Another Cone Stamp tool Tutorial

Started Feb 21, 2013 | Discussions thread
Redcrown Senior Member • Posts: 1,510
Re: Another Cone Stamp tool Tutorial

I spent some more time fooling with the technique in the video, trying to figure out how it really works. Why it works sometimes for some people and not others.

I'm still a bit confused, but here are a few things I discovered:

1. The brightness/contrast adjustment is critical, and the -50 value in "Legacy" mode is the only setting that works. I suspect that some who tried and failed had missed the "legacy" setting. That's not the default in CS6, and the video author does not mention it.

2. The radius of the Gaussian Blur and the High Pass can be varied, but they must be identical. Regardless of what radius values you use for the blur and high pass, the contrast adjustment must remain at -50 legacy mode.

3. The technique does not generate a true identical version of the original, but it's real close. Close enough to be practical. Even if you can't see a difference on the monitor, it can be measured with a "Difference" blend mode and the histogram. Here is how:

3a. After doing the "split", stamp a new "merged" layer on top of the stack and turn off the split layers, leaving only the merged layer on top of the original layer.

3b. Put the merged layer in "Difference" blend mode. The screen will probably turn black.

3c. Display the historgam and look at the "Std Dev" value. If there is absolutely no difference between two layers, the StdDev value will be zero. In my tests of the technique I get a StdDev around 0.75. That's barely at the threshold of a "visible" difference. The visibility of a difference also depends on the quality of your monitor and the quality of your vision. But at 0.75, it's good enough to be of practical use.

4. The technique is actually better than a true frequency separation (FS) for the purpose given in the video. It separates tone from texture more than the true FS technique. The true FS technique is somewhat controversial. Some people swear by it, others don't think it's worth all the effort. While it pro ports to separate tone from texture, in reality that split is about 75/25.

The "tone" layer will contain about 75% of the tone from the original and 25% of the texture. Ditto, but vice-versa for the texture layer. So, if you clone or heal on the tone layer, you are actually copying 25% of the texture as well. The technique in the video appears to give a much higher separation. I'd guess it's around 95/5.

Bottom line -While it's an interesting exercise, I think other "dodge and burn" techniques are easier and better to use for dealing with shadows and highlights.

Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow