It is my understanding that the 3 tools for improving color tones of a
photo are "Brightness-Contrast", "Levels" and "Curves". "Levels" from
what I have read is a happy medium between the other two. I'd like to
understand more about how to approach "Levels" and definitely
"Curves". Any links or tutorials that you found helpful would be
appreciated. I've visited lots of places, but have not found much
detail on this topic. Most just say use it, and I'm already doing that
:-) The March issue of Linux Magazine had a nice article on GIMP and
improving photos. It was the first item I've really seen that showed
the details of the tools and explained why.
I understand in "Levels" that the left is black and the right is
white, and the middle gamma. A good picture has tones in all 5 zones.
That's the extent of my knowledge on levels and I hope to change that
Does "Auto" change all 3 channels, or do I need to select each channel
and click "Auto"?
If you don't specify a channel, but use "Value", are you making the same
change to all 3 channels?
When using the "eyedroppers", do you select the darkest and lightest
in the picture that exist, or that you wished existed?
It would be nice if "Auto" showed the changes it made via the 3
triangle slides or a number in the text boxes. It does change the
preview image, but the graph and everything else is the same. I
usually click "Auto", "Ok", and get back in to look at the changes
made in the graph, and say, wow, that's neat, how did they do that and
why can't I do it manually? When I use "Levels" without "Auto", with a
specific channel selected, I turn the picture magenta, or green :-)
Great for aliens, not so good for humans :-)
Curves. I have spent a lot of time clicking on the line and pulling up
and down, on the left and right, not really knowing why I'm doing it.
How do you use "Curves" efficiently?
What reasoning is used to put a dot on the line in a specific position
and pull the line up or down?
Any examples or doc that perhaps helped you get started with this
This is what I'm working on:
I have a 90 year old sepia toned picture of a person's profile. I
first go Image -> Mode -> Grayscale to get black and white. When I
analyze the photo with Image -> Colors -> Levels, and look at the RGB
channels, most everything is on the left (dark) inside the first
column, with just a little bit in the next column. The chart is
largest for Red, then Green is smaller, and Blue is the smallest. The
Layers -> Colors -> Histogram confirms (mean): R (34%), G(25%),
B(15%). When I Image -> Mode -> Decompose the photo, selecting RGB,
the red layer looks the best (lightest, most visible), followed by the
green and blue (the darkest). All I did was click on the eye to enable
and disable visibility of two, while viewing the third, to see if I
learned anything from it.
I've experimented with:
Method 1: Layer -> Colors -> Levels, "Auto"
Method 2: Duplicate the layer, and change mode to Screen
Method 3: Duplicate the layer, Layer -> Colors -> Desaturate,
Layer -> Colors -> Invert, and change mode to Overlay
They all seem to turn out similar results. Actually Method 2 and 3
combined looked nice. In general, I seem to be getting too light on
one side of the face, and not light enough on the other side.
If I do a final Filters -> Enhance -> Sharpen, do I need to merge the
layers? Is the Sharpen applied to a specific layer, the sum, or the
Any suggestions or discussion appreciated?
Gimp-user mailing list