criterionforum.org

True. I don't understand how anyone can produce work like this and be truly happy with the color balance and the black levels. Which happen to be utterly terrible throughout the Serpico 4K restoration; I watched the Blu-ray last night. (At times I felt very closely reminded of the 'Manila in the Claws of Light' color scheme... restored by guess who.)

As it stands, there's just no way to convince me this was intentional as per the original cinematography. And it's not just a direct comparison to previous masters/restorations (like the MoC Blu-ray in the example I linked to) or claiming that previous gradings were too magenta. The picture of the new Serpico restoration stands out like a sore thumb to anyone who's not color-blind, to be a bit facetious.

I wonder more and more how things will change all the sudden in Europe the day 2 persons will retire.

mhofmann wrote: Wed Nov 25, 2020 6:06 pmAnd it's not just a direct comparison to previous masters/restorations (like the MoC Blu-ray in the example I linked to) or claiming that previous gradings were too magenta.

It never is. It's just trading an issue for a new one.

You can also add Alain Cavalier's Thérèse to the ECU.

that's sad to hear

Discussion of Le Cercle Rouge (Studio Canal 4K, Criterion,Studio Canal 2010), in French, with relevant quotes from Team Deakins podcast and links to captures:
https://www.dvdclassik.com/forum/viewto ... 22&t=39301

Stefan Andersson wrote: Fri Dec 04, 2020 5:02 pm Discussion of Le Cercle Rouge (Studio Canal 4K, Criterion,Studio Canal 2010), in French, with relevant quotes from Team Deakins podcast and links to captures:
https://www.dvdclassik.com/forum/viewto ... 22&t=39301

tenia, could you summarize your impressions of the Le Cercle Rouge UHD disc here for us?

I still have to see the BD. I'm not equipped in 4k and Canal mistakenly sent me the UHD screeners rather than the BD screeners so I still haven't been able to access it. Basee on what I saw, especially the BD caps from Svet in his review (and not his downconverted caps), I think it's warmer and a tad more colorful than what others would have done but I don't see anything horrifying like I saw on Serpico.

Based on the one set of DVDBeaver caps, my personal preference for color is the 2010 Studio Canal. It's the only one where flesh tones look right...but of course, maybe they're not supposed to look true to life?

tenia wrote: Sat Dec 05, 2020 7:50 pm I still have to see the BD. I'm not equipped in 4k and Canal mistakenly sent me the UHD screeners rather than the BD screeners so I still haven't been able to access it. Basee on what I saw, especially the BD caps from Svet in his review (and not his downconverted caps), I think it's warmer and a tad more colorful than what others would have done but I don't see anything horrifying like I saw on Serpico.

That was my impression too, looking at the caps - thanks!

jsteffe wrote: Sun Dec 06, 2020 4:13 pm That was my impression too, looking at the caps - thanks!

Similar opinion here. Having seen the Blu-ray, and ignoring any potential color grading backstory, Le Cercle Rouge looks a bit on the warmer, overall grey-blue-brownish side, but still reasonably balanced. Black levels are quite decent (unlike a disaster a la Serpico) and the white point isn't shifted into awkward territory (as on, say, La religieuse).

Listing updated with the Chabrol, Serpico and Thérèse.

Michael Kerpan wrote: Wed Nov 25, 2020 3:49 pm I honestly don't understand how such uniformly bad work could have become a de facto norm.

It's an agglomeration of competing drifts (in different directions) from 80 years of film stock aging, developing chemistry changes, technology changes (in projection bulbs, projection reflectors, projection stocks not to mention a ban on indoor cigarette smoking), with interesting localization variances across the globe--in addition to the measurable changes in color temperature that can be documented there is the additional confounding effect of the much briefer and recent 20 years of anchoring expectations by consumers--but the images which consumers are anchored to was heavily limited by the severe color limitations of standard definition and later rec. 709. There's are additional confounding factors to that anchoring effect in the generational preferences for magenta by home video colorists employed in the early 2000s not to mention the severe financial and time and equipment limitations they were operating under before the DVD boom and then again after the DVD crash.

