Ah, I need to check the video. I don’t have the Anolis but everyone always claims its better.

I’m an idiot you’re right for some reason thought they had to stick with the frame grab.

I do think there are healthy browns in some of the Hammer gothic sand they do come across more at times than others for some reason. One day I’d love to drool over seeing an ib print.

If this were Indicator handling the issue we wouldn’t be talking. Their Revenge of Frankenstein is astounding.

(2021-01-24, 10:22 PM)captainsolo Wrote: Ah, I need to check the video. I don’t have the Anolis but everyone always claims its better.

I’m an idiot you’re right for some reason thought they had to stick with the frame grab.

I do think there are healthy browns in some of the Hammer gothic sand they do come across more at times than others for some reason. One day I’d love to drool over seeing an ib print.

If this were Indicator handling the issue we wouldn’t be talking. Their Revenge of Frankenstein is astounding.

I have some excellent news for you which I've left you a PM about...

@pipefan413 thanks for making this thread. I know we've discussed the issues, and it really is a rabbit hole. Things seemed so much more simple when I made my custom a couple of years back! But, definitely think that release could be overhauled/upgraded if anyone wants to take a crack at it.

DE ANOLIS (2002 MASTER)

I really wanted to like this transfer but, after re-examining everything, I really don't... at least on the version available on the 2019 Studio Hamburg DE BD. It's got all the earmarks of an old transfer - magenta and overly bright. Even if it was from a newly struck IP at the time, they definitely brightened and magenta pushed this ones to accommodate CRT monitors.

To be honest, I'd be curious what this transfer looks like with @PDB doing the coloring pass he normally does to compensate for older transfers.

But, even though the brightness and the magenta cannot be corrected, what cannot be is the lack of any sort of visible grain field. This one really does look DNR'ed and any sort of filmic-ness has pretty much vanished. I know folks find this preferable because of the lack of the blue bias, but there's so much else wrong.

The 2016 DE BD might present an improvement. The fact that it's at 25fps can be easily corrected, but, looking at caps-a-holics, it doesn't look like it's got a tremendous amount of grain still. I'd be interested in seeing the Amazon Prime stream, but all these issues make it hard to recommend as a potential source for a preservation even if you color-correct and throw a grain plate...

BFI MASTER (2007; with additional footage in 2012)

I have to confess that I don't actually hate this transfer that much (sorry @captainsolo) On the plus sides, it is filmic with a pleasing amount of grain and no excessive DNR/EE. The contrast levels are pretty good too; black level is healthy and not crushed, plus it does look genuinely filmic, not an overly bright, dated transfer.

But, yeah, the color...

I'm not totally dismissive of this. Is it confirmed that their check print was the 2002 WB IP? Because, yeah, that would be absurd, but it really wouldn't be so absurd if they were referencing a different source (possibly from the BFI's own archive). So, I wouldn't call them liars off of that. That being said, it does seem to have a blue-bias which is not represented in other transfers of Jack Asher Hammer films.

As one can see with my preservation, a single-pass color correction to remove the blue tint only goes so far. You'd need to do a lot of scene by scene color adjustments to look like the old WB transfer or the US Warner BD. But, such an adjustment certainly is quite possible, even working from the finished BD. Time-consuming work but this probably still presents the best source for any potential preservation.

US WARNER BD

It really is deeply unfortunate that the US Warner Archive BD has bad black levels. The more I see it, the more I'm 100% in agreement that this was a technical glitch, and I'm pretty surprised it hasn't been recalled/had a replacement program. Their CURSE OF FRANKENSTEIN is excellent, so they can really produce good work.

Putting aside the levels issue (which, again, really is a big deal and makes it quite unwatchable IMO), the color looks pretty dead-on, minus any shenanigans with the titles not being brownish.

Interesting speculation about it being taken from an earlier version of BFI's work or at very least the original coloring files (DaVinci, Scratch Assimilate, etc). Totally possible but, with a dedicated amount of time invested in color correction, it is possible they were correcting the REC. 709 master pre-H264 compression (as said before).

