Approach Using Color Finesse's WVM and Vectorscope
I've been practicing color matching using Levels to match foreground and background objects in their R, G, and B channels. I've gotten somewhat comfortable with this process so I've moved onto trying color correction using SA's Color Finesse in AE.
I have 3 major questions:
1) What does somebody look for when using the WVM? – I imagine I'm supposed to stretch values to about 90 for contrast, but beyond that it's hard for me to determine what I should be seeing? What is clean? What is sloppy? If I align the color channel fragments on the WVM it looks worse than eyeballing it, but then I feel I'm not really "getting it" by not using the WVM as my CC yardstick. What do people look for and what do they not worry about?
2) In regards to the vectorscope? What is bad? – Like, some images are going to naturally have more color in one direction (e.g. a blue sky is...blue). So what can one look for as a correction? For example, if I have my vectorscope leaning into Cyan more than anything is that bad? What am I looking for?
3) What's a common order of ops in using Color Finesse for CC? – I would think overall Contrast and RGB gamma would be first, but when I set LEVELS in Color Finesse it shifts all my RGB channels off the scale. I adjust the pedestal but they're then chopped off. That just one example off where the process can go awry. So I was wondering section by section what is the more generic process to get started and practicing?
I know this is a lot at once, but this is a rather involved CC program for a beginner. As always, thanks a lot for any feedback or input.
Jeez, I'm not sure. I simply don't use Color Finesse enough to give you a straight answer. Every time I use it, I have to look at the manual again. The ol' memory just ain't what it used to be.
Sr. Promotion Producer
KCRG-TV (ABC) Cedar Rapids, IA
I've been playing with Color Finesse all day. I'm starting to get the hang of it...Some what.
The one thing I've run into some trouble with I think is how colors inversely affect one another. Like, I have the WVM all stretched out for each channel (RG and B). The luma is in a good spot, yet I can't seem to get cyan out of this picture no matter what I do.
The vectorscope is leaning towards the cyan, but...by playing with shadows, highlights etc. I just can't seem to get the clothes of the people to make a plain red without pushing the entire RGB pedestal back into the red as well.
This is the image:
I'm trying to figure out what technique is used when Master RGB controls don't cut it. I'm all tangled up in the shadows and highlights getting nowhere.
Thanks to you and anybody who has suggestions.
I took a quick crack at this to see what I could do.
I've found that curves are particularly useful for this kind of faded-photo situation, since the underlying problem is that you've lost most of the cyan dye from the print.
After some playing around I decided (with no real knowledge) that the guy on the left was wearing bluejeans, and the guy on the right was wearing gray slacks of some sort. So I started in the Color Finesse Full UI and selected the Curves tab. I clicked the gray point (middle) eyedropper in the lower left of the controls, then I clicked on a "middle gray" portion of the right-hand guy's pants. I chose his right (your left) leg.
The results were a definite improvement, but looked a small bit too green for my taste. So I grabbed the point in the green curve that Color Finesse inserted and dragged it slightly to the right to reduce the green in the shadows.
It looked better, but a bit too contrasty, so I clicked the Master curve and dragged a point slightly to the left to lighten the shadows and reduce the contrast.
This took a lot less time to do than to read. The results, along with my curve settings are in this screenshot:
There are a lot of other ways to attack this problem, and this isn't perfect. But the skin tones are reasonable which is the most important thing.
As for your original questions:
1) The WFM is there as a reference to let you know when things are starting to clip in the highlights and shadows and to see when there is an imbalance between the channels. My advice is to start off color correcting by eye, but making note of what the WFM looks like. Eventually you'll learn the relationships enough that you can do the inverse: correct by WFM and like the resulting image. Don't obsess about the WFM.
2) The VS is, again, just a reference tool. If you like the results visually, it doesn't matter what the VS looks like. But if the results look "off" and you can't figure out why, the VS can often show the direction you need to correct in. If the VS leans towards cyan and you don't like the cyan cast, then you can, for example, use the Hue Offset wheels in the HSL tab to drag the colors away from cyan to correct. You'll immediately see the VS display move in the same direction.
3) It depends. If the image is almost right, then the common advice is to set the levels first, then tweak the color to get the look you want. But when faced with an image like your photo, it's often best to try to cancel the color distortion first, and then tweak the levels once it's "close" to correct.
Hope that helps. A lot of learning color correction is just playing around to understand how the controls interact. It can be frustrating at first, but you'll soon learn to evaluate an image and know how to attack the problem.
It took me 0.2 sec. for my automatic white balance/shadow repair program. No guessing needed.
auto white balance program
That was a big help. I was able to get "relatively" close to what I was going for, but like you I had to leave the scopes and just play around with it to give me some kind of result.
However, instead of using curves I used the RGB tonewheels (shadows/mids/highlights). It helped me by inverse effect get the red to shine through on the far-right lady's dress. But if those are indeed blue jeans on the left-man's legs, then I needed to bring in more blue (less yellow) as they were rather grey. That would make sense though considering how yellow the room is.
I'll try using the eyedropper and curves next challenge I get.
As for the interface, I feel I was initially using the less correct RGB WVM, the one where they're all stacked on top of each other. The side by side view is much less confusing. That and my feel for the vectorscope is that it's more-a-less a "target gauge" where you want the colors to be in the middle, but not too small (dark).