S
Stroker1
Guest
This is an old rant. The other thread prompted me to tell ya'll about it.
Now, this is a quick example. Certainly not as satisfactory as I like. My biggest boo-boo in this example is that I only dealt with highlights only. Should have worked on the lowlights a tad, but I didn't. Or I should have used Histogram. More on this a bit later.
So what's the problem? Fleshtones. Two different pictures that I want to hack together, but the fleshtones are way wacky. :yech:
The little girl with the ponytails is the primary subject. Toss the in baby, the secondary subject, and I'm sure you can see the differences in the fleshtones. Major :yech:.
On Girlie #1, I picked a spot on her lapel (highlight) and sampled the colour. I got:
RGB: 245,208,183
On Girlie #2, I picked a spot on her forehead (highlight) and sampled the colour. I got:
RGB: 243,154,144
In Girlie #2, the G and B channels need to be brought up a few notches.
Enter Curves.
Added a Curves Adjustment Layer and clipped it to Girlie #2. Then I got busy in the G and B channels.
In G, I brought 154 up to 208.
In B, I brought 144 up to 183.
After that, it was a small matter of eyeballing it for tweaks. No big deal since the above gave me a *very solid* foundation to work from.
The bad thing is that there is some extra green in Girlie #2's cheeks. I should have tweaked some of that out.
Variations -
1) Sample highlights and lowlights. Then get busy with Curves. Twice as much work as the above, but certaintly worth it. I should have done some lowlight work in my quick example.
2) Sample highlights, midlights, and lowlights. I tried this once and got excellant results. You would have sworn that both subjects were in the same photo to begin with. It was spot on all around.
3) Histogram. You can use Histogram to get an average of the seperate channels. Make a decent selection that includes a full range of tones in both subjects, get averages from Histogram, then get busy with Curves. This is certainly functional, but I've found that a little more tweaking usually has to be done, which is no big deal. (Histogram is an unsung hero IMHO. Useful info can be culled from Histogram.)
Cool little technique. I like it and use it. Gotta love Curves. I'll tell ya'll about a few more Curve tricks one of these days.
play.fiddle.learn
Now, this is a quick example. Certainly not as satisfactory as I like. My biggest boo-boo in this example is that I only dealt with highlights only. Should have worked on the lowlights a tad, but I didn't. Or I should have used Histogram. More on this a bit later.
So what's the problem? Fleshtones. Two different pictures that I want to hack together, but the fleshtones are way wacky. :yech:
The little girl with the ponytails is the primary subject. Toss the in baby, the secondary subject, and I'm sure you can see the differences in the fleshtones. Major :yech:.
On Girlie #1, I picked a spot on her lapel (highlight) and sampled the colour. I got:
RGB: 245,208,183
On Girlie #2, I picked a spot on her forehead (highlight) and sampled the colour. I got:
RGB: 243,154,144
In Girlie #2, the G and B channels need to be brought up a few notches.
Enter Curves.
Added a Curves Adjustment Layer and clipped it to Girlie #2. Then I got busy in the G and B channels.
In G, I brought 154 up to 208.
In B, I brought 144 up to 183.
After that, it was a small matter of eyeballing it for tweaks. No big deal since the above gave me a *very solid* foundation to work from.
The bad thing is that there is some extra green in Girlie #2's cheeks. I should have tweaked some of that out.
Variations -
1) Sample highlights and lowlights. Then get busy with Curves. Twice as much work as the above, but certaintly worth it. I should have done some lowlight work in my quick example.
2) Sample highlights, midlights, and lowlights. I tried this once and got excellant results. You would have sworn that both subjects were in the same photo to begin with. It was spot on all around.
3) Histogram. You can use Histogram to get an average of the seperate channels. Make a decent selection that includes a full range of tones in both subjects, get averages from Histogram, then get busy with Curves. This is certainly functional, but I've found that a little more tweaking usually has to be done, which is no big deal. (Histogram is an unsung hero IMHO. Useful info can be culled from Histogram.)
Cool little technique. I like it and use it. Gotta love Curves. I'll tell ya'll about a few more Curve tricks one of these days.
play.fiddle.learn
Curves ROCK! 
I'm beginning to learn that...