About color grading (and timing)

[spoRv]

From Wikipedia:

Color grading is the process of altering and enhancing the color of a motion picture, video image, or still image either electronically, photo-chemically or digitally. The photo-chemical process is also referred to as color timing and is typically performed at a photographic laboratory.

From now, let's use timing only for printed film, and grading for all the other cases.

The original negative has no timing at all, right? So, if a release is based on a film original negative scan, it *should* be regraded; at the contrary, if the master is taken from a release print, the timing could be preserved.

The former, if the intention is to have a color grading equal (or similar) to the one watched in the theaters, needs some kind of reference; a release print could be expected, that usually is not used as a master for various reasons - generation quality loss, used print often needs cleaning due to age, dirt, wear etc.

Of course, the whole process will be made in the digital domain; this *should* lead to a different result of the color timing, because the latter was made photochemically, and I think it's impossible to recreate perfectly those nuances - due to several reasons expert could surely explain better.

So, let's say that 100% is the right, original, color timing seen in the theaters at the day the movie was projected for the first time... let's say also studios are so lucky to find a first generation release print, projected maybe once, then stored in the proper conditions, the best ever found... now, given the right conditions - lamp color, best scanner etc. - it is supposed to obtain a very near copy, color wise, of the release print... let's say 95% - I think is impossible to better this result IMHO.

A negative scan will be better in resolution, but, after the proper, carefully chosen, scene by scene, shot by shot, digital color regrading, how close to the original color timing could it be? I mean, could be better, color wise, than the scan of a release print?

Frankly, I can't answer, but, according to what I read all around the net, the situation where the colors of some scenes are really close to the release print, while others are a bit, or a lot, way off, is due to the fact that it's not (always) possible to recreate digitally the photochemical process...

At the end, let's say that the best color grading of a negative scan could reach an overall 90% - and this will be a great result - but this is the average quality... in few scenes, it will be 100% like the release print, while in others will be 90%, 80% and so on, while the release print scan would have the quality consistent for the whole movie.

So, thinking about the techniques we, project makers, use for our project, it would be possible to obtain a very detailed version of the negative, with the color of the positive print... and it's quite easy... how?

Well, given the fact we have both O-negative and a very good positive prints, we should scan them at the best resolution possible - 8K will be great, but 6K should do; 4K is good, but 2K should be avoided; using the same scanner, they should be spatially aligned; so, we could take the negative luma, the positive chroma, and combine them, obtaining the same resolution of the negative, and a reduced chroma resolution of the positive, that will be more than enough for any BD (or Ultra-BD).

What do you think?

[guiser]

From Wikipedia:


So, let's say that 100% is the right, original, color timing seen in the theaters at the day the movie was projected for the first time... let's say also studios are so lucky to find a first generation release print, projected maybe once, then stored in the proper conditions, the best ever found... now, given the right conditions - lamp color, best scanner etc. - it is supposed to obtain a very near copy, color wise, of the release print... let's say 95% - I think is impossible to better this result IMHO.

Even in this case I wouldn't think a raw scan would look like the original, projected source. The main reason being dynamic range. Scanners are limited, so multiple scans must be made to preserve the entire range, then tradeoffs have to be made to get the final frame image back into the reduced color space of the format being used.

I'm no expert here and have barely any experience with video, but for photos it can certainly be quite difficult to get slide and even some printed images to look correct when scanned (and especially when re-printed again, even with a calibrated workflow). There was some discussion of this very subject over on ot (talking about 8 or 16mm I think), and I would certainly be interested in more info if people have it.

[spoRv]

Forgot to mention the following:
is the bulb the same used at the time of theatrical projection? if not, is it a perfect replica?
is EVERY hardware perfectly optimized/calibrated?
has EVERY person involved 100% perfect color vision?

and many other factors that don't come to mind (as usual) but I'm sure there are...

[CatBus]

I'm usually pretty satisfied with a clear intent to preserve theatrical colors.  It won't ever be perfect--compromises and mistakes will be made--but as long as the goal is to make it look like it did, it probably won't stray too far from the mark.

Asking for technical perfection is just too much--and frankly, in this revisionist world, an intent to preserve theatrical colors is a pretty tall order in its own right.

That said, apparently the guy behind the Do The Right Thing Blu-ray didn't know about the gold filter and really thought he was, er, well, doing the right thing.  So a single error can be really bad, and good intent has to be paired with some sort of quality control.

[PDB]

Agreed, there will always be a compromise trying to get the theatrical colors. Sometimes it simply boils down to the fact computers can't always get 100% the look of chemical timing. I know there can be a big discussion about bulb colors and print variations but even now in our digital world you can see a DCP and it will look different based off what projector you see it on (DLP, LCOS, Laser). But I think the desire and intention is what matters and if its found to be wrong, there is always version 2.

[Laserman]

From Wikipedia:
So, let's say that 100% is the right, original, color timing seen in the theaters at the day the movie was projected for the first time... let's say also studios are so lucky to find a first generation release print, projected maybe once, then stored in the proper conditions, the best ever found... now, given the right conditions - lamp color, best scanner etc. - it is supposed to obtain a very near copy, color wise, of the release print... let's say 95% - I think is impossible to better this result IMHO.

Even in this case I wouldn't think a raw scan would look like the original, projected source.  The main reason being dynamic range.  Scanners are limited, so multiple scans must be made to preserve the entire range, then tradeoffs have to be made to get the final frame image back into the reduced color space of the format being used.
I'm no expert here and have barely any experience with video, but for photos it can certainly be quite difficult to get slide and even some printed images to look correct when scanned (and especially when re-printed again, even with a calibrated workflow).  There was some discussion of this very subject over on ot (talking about 8 or 16mm I think), and I would certainly be interested in more info if people have it.

Forgot to mention the following:
is the bulb the same used at the time of theatrical projection? if not, is it a perfect replica?
is EVERY hardware perfectly optimized/calibrated?
has EVERY person involved 100% perfect color vision?
and many other factors that don't come to mind (as usual) but I'm sure there are...

If scanning on a newer scanner, ,and the ACES colour space, the entire dynamic range of the negative is easily captured via a triple flash process, and it is basically possible to completely recreate a chemical grade if you had limitless time and resources. The reality is of course that studios and workers only have a finite amount of time and expertise to spend on it, and don't always want the look of a chemically timed print, and its random fluctuations and sometimes undesired effects.
To answer other questions, projection lights varied a lot, there were Xenon lamps and Carbon Arc rods for example, which have a different spectral response, the size and composition of the screen itself, the age of the Xenon lamp, the reflector condition and the condition and age of the print all contribute to the colour of the projected image.
Also, LPP looks different to SP, looks different to IB Technicolor etc. and even within a print run the colour varies from print to print.
So although you could, conceivably recreate the look of a particular print, it would take so long as to be unfeasible, and not be desirable anyway as each print is different, and each venue would have the same print look different.
What you can do is work out if the prints you have were intended to be projected via Xenon or Carbon, look at multiple prints to look for consistency, and have an understanding and notes if possible from the shooting process. You then also need calibrated equipment that can display the target colour space correctly. You then need years of experience. Then you can do a very good grade that will match, for all intents and purposes, the original theatrical presentation.
Of course then it is shown in homes with large amounts of ambient light, uncalibrated displays and incorrect levels, but that is outside your control.
At least if the home viewer goes to the trouble to calibrate their viewing environment, it will be possible for them to get very close to the original presentation.

[spoRv]

...I'm waiting a long article about film scanning from you, now!