steelneck
Hello.
I want to share a little knowledge about scanning negatives in Linux. Many Linux-newbies curse over Xsane, especially when trying to scan negatives, they have a really hard time getting it right. Once i had those problems too until i learned the power of Xsane-design. Most people have not the right medium definitions for the film they are trying to scan, most have not even heard of "medium definitions" and those who has have, have usually tried to use the GUI to make one, and ends up in even deeper trouble.. Forget the GUI for making a medium definition, it will only make things more confused, edit the file containing the definition directly instead.
The medium definitions lives in a hidden file under your home dir, those files beginning with a period, in this case the ~/.sane/xsane/xsane.mdf In this file you specify the levels for black point, white point and gamma for each channel, and this is not the usual 0-255 levels as in 8-bit editing, this is in much higher bit depth when Sane is talking directly to the scanner hardware, it is 0-6553600 for each channel as in 48-bit color.
When you have made a medium definition for you film, and scanner(!) - not all scanners are the same -, you will find colors much easier to get correct. Most of the time all you have to do is to adjust light and contrast in the simple mode, the gray channel. Even the auto button and batch scanning will work quite nice. Those coming from the "digital darkroom" usually also find the light, contrast and gamma controls a bit akward, but for anyone coming from the real darkroom, light and contrast will feel quite natural, once you get that medium definition at least in the ballpark.
As most of you know, a scanner can handle a lot more density range than a color negative, leaving quite some room on the left and right side of the histogram, the film base is not really black and highlights are not white. We also have the issue of the orange mask that is a little different in different films, the channels do not align, different amount of room for each channel on each side of the histogram, a mess. But with a good medium definition.. viola! All three channels now aligns pretty well in the histogram = go dark at the same levels and blows out at the same levels. This is why adjusting even works quite well even in the simple mode with a good definition. One could say that we have white balanced the scanner to the film in question. You got your self a decent starting point.
In the real darkroom we could think of it as adjusting the color filters. And when those are right, a thin, overexposed negative just needs a little less exposure on the paper, that corresponds to the light slider in Xsane. And the contrast slider corresponds to longer development time of the film.
The xsane.mdf starts with:
"MEDIUM_DEFINITIONS"
12 #The number is just an example
This says that the file contains 12 different definitions, so if you have 8 definitions and add one more, you have to change the number to 9 or else xsane will get confused , and scary things might happen.. huuu. This file should also be edited with Xsane shut down, this is because xsane write this file when it get closed, it overwrites it with the values it read when it was started. So edit with xsane closed.
To give an example of a definition, here comes my definition for Fuji Reala:
"medium-name"
"Fuji Reala"
"medium-shadow-gray"
0
"medium-shadow-red"
949000
"medium-shadow-green"
403000
"medium-shadow-blue"
280000
"medium-highlight-gray"
6553600
"medium-highlight-red"
3800000
"medium-highlight-green"
2480000
"medium-highlight-blue"
2090000
"medium-gamma-gray"
65536
"medium-gamma-red"
65000
"medium-gamma-green"
53000
"medium-gamma-blue"
50000
"medium-negative"
1
Note the last entry "medium-negative" and its value 1, this tells xsane to invert the negative to positive (0 is for slides), and thus becomes "medium-shadow-red" the right "highlight" slider in a normal levels tool. In this case it is pulled up from 0 to 949000. In an 8-bit histogram that would be 37. And of course the opposite is true for "medium-highlight-red", that controls the shadow levels once the image is inverted. It can take some time to wrap ones head around this, thinking inverted in 48-bit color, but after some experimenting you get the hang of it.
You can copy my values above, but do not expect it to be right for your scanner or film, though it can be used as a starting point just to wrap your head around the concept.
What you want to accomplish is that all channels get equal levels in extreme highlights and shadows, and of course that grey becomes grey. But do not overdo the white and black point, you want a bit of room on both sides of the histogram when you do a preview-scan with all sliders set to neutral. If you do not have some room on each side, with neutral unadjusted light and contrast sliders, you loose the great exposure forgivness that negative film has.
