RustyBug Online Upload & Sell: On
|
solarishead wrote:
I tried the technique with this image however, and it didn't work well because there is literally nothing in the image that is white so I think the techniques mentioned above are probably more appropriate.
Nice video ... pretty quick and to the point. I thought he might "tweak" on the curves after using the eyedropper, but he didn't go there. I reckon that could be the "next" video.
The one thing to bear in mind regarding the "chosen points", is that A) there is an assumption being made that the point is SUPPOSED to be neutral. B) there is an assumption being made that different points are SUPPOSED to be neutralized.
The caveat to "A" is that we can take ANY COLOR ... and if the tonal value is high enough, it will appear white. Thus, not all "brightest" points are always THE point for choosing as the basis for assessing the light color. Most times, likely ... just be watchful / studious of the light and an area's inherent color properties to not be fooled, i.e. (think about what's happening vs. blind faith on the brightest area).
The caveat to "B" is that sometimes we actually want the adjustment of an area toward neutral ALLOWED to push / pull / shift the other areas to ACCENTUATE the color difference of the lighting when in variable / mixed lighting vs. uniform lighting. Here again, that "think" thing comes in handy.
The 50% gray layer technique in the video ... could be a handy way of finding your middle tonal values. I think you could also do something similar with analyzing at different values (i.e. 25% gray, 75% gray, 90% gray ... 10% hue, 90% hue, etc.) by using different fill. That one will be a nugget to tuck away for some some additional refinement (say like finding the most offending area of a color cast).
The threshold layer ... yup, it helps me find the points. Main diff, being I select multiple points vs. only a single point. Sometimes, a single point will be angled toward the sky / nearby object / grass, etc. and be reflecting the reflected light. So, like how we have global adjustments and selective adjustments for contrast, sharpening, etc. ... same here. By selecting multi-points, it helps me to look and see if / where I'm contending with global vs. selective issues. In my "pick 4" points, 3 outa 4 (Sesame Street's one of these things is not like the others) usually is my baseline global (noting why #4 isn't following suit) adjustment. So, I try to pick points not only in reference to their tonal value, but also their angular position to the source(s) of my color.
One that comes to mind is the black shoes on Ben's nephew wedding shots. The point on the shoe can be chosen on the side or back of the shoe (which may be darkest), but if the shoe is in or near the grass or colored wall, etc., it will picking up those colors. However, if we choose a point on the toe (which may NOT be darkest tonal value), it is pointing upward toward the sky and receiving the same source light that is acting as our key / fill light.
The threshold is a good tool ... but, here again it too isn't always a stand alone tool to be used by itself. You may still need to use your good ol' noggin' a bit too.
Just out of curiosity, when making many adjustments to an image, when is the best time to perform color correction? Specifically wondering if it's best to do first or last, or ever both?
I do mine on my first adjustment layer (PS). I seem to recall Scott Kelby advocating color first as well (but, don't hold me to it, that was years ago I glanced at one of his books).
(Technically, I have my original source layer, and a duplicate of that, then a color balance layer.)
Here's the reason:
If I have my color properly neutralized to the light, then my image's color contrast is at its optimum amount of hue contrast without an overburdening cast limiting the amount of hue contrast that I have to work with in the subsequent adjustments. Meaning, I don't have to ratchet saturation or contrast to get t contrast that already exists, but is being covered up by the overall cast (kinda like a hue "veil", acting somewhat like veiling flare for contrast reduction). In this manner, I do less damage / artifacts in the generation of the "overcooking" process to achieve similar contrast levels.
Kind of a "less is more" perspective. Less cast = more contrast. Less push = more clean.
For artistic rendering ... I can always take and reduce the amount of adjustment to "put it back in" if there are those times that I really feel it is pertinent to the mood or color of the light for it to be revealed.
But, if I go on about the other tasks of pushing things around, and THEN try to correct color, it can be like reigning in two horses going in opposite directions. That's not to say there isn't some additional refinement of color in intermediate or later stages, but in Demming fashion, I find it prudent to address the lion's share of it earlier, rather than later, as it has cascading implications on some of the other processing decisions to be made.
|