Nice tut and nice results, gaross!
Only question I have is why work in CMYK space, which is mostly used for printer ink output? Is there a disadvantage to working in native RGB, (other than the fact that us old graphics lab guys tend to think in CMY percentages rather than RGB log scale)?
It would be nice to see the "after" result without the split screen, because my eyes would naturally gravitate to the "white" wall over your left shoulder.
If I "was" to work in CMYK, I would leave M anchored where it is, lower C and raise Y accordingly. Then adjust K slightly if necessary to preserve or adjust the perceived luminosity. This preserves more native bit depth and reduces theoretical Q noise slightly, compared to raising (or lowering) all three color channels unilaterally. Same would be true in RGB space as well, although the effect on luminance would be reversed.
Again, excellent tutorial. I'm sure you understand that very few people actually have the color acuity and understanding to make fine subjective adjustments sensibly.
Avoiding over exposure and making sure to white balance saves a lot o grief in post.
I'm sure this will seem unorthodox to the purists in our midst, but I find colour curves is helpful in correcting problems in the edit bay. Sony colour correction is also useful for serious mistakes. But I have to admit I'm more or an eyeball guy than a scopes and waves type.
I do try to lean heavily on zebras, wave forms and focus assist while on location.
Same goes for watching levels and keeping an ear on audio.
In a pinch iZotope RX is useful for solving issues in post, especially BG noises, RF hum and so on.
But for video images colour curves is the first tool I reach for, and go from there.
Thanks for the tutorial, but I am far to old and stupid to digest and retain all that.
Guess for me colour is more of an art than a science.
Techies will argue against that, I know.
After doing so many sundown events, with the color temps change faster than a baby's diaper I had to figure out a way to standardized on an easy method of CC work. Every 5 minutes of footage I take a screen shot in HD, drag it into Photoshop CS5.5 and use simple Auto Tone, Contrast and Color as a starting point. Then I drag that corrected clip back into SVP and place it into a bin called Color Corrected with the clip labeled with the time. This typically gives a good baseline scope reading with the scopes and waveforms reading correctly. I love the scopes - they make all my footage TV perfect. Good blacks, blues and face tones.
I then use SVP Color Match. I am always impressed and the results are typically better than my guess, especially with face tones. It usually matches all the better cameras I use including the contrasty AVCHD hand held Lumix. You can still see the camera differences like lens variations but it works great for what I do.
Your Photoshop technique for skin tones supports something I was told by a DaVinci colorist back in the eighties that I always keep in mind. "Realistic skin tones are found in the blue Gamma knob." Your grab of the curves window in PS shows that's what you found also.
Even after white balancing, sometimes even shooting chip charts to align the color vectors, skin just doesn't quite look real in video (IMHO). After getting as close as possible with the CC wheels, opening up Color Curves and adjusting the Blue Gamma will eventually hit a sweet spot for me.
CMYK is only used if we want to get a good print.
This supposes you use a printer profile (ICC)
To do it, you print an IT8 sheet, measure all the boxes with a spectrophotometer, do your Icc, print again with color management enabled. Doing this wil ensure what you are seeing (RGB) is what you will get (CMYK for 4 colors, but you can also print with 6 colors (adding light cyan, light magenta) or more...)
Back to Vegas : it only uses RGB, no CMYK. Unless you use a CMYK still picture, no need to check your files with Photoshop.
You can use a color picker (I learned this word here) to adapt your method.
NB a CMYK still picture means a picture with a CMYK embedded profile.
That's a good and very useful tutorial with excellent results. Thank you.
I don't have Photoshop and don't think I'm prepared top lay out the money for it as I shall probably use it only rarely. I do agree, however, that accurate skin tones are of vital importance and that, if these are right, the rest of the image will probably fall into place or at least seem more acceptable. I've always relied on White Balance, Colour Curves (using the blue line) and the FBmN Exposure Plug In and have found these able to deal with just about every problem I've faced. It also means working just within Vegas without having to import or export images and this saves time and effort.
"Switching the sample point in the Info window from RGB to read CMYK does not convert the workspace to CMYK."
Ah, that clears it up. I assumed your instructor was having you change the Image Mode to CMYK, where profiles and the other things mentioned would have come into play.