Regardless, this remains a pretty useless source for a preservation starting point. I experimented trying to overlay the blacks from the BFI master to replace the crushed detail but because of some framing discrepancies, that is tricky to say the least. If someone manages a solve for that, then certainly this door could be reopened but until then...

JAPANESE REELS

I've previously erroneously said that these might be a good color source (apologies @pipefan413). Well, I was 100% wrong! Certainly a lot of red fading. Useful for seeing how tight the blacks can get in a print but that's about it.

(2021-01-25, 05:18 PM)The Aluminum Falcon Wrote: DE ANOLIS (2002 MASTER)

I really wanted to like this transfer but, after re-examining everything, I really don't... at least on the version available on the 2019 Studio Hamburg DE BD. It's got all the earmarks of an old transfer - magenta and overly bright. Even if it was from a newly struck IP at the time, they definitely brightened and magenta pushed this ones to accommodate CRT monitors.

I think you're getting muddled here (understandably because this is all a big bloody mess). I noticed a few times that you've called the 2019 release an "Anolis" one, and here you seem to be talking about the 2019 release again since you say it's the 2002 master, but the 2019 one is a Studio Hamburg release that doesn't seem to have anything to do with Anolis. Studio Hamburg has consistently released the 2002 master but Anolis released the 2007 master back in 2017 (and further extended the 2012 version of it), and *that* release mentions Studio Hamburg, presumably just because it was licenced to Anolis by them or something. But the reverse does not occur: Anolis isn't mentioned anywhere on either of Studio Hamburg's releases (2016 or 2019).

So, recap:

• ????, US, web stream: streaming version of 2002 master, US titles, very little to no DNR/cleanup (hooray!) - runs at 24/1.001 fps
  
• 2013, GB, Lionsgate: 2007 master, 2007 cut and 2012 extended cut, both UK titles, only 2007 cut has Universal logos - runs at 24 fps
  
• 2016, DE, Studio Hamburg (only one I don't have yet): 2002 master, US titles - runs at 25 fps
  
• 2017, DE, Anolis (additionally has Studio Hamburg logo on back): 2007 master, with 2007 cut and newly extended 2017 Anolis cut with dumb sound editing, both UK titles but again only the 2007 cut has Universal logos - runs at 24 fps
  
• 2018, US, Warner: 2007 master with colour fixed but levels totally ruined, 2007 cut only, UK titles and Universal logos - runs at 24/1.001 fps
  
• 2019, DE, Studio Hamburg: 2002 master again, with more DNR *and* somehow also more magenta push than the streaming version (like, surely if you were doing DNR, you'd *fix* the magenta push instead of making it worse?!) - runs at 24/1.001 fps
  

(Edited to add frame rates since they're all horribly inconsistent as well!)

There's also the 2020 Spanish one which unfortunately seems to be released by Resen, who (for those who aren't familiar) are bootlegging scumbags who don't really deserve my money, imo.

But yes, I agree with everything you said, if you were talking about the 2019 Studio Hamburg release as I think you were: their encode is cleaned up and more magenta, which is why I'm saying the Prime version is better. Still a little magenta but nowhere near as noticeable. And has a bunch of flecks that aren't there on the SH version, so SH definitely did some cleanup, for better or worse.

(2021-01-25, 05:18 PM)The Aluminum Falcon Wrote: To be honest, I'd be curious what this transfer looks like with @PDB doing the coloring pass he normally does to compensate for older transfers.

I mean, by all means I'd support that if @PDB wants, I'd just say don't use the Studio Hamburg encode, use the streaming version instead (PM me obv) because...

(2021-01-25, 05:18 PM)The Aluminum Falcon Wrote: But, even though the brightness and the magenta cannot be corrected, what cannot be is the lack of any sort of visible grain field. This one really does look DNR'ed and any sort of filmic-ness has pretty much vanished. I know folks find this preferable because of the lack of the blue bias, but there's so much else wrong.