Starting from the assumption that it's ever been correct on home video is going to lead to a lot of people calling something new incorrect. Starting from the assumption its correct on aged prints is equally problematic. Trying to correct previous home video looks is going to make some customer mad. failing to correct previous home video looks is going to make some customer mad.

Basically, the yellow effect could be calculated by the =~800K shift to blue that happened when the industry switched from carbon arc to xenon projection. but this would only apply to that portion of film history. and the contra color shift (back to yellow) to correct things being too blue for decades is being applied across the board.

The cyan bias on the other hand could be an attempt to correct the magenta bias of the early 2000s, and in solving that problem they've simply gone too far in the other direction.

And this is setting aside the projection and print replication problems that were created by the multiplex booms of the 80s and 90s. Dim images most saw in skinflint theatres are going to privilege the blue side of the spectrum, anchoring memories further away from yellow. Additionally, color replication and fidelity) was pretty poor for fifth and sixth generation prints most people saw in theatres (color quality went down with each subsequent geneation, but many generations were necessary given the logistical limitations of striking 2000-3000 prints for simultaneous domestic release).

Again, I recently saw vintage copies from movies from the 60s, 70s, 80s and 90s, and none looked like the Ritrovata color-signature, and even less so like Eclair ones.

Also, I'm quite certain not many people unhappy with these signatures are regretting the older magenta-pushed gradings. It's not a question of comparing these looks and saying now it's different so it must be bad. It's a matter of having traded a structural signature (magenta pushing every movie and handling contrast in a certain uniform way) for another one (elevated black levels and either a dull yellow drift or a steely blue one). The access to digital tools like LUTs makes me think this actually is the core issue : the over-use of similar (maybe even identical) LUTs from movie to movie within the same lab, leading to the list I've compiled.

It's the Laughing Cow process : no matter what you input, the output is always pretty much the same.

This also means that whether you have a reference print and a technical referent doesn't matter anymore : the lab used will lead 95% of the obtained color grading anyway. Alain Cavalier might not have supervised the grading of Thérèse since having the movie restored and graded at Ritrovata, Hiventy or VDM rather than Eclair would have made a much bigger difference than his supervision.

The fact is reference prints weren't more or less accurate 7 years ago, but neither Eclair or Ritrovata were grading movies this way. And they won't be more or less accurate once their colorists will retire, but we'll probably see the disappearance of these gradings with their retirements (except of course if they train their successors to do it the same way).

movielocke wrote: Tue Dec 08, 2020 4:52 amAnd this is setting aside the projection and print replication problems that were created by the multiplex booms of the 80s and 90s. Dim images most saw in skinflint theatres are going to privilege the blue side of the spectrum, anchoring memories further away from yellow. Additionally, color replication and fidelity was pretty poor for fifth and sixth generation prints most people saw in theatres (color quality went down with each subsequent geneation, but many generations were necessary given the logistical limitations of striking 2000-3000 prints for simultaneous domestic release).

Would prints of very different movies be poorly replicated and/or fading in very similar fashion, leading to differnt protographies now looking very similar ? Because that's what we're getting here : no matter what you input, the output is always about the same, while I doubt random issues in print replications or in fading prints would all lead to a similar look.

Remember also that everything we see from Ritrovata or Eclair could be completely mixed up if the rightholder would have just chosen the other lab for the restoration : Thérèse now looks the Eclair way, but they could very much look the Bruce Lee movies if the rightholder had chosen Ritrovata for the restoration and grading instead.

I understand where you're coming from with the reminders about the upstream variabilities, but downstream, the results are extremely bound to the lab chosen for the grading, so anything we see from these lab and deem "correct" is pretty much a matter of chance.

