=255,255,255) first. Then try to pick the points where red, green and blue look right, then move them separately at where the rightest peaks of histogram reach at the right edge (#2, #3)
If the canvas isn't white, then adjust it to the color as you see.
Secondly, adjust the dark point. Unlike the white, paintings aren't always true black, and since screening divided the dark colors randomly, the histogram has lost the correct blackness. I've just adjusted black loosely in #4.
Then I should recover the contrast of the dark parts which were lost during the printing process. Using master curves, add contrast on the dark evenly, then reduce the brightness of the bright sections to the point the image looks a bit dull and bright. Adding saturation later will make the image sharper and darker (#5).
Also, if the image has strange hue, correct it with tonecurves (#6). I use two points per channel to adjust color balance. If I try to adjust color balance with only one point, the effect will be too strong in the dark areas.
After finishing the tonecurves adjustment, add a bit of saturation.
The above process is a standard way of color adjustment, but some scans have problems that may be difficult to solve with standard procedures.
Hue shift of highly saturated red/magenta is one of those problems. In this case, I use replace color tool.
Here is the final result
(edited from midzki's original post, forum #4619)
Basically, color should be adjusted via tonecurves, and then saturation a bit. Never use brightness, contrast, etc. because they ignore certain color conditions.
You can use your scanner's driver to adjust colors, but this tutorial will use Smart Curve because it can show color histograms accurately and visually. It's an 8bf plugin for Photoshop, and also works on Xnview or Irfanview, etc.
Ideally, color adjustment should be done twice, when scanning and after filtering. But we'll use the raw image resized with bspline (and slightly blurred to simulate a filtered image) here.
#1 on the following image is a 16bit/channel raw scan without any color matching profile.
As computer paintings are drawn on white canvas, I adjust white at the brightest color (R,G,B=255,255,255) first. Then try to pick the points where red, green and blue look right, then move them separately at where the rightest peaks of histogram reach at the right edge (#2, #3)
If the canvas isn't white, then adjust it to the color as you see.
Secondly, adjust the dark point. Unlike the white, paintings aren't always true black, and since screening divided the dark colors randomly, the histogram has lost the correct blackness. I've just adjusted black loosely in #4.
Then I should recover the contrast of the dark parts which were lost during the printing process. Using master curves, add contrast on the dark evenly, then reduce the brightness of the bright sections to the point the image looks a bit dull and bright. Adding saturation later will make the image sharper and darker (#5).
Also, if the image has strange hue, correct it with tonecurves (#6). I use two points per channel to adjust color balance. If I try to adjust color balance with only one point, the effect will be too strong in the dark areas.
After finishing the tonecurves adjustment, add a bit of saturation.
The above process is a standard way of color adjustment, but some scans have problems that may be difficult to solve with standard procedures.
Hue shift of highly saturated red/magenta is one of those problems. In this case, I use replace color tool.
Here is the final result
(edited from midzki's original post, forum #4619)
Updated by midzki over 13 years ago