... of this. Like I said, it's not as bad on the streaming version of the master, which has visibly less cleanup done. It's obviously lower bitrate though. CBR is at 10 Mbps, VBR is around 13.8 Mbps, and yet both still generally look better than the 2019 SH BD encode. I'm quite sure neither is likely to be totally free of artefacts, I haven't watched either the whole way through under a microscope or whatever... but there's definitely stuff lost on the SH one that's present on the streaming version.

(2021-01-25, 05:18 PM)The Aluminum Falcon Wrote: The 2016 DE BD might present an improvement. The fact that it's at 25fps can be easily corrected, but, looking at caps-a-holics, it doesn't look like it's got a tremendous amount of grain still.

I too would like to check it but it's extremely expensive. Like, $93, roughly. I mean... really. Honestly, I paid almost that for the Anolis one already, but at least that one was definitely unique in terms of having the extra runtime (I wanted to see how they handled grading and sound editing). I think the 2016 SH one is basically another encode of the 2002 master with potentially quite small differences in colour grading and possibly fidelity/detail, which is not without merit, but I can't afford to throw $93 at it all on my own. I'd definitely contribute a chunk of it, but I can't afford the entire asking price and the sellers aren't likely to budge very far because it's long OOP and both of the ones I found are still sealed.

If anybody else is also curious and wants to help feed my curiosity with cash, I could buy it with some financial help. I'd obviously report back with comparisons and whatever I've gleaned from looking at it against the other transfers of the same master. PM me if interested.

Would be good to be sure because I have literally every other release bar this one, just don't get hopes set too high in case it does turn out to be nonsense.

(2021-01-25, 05:18 PM)The Aluminum Falcon Wrote: BFI MASTER (2007; with additional footage in 2012)

I have to confess that I don't actually hate this transfer that much (sorry @captainsolo) On the plus sides, it is filmic with a pleasing amount of grain and no excessive DNR/EE. The contrast levels are pretty good too; black level is healthy and not crushed, plus it does look genuinely filmic, not an overly bright, dated transfer.

But, yeah, the color...

Agreed. I don't know if folk are getting mixed up because there's so much flying around here but I don't agree that the BFI one has crushed blacks, there's blatantly a huge amount of detail lost in the Warner encode of it that again I'm convinced is just a dumb levels mistake. It's pretty clear in the next comparison video but I haven't uploaded yet because I'm waiting for 500 MB Vimeo upload limit to refresh, so it'll be a few days away.

(2021-01-25, 05:18 PM)The Aluminum Falcon Wrote: I'm not totally dismissive of this. Is it confirmed that their check print was the 2002 WB IP?

Yes, according to the BFI's head of image quality, who specifically cites that. It's on the UK BD extras.

(2021-01-25, 05:18 PM)The Aluminum Falcon Wrote: Because, yeah, that would be absurd

Indeed it is!

(2021-01-25, 05:18 PM)The Aluminum Falcon Wrote: That being said, it does seem to have a blue-bias which is not represented in other transfers of Jack Asher Hammer films.

... Or in literally any other transfer of this film that has ever been made, including the early 90s LaserDisc ones made by Warner that look nothing like the 2002 Warner master (or at least not very close, but certainly more similar than either is to the BFI's blue tinted monstrosity). Yeah, I honestly don't think they got it right here and not using reference just seems plain silly to me. He claims to have graded it to "what he saw in that check print" (which he'd just cited as the 2002 WB IP) and yet Warner's own release of that looks nothing like this. So did Warner totally change the colours because they thought they were dumb? Or did the BFI totally change them for... some unknown reason? We can't really know without scanning an unfaded print, but that dark blue push isn't on either LaserDisc and I'm pretty damn sure it wasn't on the VHS either. So either Warner has been consistently getting it wrong for *years* despite scanning multiple different sources and having access to just about every conceivable reference one could dream of having on hand, or the BFI just made a tit of it and then nobody else fixed it until Warner tried but made their own mistake with the levels in 2018. What a mess, man.