I'll try one day on a few select examples to show how a given movie would look if graded by the different labs (Eclair and Ritrovata in particular, but MPI and Paramount too would be neat, but I'm not an expert in Photoshop so it won't be easy - if somebody more expert there want to try, I'll be more than happy to supply screencaps).

Yes the LUTs are the central problem, to some extent the LUTs are trying to correct things like the blue shift with xenon projection. And if you're watching vintage prints with a xenon bulb and reflector, you're getting about 800K more blue than the print was timed for.

iirc Ritrovita applies a LUT for their film-out deliverable but doesn't adjust LUTs for the deliverables for DCP, broadcast and home video color spaces, right? that always has struck me as a significant central problem.

This is my understanding too, though I'm not entirely sure. I however wonder how/if applying this LUT can uniformise all those movies gradings if they're different before the LUT appliance, ie if this mistake would explain the issue, or if it's not the reason for applying it but how it's applied that is the issue (for instance, using similar LUTs from movie to movie).

movielocke wrote: Fri Dec 11, 2020 9:35 pm iirc Ritrovita applies a LUT for their film-out deliverable but doesn't adjust LUTs for the deliverables for DCP, broadcast and home video color spaces, right? that always has struck me as a significant central problem.

I have been wondering about similar things, but then again, Ritrovata for example have had years by now to grab some of the home Blu-ray releases off a shelf and say "wait a minute - we care about our restorations and something doesn't look right here," and that certainly hasn't happened. So unfortunately there's not a high likelihood of an undiscovered mistake here anymore, and I by now assume this is exactly how they want their output to look on consumer formats.

The new Blu-ray release of Madame Claude has the very distinct Ritrovata look.

And Svet seemingly mixes up in his examples Eclair and Ritrovata, which have very different and distinct issues (though are probably part of the same larger kind of questionings).
At this point, he could have mentioned Fox Deluxe titles.

Looks like we can add Djibril Diop Mambéty's Hyenas to the ECU based on the Kino blu.

I'm not overly convinced by what I saw from Hyenas and it definitely has a certain Eclair touch in the way it handles contrast and yellows. However, it doesn't have the typical steely blues, including during low-lit scenes. I checked the tech details (there's a text panel at the end of the restoration) and it's not the same colorist than usual (Aude Humblet did Hyenas, instead of Bruno Patin). That might explain why it's not as heavy as usually while still retaining this certain touch.


On a side note : as mentioned elsewhere on the forum, Retro-HD has been down since March, and it turns out it's unlikely to come back up again. I'll hopefully be able to transfer to my new site the resources I linked to illustrate these examples, but in the meantime, most of these links are currently dead.

It looks like The Mattei Affair is steaming on Criterion Channel through a source clearly looking like a Ritrovata grading. I've added it to the list.
Also added : Just Jaeckin's Madame Claude.

Would people be able to list here the Fox movies concerned by the tealed color gradings ? Movies like Desk Set, Wild River, etc.

Edit : I might also need the support of someone with digital coloring skills good enough to try and reproduce the labs color signatures on a few select screenshots.

tenia wrote: Mon Apr 26, 2021 9:23 am Would people be able to list here the Fox movies concerned by the tealed color gradings ? Movies like Desk Set, Wild River, etc.

Edit : I might also need the support of someone with digital coloring skills good enough to try and reproduce the labs color signatures on a few select screenshots.

Just from what I own...(besides the two you mentioned):
The Best of Everything
Bigger Than Life
Black Widow (1954)
The Bravados
Broken Lance
Bus Stop
Fantastic Voyage
From the Terrace
Garden of Evil
Hilda Crane
House of Bamboo
The Inn of the Sixth Happiness
The King and I
Love Is a Many-Splendored Thing
Peyton Place
Violent Saturday (I have a suspicion this must of been one of their earlier attempts because it's not as "aggressive" as something like "Bigger Than Life")
Voyage to the Bottom of the Sea
Warlock (Twilight Time disc at least)

Hi fellow Criterionforumers. I wrote an article for MUBI Notebook that was published today. I did some research and tried to get to the bottom of just why there is so much variation in terms of quality when we see home video/film restorations today. I sort of set out to write a hit piece on Ritrovata, but through a number of conversations found the issue more interesting and nuanced than I was expecting.