Hope i been to some help.
I want to share a little knowledge about scanning negatives in Linux. Many Linux-newbies curse over Xsane, especially when trying to scan negatives, they have a really hard time getting it right. Once i had those problems too until i learned the power of Xsane-design. Most people have not the right medium definitions for the film they are trying to scan, most have not even heard of "medium definitions" and those who has have, have usually tried to use the GUI to make one, and ends up in even deeper trouble.. Forget the GUI for making a medium definition, it will only make things more confused, edit the file containing the definition directly instead.
The medium definitions lives in a hidden file under your home dir, those files beginning with a period, in this case the ~/.sane/xsane/xsane.mdf In this file you specify the levels for black point, white point and gamma for each channel, and this is not the usual 0-255 levels as in 8-bit editing, this is in much higher bit depth when Sane is talking directly to the scanner hardware, it is 0-6553600 for each channel as in 48-bit color.
When you have made a medium definition for you film, and scanner(!) - not all scanners are the same -, you will find colors much easier to get correct. Most of the time all you have to do is to adjust light and contrast in the simple mode, the gray channel. Even the auto button and batch scanning will work quite nice. Those coming from the "digital darkroom" usually also find the light, contrast and gamma controls a bit akward, but for anyone coming from the real darkroom, light and contrast will feel quite natural, once you get that medium definition at least in the ballpark.
As most of you know, a scanner can handle a lot more density range than a color negative, leaving quite some room on the left and right side of the histogram, the film base is not really black and highlights are not white. We also have the issue of the orange mask that is a little different in different films, the channels do not align, different amount of room for each channel on each side of the histogram, a mess. But with a good medium definition.. viola! All three channels now aligns pretty well in the histogram = go dark at the same levels and blows out at the same levels. This is why adjusting even works quite well even in the simple mode with a good definition. One could say that we have white balanced the scanner to the film in question. You got your self a decent starting point.
In the real darkroom we could think of it as adjusting the color filters. And when those are right, a thin, overexposed negative just needs a little less exposure on the paper, that corresponds to the light slider in Xsane. And the contrast slider corresponds to longer development time of the film.
The xsane.mdf starts with:
"MEDIUM_DEFINITIONS"
12 #The number is just an example
This says that the file contains 12 different definitions, so if you have 8 definitions and add one more, you have to change the number to 9 or else xsane will get confused , and scary things might happen.. huuu. This file should also be edited with Xsane shut down, this is because xsane write this file when it get closed, it overwrites it with the values it read when it was started. So edit with xsane closed.
To give an example of a definition, here comes my definition for Fuji Reala:
"medium-name"
"Fuji Reala"
"medium-shadow-gray"
0
"medium-shadow-red"
949000
"medium-shadow-green"
403000
"medium-shadow-blue"
280000
"medium-highlight-gray"
6553600
"medium-highlight-red"
3800000
"medium-highlight-green"
2480000
"medium-highlight-blue"
2090000
"medium-gamma-gray"
65536
"medium-gamma-red"
65000
"medium-gamma-green"
53000
"medium-gamma-blue"
50000
"medium-negative"
1
Note the last entry "medium-negative" and its value 1, this tells xsane to invert the negative to positive (0 is for slides), and thus becomes "medium-shadow-red" the right "highlight" slider in a normal levels tool. In this case it is pulled up from 0 to 949000. In an 8-bit histogram that would be 37. And of course the opposite is true for "medium-highlight-red", that controls the shadow levels once the image is inverted. It can take some time to wrap ones head around this, thinking inverted in 48-bit color, but after some experimenting you get the hang of it.
You can copy my values above, but do not expect it to be right for your scanner or film, though it can be used as a starting point just to wrap your head around the concept.
What you want to accomplish is that all channels get equal levels in extreme highlights and shadows, and of course that grey becomes grey. But do not overdo the white and black point, you want a bit of room on both sides of the histogram when you do a preview-scan with all sliders set to neutral. If you do not have some room on each side, with neutral unadjusted light and contrast sliders, you loose the great exposure forgivness that negative film has.
Hope i been to some help.
