In message <4ef8950f.6050...@pilobilus.net>,
Steve Kinney <ad...@pilobilus.net> wrote:
> Can you recall the specifics of how you tamed the streak?
I know that I duplicated the base layer, used a mask to isolate the
streak (so that it would be the only visible part of its layer), and
applied color corrections to it until it closely matched the
un-streaked part of the layer below. Then I merged it down into the
base layer and worked on color correction image-wide. The details
are lost in the mists of time.
> >You face three problems: An unwanted color, an unwanted
darkening, and loss of detail.
> I'm going to take issue with one small part of what you just said.
Specifically, the last part... loss of detail. [...] from where I
am sitting, it appears to me that underneath the green streaks there
is very nearly just as much detail as in other similar parts of the
The eye and brain are excellent correction filters and fill in an
amazing amount of detail from subtle cues. In the most heavily
damaged parts of the image, contrast is very limited. Working
carefully on small patches of the image with varous filters and
effects I have been able to get "nearly perfect" brightness and
color - but the results are distinctly blurry, in comparison to more
or less identical regions that were not heavily damaged.
> [...] if I just select that whole stripped area and then start
fiddling things, correct me if I'm wrong, but won't that likely
create some result that is less than optimal for each of the two
sub-sections of the stripe (i.e. the lighter part and the darker part)?
Right you are. I was only looking for a crude approximation to
guide "from scratch" reconstruction of the damaged area.
> from where I'm sitting, there is still plenty of _real_ underyling
image detail (under the green stripe) that I would sort-of like to
preserve, if possible, rather than just paining over or replacing
outright with other little chopped up parts of the undamaged
background or, God forbid, my own clumsy attempts to hand-paint in
some details. (I am as clumsy with a brush as The Hulk is with a
To paraphrase from Mad Max, "Accurate detail is a question of time.
How much time do you have, how much accurate detail do you want?" :o)
> I understand only a bit about layers. Should I be separating this
image into red, green, & blue layers and then be attempting my
repairs primarily or exclusively on the green layer?
Um, nope. Those are channels, not layers.
Layers are like multiple images stacked on top of each other. The
usefulness of this is largely due to a feature called a layer mask:
In the Layers tab of your dialogs dock, layer masks (when present)
appear as a second rectangle beside the first, always in black and
white. Masks have a stencil like effect: Where a layer's mask is
white, the layer is opaque and visible in the finished image (unless
it is under another opaque layer in the stack). Where the layer's
mask is black, the layer is transparent and invisible in the
finished image. Shades of gray are partially visible, the lighter
they are the more of the layer "shows through".
Confused? Don't feel bad, it took me forever and a day to get used
to using layers with masks to repair and compose images. But today
they seem entirely natural to me and I couldn't live without them.
All I can suggest is to find tutorials that make use of layers, work
through them, and be persistent until it starts to make sense.
> Also, if you could give me a one sentence definition/description
of "gradient masking" I'd appreciate it.
I probably shouldn't have said that because I am not sure anyone but
me uses the term "gradient mask". That is when you add a mask to a
layer, and make part of that layer blend smoothly into the layer
below by using a gradient from black to white on the mask. This
results in the layer fading from invisible to fully visible as you
move from the black to the white part of the layer mask.
Example: Imagine a flash photograph of a line of people on a stage,
taken from a seat at the far end of the front row. The people on
the end of the stage nearest the camera will be properly exposed,
but those at the far end of the stage will be underexposed. Just
increasing brightness to bring out the underexposed far-away people
makes the ones close up way too bright, with "blown out"
highlights. What do do?
Make a copy of the base layer (i.e. the original image) as a new
layer, and brighten the whole new layer until the underexposed
people at the far end of the stage are clearly visible. Add a mask
to the altered layer (right click its thumbnail in the Layers dialog
and select "Add layer mask"). Click on the new mask to select it,
then use the "Blend tool" to fill the layer mask horizontally with a
smooth gradient from black to white, making the too-bright people on
stage invisible (black end of the gradient), while leaving the
previously underexposed people at the far end of the stage fully
visible (white end of the gradient). Real life example, works well
in many similar cases.
> With regards to the clone and smudge tools, I haven't actually
delved into those myself yet [..]
The Clone tool enables you to "paint" a continuous sample from one
location in an image, to any other location in the image. A vastly
useful retouch tool.
The Smudge tool is for smearing pixels around with whatever brush
you have selected. Great for softening sharp edges and making
"seams" in cloned or pasted regions vanish.
With these and other tools, remember that ctrl-Z is your friend.
Easily your best friend. :o)
> >For a beginner this is a fairly major hobby project...
> I was hoping not. And like I say, I do believe that if I had
dodge&burn equivalents for contrast, saturation, and hue, I'd have
already cleaned up thsi image, toot sweet.
And for my part, I am hoping that someone on the list will point out
exactly how to do that with the existing tool set. No matter how
long I use this thing, I can still be very pleasantly surprised when
someone points out something I never noticed, or a simple method
that never occurred to me before.
gimp-user-list mailing list