Hope you read and enjoy. Thanks so much to our own James Steffan, Tenia, and David Mackenzie for speaking to me about the issue!

To follow this piece by Drucker, here's mine (in French), focused on Ritrovata and Eclair, but with a pattern analysis prism.
The idea isn't so much to rant against some labs' practices and the uniformised results than to simply try and demonstrate there is something definitely happening (1) (not that there's still any doubt in the case of Ritrovata, since this has been explicitly confirmed to me by the lab) and to try several of the usual nay-sayers hypothesis and demonstrate it's easy to debunk them (2). I tried to construct in the most analytical-minded effects-oriented way possible rather than trying to go into guessing what could be the exact technical reason (say, the excess use of LUTs), but I'm sure it'll catch some flak from some naysayers nevertheless, despite what I would argue to be a rather hefty quantity of focused examples.

EDIT : as a clarification for my bit in Drucker's piece : I can't check RGB values on my TV screen since I don't have a probe for this. However, I'm equipped on my computer to properly capture BD content and measure RGB values on these screencaps. What I meant by the RGB strange values visible on some BDs is that normal computer screencaps CAN (and often will) go down to 0 0 0 for pure blacks and up to 255 255 255 for pure whites. You should in any case at least have black bars (for anything not 1.78) being pure blacks at 0 0 0. However, several movies clearly at least have black levels capped, in these conditions, way above 0 0 0. One of the most egregious examples I've seen is Gaumont Un témoin dans la ville : despite black bars firmly standing at the proper 0 0 0, black levels within the movie's frame never go below 15-16, and often don't go below 25 or even 32 on dark night scenes. There's also Spotlight on a Murderer, whose abnormally elevated black levels were fixed by Arrow for their release. In such cases, it's hard to say if the issue solely stems from the disc's or the master file's encode, if it's down to a restoration choice or a combination of both, but in any case, this clearly is introduced during the recent digital work (and it has been confirmed to me by Eclair that they do choose to slightly raise black levels on some of their works, though they tend to do it on a case by case basis and have been dialing down how high they were raising them).
In at least one case, Elevator to the Gallows, it is however 100% coming from the BD's encode, and the issues were fixed on the German disc.
And in at least a few other cases, black bars aren't even pure blacks, but can be stuck at 4, 5 or 6 (depending on the movie), which isn't quite re-assuring encode-wise.

Finally, in the case of color movies, these elevated black levels WILL make the color shifts structural to the labs' signatures more visible, because these dark greys WILL absorb the color shift from the signatures. That's what you see in the first set of screencaptures of my article and which makes the Ritrovata signature so easy to spot in this kind of shots.
And in some cases, the digital technical text panels detailing the restoration workflow do show all this, the elevated black levels and/or the color shift. The technical text panel on Muriel is drifting in blue-green, for instance.

So while there might have been a poor phrasing on my side, the matter is real and impacting the fidelity to the original elements by introducing new digital issues. It can be hard to state exactly what is coming from the encode, the restoration deliberate choice to have elevated blacks, or a combination of both, but it is there. Of course, if anybody has the proper way to describe and label this matter, please let me know.

Here are some examples for B&W movies, followed by Ritrovata color examples. 1st text panel is from Criterion's BD of Muriel, 2nd screencap from Gaumont's BD of La menace. Caps 3 & 4 are from Gaumont's Un témoin dans la vile, caps 5, 6 & 7 from Gaumont's Plein feux sur l'assassin. Remaining caps are a mixture of Ritrovata restorations.


Note : I'm quite certain the issue is also capping white levels, but the effect is mostly visible on black levels, because milky blacks have a much bigger dulling effect that clipped whites, especially now that the industry moved away from industrially clipping highlights.