Page 1 of 1

Posted: Fri Jan 28, 2011 11:58 pm
by mikemeister_admin
If I don't have a color cast and then I use the saturation slider in LAB to rotate the curves in the a and b channels, then this shouldn't introduce a color cast. It modifies green and magenta in the a channel by equal amounts and it modifies blue and yellow in the b channel by equal amounts.

But what happens now if, for example, I introduce 2 points on the LAB curves by setting contrast pins? If I rotate the L channel, no problem. If I don't remove the contrast pins from the curves in the a and b channels, then does rotating the curves in the a and b channels risk introducing a color cast (week 6, example 4)?

More generally, doesn't modifying the blue and yellow parts of the curve in the b channel by different amounts risk introducing a color cast?

Thanks,

Richard

Posted: Sat Jan 29, 2011 4:41 pm
by ggroess

If I don't have a color cast and then I use the saturation slider in LAB to rotate the curves in the a and b channels, then this shouldn't introduce a color cast. It modifies green and magenta in the a channel by equal amounts and it modifies blue and yellow in the b channel by equal amounts.


IF..you do not have a color cast then it should not introduce one...Big If...When you guys post images the first thing I do is over saturate them to see if they have a cast.  if they do that becomes part of the discussion.  The saturation slider moves the a and B channels equally so if a cast appears it was there to begin with.


But what happens now if, for example, I introduce 2 points on the LAB curves by setting contrast pins? If I rotate the L channel, no problem. If I don't remove the contrast pins from the curves in the a and b channels, then does rotating the curves in the a and b channels risk introducing a color cast (week 6, example 4)?


When you use a contrast pin in LAB you change the slope of the curve in between the two points and the rest of the curve responds by flexing.  There is still a zero point somewhere on that new curve and you need to move the linked points around together to get the zero point back.  Each channel is separate at that point so adjustments to the A channel with a contrast pin will only effect the a channel.  This is the reason I tell you to set a marker hue clock.  You will have to move the linked points together to restore the channel values in the hue clock in order to not introduce a cast.  So if you are adjusting the A channel you want to set a hue clock on something that might be stronger in the B channel.  That way you do not set the hue clock in the middle of your contrast pins range.  Neutrals are a prime choice but not every image has a neutral.


More generally, doesn't modifying the blue and yellow parts of the curve in the b channel by different amounts risk introducing a color cast?

Thanks,

Richard


In Lab anytime you mess with the zero you get a color cast.  Yellow Blue is no different than Red Green...the difference is in the visual perception of the result.  We tend to see color casts easier in lighter areas of the image.

I'm trying to think of a better example for you....this might be better as a video answer..I'll try this weekend to make a short video that shows what is going on here....it is a geometry problem and well it's been 30 years since me and geometry had a heart to heart...

Greg

Posted: Sat Jan 29, 2011 6:26 pm
by mikemeister_admin
Thank you very much. It is becoming clearer now.

Setting a neutral and then placing contrast pins isn't a good idea from what I can see. It generally messes things up.

So in each channel, you should move the contrast pins around together to get back to your zero point before going on to the next channel. Clear in theory, not always easy without some practice.

I was a math major and geometry wasn't my strong point, and luckily I haven't had to use it now for about 40 years.

Richard

Posted: Sat Jan 29, 2011 7:59 pm
by ggroess
I have posted a video on the forum about this...

http://www.curvemeister.com/forum/index.php/topic,3486.msg15659.html#msg15659
Greg