If you have time, play with the RGB primary adjustments instead. They correspond to Vegas' 0-225 (integer) log scale. Even though I spent 40 years working in cmyk percentages, RGB is much friendlier once you become acclimated. ;?)
Your instructor is obviously an old graphics print type, or has a military photo background (that's the only way it was taught back in the day).
5% up, 5% down. The readouts are the same; the corrections are different.
I'm not suggesting that anyone change what they are comfortable with; I worked in cmyk percentages for years before undertaking the transition to imaging.
Just that not everyone is aware that there are some very real differences (33% error in the first example).
I can't be bothered to go through the maths to sheck it, but my guess is that this boils down to a method of adjusting things so that your skin tones lie on the so-called "flesh line" on the vectorscope.
I keep trying but never fully understood how to use Vectorscopes, Histograms, Waveform, RGB Parade, color wheels, etc in Vegas. If there's a tutorial out there I'd love to see it. Most CC tuts I've see seem to be aimed at "movie looks", etc. not fleshtones.
Generally skin tones are around 60-80% on the waveform, but its really more about the look and what you are shooting. Clearly you do not want skin tones on someones forehead topping 100%, unless the character is on another planet and that is the look you are going for. That % is normal for well lit image or headshot. If your character is lurking in the shadows, the % will be closer to the murk.
The vectorscope has to do with color legality of broadcasting analog signals, today that equates to legal range for the codec you are working with. Since you are working from the codec this will not be out of range, since the codec has already recorded the information. Once you start correcting, you can push beyond the limits, and scope is suppose to indicate this. In transmission, scope is diagnostic tool where you send color bars from one end, and then adjust the receiving end to match bars. Vectorscope is somewhat old school, the RGB levels is the same view but more in a digital sense. RGB levels clipping the top indicate the image sensor or codec was not able to record more information in that color channel at the point. With a really dark image, it also indicates where in the image no information or detail was recorded. Either is fine as long if that is the intent of the look.
Scenario 1: I buy a jacket on eBay for $100. I decide to mark it up 50% and list it on Craigslist. It sits there for a couple of months. I decide to sell it at my cost, so I advertise it at 50% off and to my surprise, it sells the first day. Have I actually broken even, have I made money, or lost money on the deal? How much? What was my mistake?
Scenario 2: I buy a camcorder on eBay for $100. I decide to mark it up 50% and sell it on Craigslist, so I add $50 to the price. It doesn't sell, so once again I decide to sell it at my cost. I take $50 off my listing price, and it eventually sells. Did I do the right thing? Did I break even?
Back to color correction. Subtracting 5% is a 33% larger correction than adding 5%. This should have been the first thing your instructor explained if he was going to teach cmyk percentages. RGB integer and decimal corrections don't have that problem.
Math notwithstanding, you've got a heckuva set of eyes for color correcting. Not a common talent.
For someone who likes to keep it simple, however, you've made "some" things fairly complicated ;?)
Percentages are dependencies, as I demonstrated above. Your instructor either did not take these dependencies into account, or did not explain them well.
The result is that using the formula, the values of (M, Y) will vary geometrically with the value of C, presenting the user with a moving target as a starting point. The logic of that just doesn't resonate with me. Best. ;?)
The wall behind is supposed to be white. It's blue on all screenshots.
Only a mask could fix it ?
Les goûts et les couleurs...
Americans prefer blue dominant pictures. This explains the Kodak look
Europeans prefer red dominant pictures. This explains the Agfa look
Asian people prefer green dominant pictures. This explains the Fuji look
This is why Icc printing profiles are different in each country, we Europeans use Eurostandard...
These days I shoot a lot of auto white balance and with each setup or location I try to find something white or grey to shoot so I have something to use with they eye dropper white balance tool in post. It works very well.
Everyone is looking at the examples on their own monitor with their own eyes.
Once again, comparison is ridiculous without a common, stable reference.
Likewise, without a neutral reference, deciding whether a flesh tone is yellow or green, red or magenta, etc. is equally futile. A slight blue/cyan bias in the whites of the eyes (unjaundiced) can give one a clue in these situations.