[Tig] Universal Color Metadata challenge

Frank Hellmann frank at opticalart.de
Wed Sep 5 10:13:12 BST 2007

Hi Guys!

This has often been discussed and there are some problems associated
with the current workflows. I don't think that there can be a "fits all"
kind of methology here.

Editing & DI:

For the more down to earth productions there is no chance to have their
video dailies/rushes done on a calibrated scanner/telecine. Most of the
time these are just run through which telecine is available at the lab
and at what look the operator seems to like that morning.  Afterwards
scans will be done in a calibrated logarithmic fashion and will give
completely different graded images to start from. In my opinion this is
what breaks the use of CDLs on editing systems and the transfer to of
them to DI systems today. For this to work right everybody needs to have
the same images to start from, which is usually prohibited by the budget
(most people just can't afford all their footage being scanned).

However, it will work if you stay digital (aka. HDCAM, D20, Viper
data/video shot) throughout your production.


Here we have at least the same images people work on, so this is not a
problem. A system to transfer grading informations needs a couple of
things to be taken into account:

- Units: How does a number of X steps brighter relate to systems that
use printer light points or systems that use video levels for grading?

- Timing: How are image clips organized on a timeline? Do we have
animated parameters?

- Operators Stacks: How are operators organized and in which order are
they applied? Simple example: Blur operator used to blur a mask operator
that is fed into a colorcorrector operator.

- Basic Functionality: What operators do we need as basic functions?
(Primary grades, Masks, Blur/Defocus...)

- Extended Functionality: What operators are not available in all
systems and have to be emulated via Masks or Difference Images? For
example: 3D-Keyer output saved as mask image. B-Spline drawn mask saved
as image. Special Grading Plugin saved as difference image.

- Image Post Processing: What 2D/3D-LUTs are applied for viewing?
Standard Format possible?

- Fileformat & Compression: In which container do we stuff this
information into? Especially masks and difference image sequences need
special consideration, as these tend to get very large.

As you see this will get very complex in a minute if we want to have a
"does it all" format. Also getting the informations in and out of
compositing applications might give headaches as well. Just imagine
Fusion producing a flow of 50+ nodes just to load a graded image
sequence in. Then you start working on the image (adding more
compositing nodes) and need to change one of the grading parameters. To
save that back out for use in the DI application will be really


Stu Maschwitz wrote:
> The Craig Fearing page:
> http://tig.colorist.org/wiki3/index.php/Craig_Fearing_Page
>  ====
> I have just posted an article on my blog about what I feel is a  
> tremendously important issue facing the film industry right now, and  
> it relates to all of you.  I would appreciate any thoughts or  
> feedback from this group.
> http://prolost.blogspot.com/2007/09/film-industry-is-broken.html
> -Stu
> --
> stu maschwitz
> www.theorphanage.com
> prolost.blogspot.com
> _______________________________________________
> No personal abuse; absolutely no advertising or marketing on the main TIG mailinglist.  Emergency
> tig contact address is rob (at) calarts.edu
> Tig mailing list
> Tig at tig.colorist.org
> http://tig.colorist.org/mailman/listinfo/tig

OPTICAL ART             Film & Special-Effects GmbH    Waterloohain 7a
Frank Hellmann          DI Supervisor                  22769 Hamburg
frank at opticalart.de     http://www.opticalart.de
Tel: ++49 40 5111051    Fax: ++49 40 43169199	
Sitz der Gesellschaft:  Hamburg, Amtsgericht Hamburg HRB 384 70
Geschäftsführer:        Christian Burgdorff, Harald Lehmann 

More information about the Tig mailing list