[Tig] LUTs compared to grading linear

[Dan C. Tatut]

Dave

LUTs again... first of all you need to have a good understanding of what a LUT is (which I assume you must have by now). 

To correctly implement a facility-wide calibrated workflow you must keep one rule in mind: BE CONSISTENT AND COHERENT. Very important.

Then regarding log data and log workflows, the theory is quite simple:

When you scan film, or telecine transfer film in log mode, you end up with density measurements for that peice of film. 

Your goal while grading is to have onscreen colors that are as close as possible to the final output.

This means that you need to characterize the output media (output film characteristics, how target film responds to light that goes through it, sould it be laser or anything else)

Then there is the display device you use to grade. This device has its own characteristics even if it's a standard model (say a broadcast monitor). One thing to remember is that all the devices (electronic or not) tend to deviate from their original calibration. Even two devices of the same manufacturer/model will deviate differently as time goes by. You need to characterize them as often as possible (once a week would be a minimum). 

A LUT (either 1D or 3D) is a combination of the output film's reponse to the input light and the display device characteristics. This is a unique combination per display device. While you can have the characteristics for a specific stock (which can be shared across systems), the actual display device characteristics MUST be measured uniquely per device!!!

The actual "technology" used to combine the film output characteristics with the display device characteristics can be either a mathematical model, empirical (measures), by using your "accurate" eyes, etc.

What you must remember is be consistent when calibrating your machines (do not change your technique/method between display devices).

The last thing to remember is that a calibration will ALWAYS BE AN APPROXIMATION not a true matching, it's impossible because the technology used assumes samples of color spaces not an actual full description of the color spaces involved.

I've worked with some of the technologies available and also saw people calibrating their tools using their eyes perception of colors. Suprisingly, not using mathematical models can give some times extremely good results...

Best, Dan

Dan Tatut 
CEO 
CHROME Imaging 
105 Rue de Lyon 
CH-1203 Geneva 
Switzerland 

Phone: +41 22 807 23 60 
Fax: +41 22 807 23 70 
Mobile: +41 78 659 11 04 
WWW: http://www.chrome-imaging.com 

	-----Mensaje original----- 
	De: John David Hukom [mailto:judgedave at mac.com] 
	Enviado el: Dom 23/07/2006 03:29 p.m. 
	Para: tig at tig.colorist.org 
	CC: 
	Asunto: [Tig] LUTs compared to grading linear
	
	
	To the TIG

	I just want to understand the logic behind logarithmic transfers and figure out how to implement a facility wide workflow working in log space . 

	I have had countless discussions about the benefit of log and somehow on the scanning side I am convinced. It does keep most of the colors. (Well if your telecine is calibrated of  course)

	Assuming the telecine gives out calibrated scans in log then I am happy with getting log compared to getting a linear scan. I really get the latitude.

	Here comes the hard and confusing part. 

	Implementing a look up table. 

	Are there standard look up tables that I can just use ... I would think like broadcast standards there are standards LUTs.

	The reason for me looking for standard look up tables eventually I have to tell my clients categorically that what you will see is what most will see. I cannot tell them an arbitrary answer like " because we say so"

	Even if we have done internal testing "if that testing is not standard" it will still be a long argument .

	I would like to adhere to a standard LUT like 6500 is a standard. This is something I can implement facility wide.

	The main problem is that when i turn on the default LUTs that come with our grading system I dont like what I see. The colors are sh at tty and I cannot in my best smile tell the client "thats what they should expect"

	Currently I skip the LUT stage I scan in LOG then grade in linear straight away. I hope this is not a bad idea. If someone knows that I will end up in a dead end I really appreciate your input.

	Thanks



	=================
	Dave Hukom
	Chief Executive Officer

	Mobile +(63) 917 839 6179
	Tel +(63) 2 848 - POST
	Email judgedave at mac.com
	Fax. 893 7655

	Ignite Media Inc.
	"Bringing life back to film without using any."
	Im Post

	Ground Floor
	Port Royal Place
	118 Rada St. 
	Legaspi Village
	Makati City 1229
	Philippines