Fw: [Tig] Color Grading in NLEs

Leffels acleffel
Thu May 8 16:31:59 BST 2003


----- Original Message ----- 
From: "Leffels" <acleffel at wideopenwest.com>
To: "Steve Hullfish" <steve.hullfish at bigidea.com>
Sent: Thursday, May 08, 2003 10:28 AM
Subject: Re: [Tig] Color Grading in NLEs


 I guess we agree to disagree. Curves as designed so that you can do one
 channel at a time. Even if you know what you are doing, you are still
 guessing as to the outcome. You can only use them in a point and click
 fashion. It would be far more valuable to have knobs or balls to manipulate
 curves interactively. The whole concept of using curves is also somewhat
 perverted ( not necessarily a negative thing ). By manipulating the curve
 you change the relationships of the channels from the original captured
 image. This may be fine for the specific moment in time you are looking at
 the image, but what if later you want to brighten, darken, contrast, or
stretch the values? You have now set a new un-predictable curve with
channels that won't necessarily track toward Whte or Black. You may think
 this is fine, but I don't. By using RGB manipulations, layered corrections,
 luminance manipulation, etc. - You can radically distort the image with
 predictable results. I'm not saying I don't use curves... I do use them. I
 use them to finessese an image. To open certain areas or create a wash, or
 cross color the blacks....but I do not use them as a mainstay of color
 correction and I would like to see systems that actually offer people tools
 they can wrap their heads around with some real fundamentals in color
 correction. The impulse with curves is for people to jump on, "whack" them
 around and stop when they see something they like. Believe me, there are
 plenty of sessions where I do the same thing. It's like the Davinci is a
roulette wheel and I spin til I win.... But for the average desktop user (
 see: non-colorist ) I don't think curves are intuitive, I don't think you
 learn anything by using them, I think they are sloppy, and I think they are
 inherently designed as a clunky click at a time interface.


> >You can affect the gamma knee and the gamma shoulder and the gain and
> > you can cause some pretty severe banding (or worse), but at least you
have
> the control if you >need it.  For color casts, I would agree that using
> Curves is much more difficult than using Hue Offset wheels (like the
> traditional
> > RGB balls).


 Again, these tools are designed for "subtle" manipulation. A little goes a
 long way. Try explaining that to your basic independant FCP user with a
 client that wants their DV to look like the Matrix. My basic worry for
 workstation color correction is that not enough time or effort is going
into
 ground - level, fundamental, functional color correction. I have Gamma Knee
 and Gamma Shoulder controls on a system I paid $450,000 for. I never use
 them. Ok, maybe a couple times in 4 years.


> > I don't use histograms much myself, but why would you dismiss having
MORE
> information about your image? Histograms are
> > traditionally the tools of the computer photo retoucher. These are
people
> that are your colleagues from the print side of
> > imaging. They have to do very complex color grading that is every bit
the
> equal of what a film/video colorist has to do -
> > minus all that pesky movement. I know that they rely on Histograms and
> Curves to do what they do.
>


 Um ... Hello Steve? That pesky "movement" is what we are talking about
here.
 We're talking about color correction for moving images, and they are far
different to work with than stills. Conceptually they share alot, but to
 interface is a whole different thing. BTW, I used to do re-touching as well
 as traditional photography. I dismiss having more information about the
 image because histograms are nothing more than a graph illustration of your
 image. If it was layed out as a wave form or vector scope sort of interface
 you could grab and move I might agree that it's useful for motion based
 applications. Again, I'm not saying people can't or don't use these
 things... I know they do. But there are much better ways of doing things
 that are far more intuitive, far more interactive, and quite frankly make
 far more sense as ways to manipulate an image. Have you ever tried to
 correct an image in let's say Photoshop by just manipulating the histogram?
 You might as well load up the remmington and pull the trigger. Those
 re-touchers you mention rely on curves and histograms because they've never
 been given better tools, and that's what they learned on. Shouldn't we as
 colorists with voices to the manufacturers -- try and put a stop to that
 nonsense... or at least point out the faults?


> > Colorists have had a pretty stable palette of tools and means of image
> analysis for the last decade or so, but that
> > doesn't mean that they're the only tools or means to analyze an image
that
> are useful.
>

 Nope  -- agreed. You can paint with many different brushes, or takes
 pictures with a plastic lens or a Hasselblad. It can all be art.. and can
 all come down to whatever the user likes most. However, there are a
 tremendous amount of people in the world who are going to start muddling
 with color correction for desktop projects... isn't now the time to give
 them some powerful tools?



> > Also, you mention color correcting on DS. I agree that you have a
problem
> there. Avid Symphony has solutions for the
> > problems you discussed of how to save corrections and the next version
of
> DS (Nitris) has taken almost all of the Symphony
> > color correction engine into its code. I don't know if you can save
> corrections in Nitris. You certainly can in Symphony.


 Even on the Symphony ( we have 6 of them) I'm not sure I agree there is a
 way for an editor to save corrections that they can use later with some
 degree of predictability. I know you can easily drop a correction onto all
 the shots that are like it... or actually pick which shots get the
 correction you have created, but I don't think there is a good image
 management system in place yet. What they haven't considered is what
happens
 whenyou get a whole bunch of new footage from a new location -- but it has
 some of the same problems that the job you worked on 2 weeks ago had?
 Wouldn't it be great to have a seperate image management bin where you
coulc
 see  before and after pictures of what the correction does? then you might
 actually be able to make an intellegent prediction... without having to
 remember what that correction you used last week actually consisted of on
 your own....


 Sorry to draw this out, but I really feel like this is the singular most
 important issue we face as colorists. It really bothers me to see people
 using tools in a cheap and sloppy way. The potential for that to happen is
 gigantic with the oncoming desktop world. Every Joe or Ginny who buys color
 correction software will be trying to sell it on their projects. Even your
 basic independent editorial house. The owner of a major editorial company
 with offices in LA, Chicago, New York - etc.. for instance, believes that
 all editors will be correcting their own work, and the colorist position
 will disappear. I don't think that's true, but let's at least push the
 manufacturers to put some decent, well thought out tools in their packages.


 Time to wake up....

 Yours in Twisting,


 Craig Leffel

 Optimus








More information about the Tig mailing list