(2021-01-25, 05:18 PM)The Aluminum Falcon Wrote: As one can see with my preservation, a single-pass color correction to remove the blue tint only goes so far. You'd need to do a lot of scene by scene color adjustments to look like the old WB transfer or the US Warner BD. But, such an adjustment certainly is quite possible, even working from the finished BD. Time-consuming work but this probably still presents the best source for any potential preservation.

Yeah, there would be merit in that, imo. I would try it myself but honestly with the push being *so* extreme I'm not convinced that my limited colour grading abilities would get it to where I need it to be. I might try pissing about with @TomArrow and Dre's tools to see how far that takes me. The good news though is that it isn't the *entire* film that's f***ed, although it is probably most of it; you could likely leave some scenes completely alone, use one correction for the majority of the scenes (green push), then another single adjustment / LUT for the dark blue push (mostly just the end).

(2021-01-25, 05:18 PM)The Aluminum Falcon Wrote: US WARNER BD

It really is deeply unfortunate that the US Warner Archive BD has bad black levels. The more I see it, the more I'm 100% in agreement that this was a technical glitch, and I'm pretty surprised it hasn't been recalled/had a replacement program. Their CURSE OF FRANKENSTEIN is excellent, so they can really produce good work.

Putting aside the levels issue (which, again, really is a big deal and makes it quite unwatchable IMO), the color looks pretty dead-on, minus any shenanigans with the titles not being brownish.

Interesting speculation about it being taken from an earlier version of BFI's work or at very least the original coloring files (DaVinci, Scratch Assimilate, etc). Totally possible but, with a dedicated amount of time invested in color correction, it is possible they were correcting the REC. 709 master pre-H264 compression (as said before).

Aye. I mean I have no idea, I'm just really impressed with the grading and I kinda can't see how they could achieve results that good from a foundation that awful, hence me wondering out loud whether they maybe had a pre-blue-tint version of the master. Maybe I'm just underestimating Warner's grading abilities. But the levels thing is bloody awful, and I'm convinced it's a technical mistake rather than any kind of intentional grading decision. It isn't just black crush it's also everything else that comes with screwing up levels: highlights also blown, colours all shunted into stupid levels of saturation.

(2021-01-25, 05:18 PM)The Aluminum Falcon Wrote: Regardless, this remains a pretty useless source for a preservation starting point. I experimented trying to overlay the blacks from the BFI master to replace the crushed detail but because of some framing discrepancies, that is tricky to say the least. If someone manages a solve for that, then certainly this door could be reopened but until then...

Aye, I've definitely thought about AutoOverlay but having not yet used it, I don't know how that would work when the levels are totally different: the problem is that if I'm understanding it right, the default behaviour is to overlay one image onto another and change the colours of the first one to the one it's being overlaid onto, but here, we'd want to take colour information from the WAC, missing detail from the BFI, but also generate missing colour information that should be in the WAC from the missing detail in the BFI... I don't honestly know if that would work at all.

(2021-01-25, 05:18 PM)The Aluminum Falcon Wrote: JAPANESE REELS

I've previously erroneously said that these might be a good color source (apologies @pipefan413). Well, I was 100% wrong! Certainly a lot of red fading. Useful for seeing how tight the blacks can get in a print but that's about it.

Well, I mean there's definitely fading, but it isn't necessarily a total writeoff, I guess. It does however serve as yet another piece of evidence to discredit the BFI's grading. If it was simply red fading removing that blue, I'd expect the thing to be completely pink in order for us not to be seeing any real trace of the extremely oppressive dark blue that's in the BFI version, but we don't see that: it's not *terribly* unbalanced, it's just a bit brown.

Having a fiddle in Resolve and getting depressed. Honestly, I don't think this 2007 mess is salvageable. It can certainly be improved, but it cannot be fixed.

I previously said that the Warner Archive Collection more or less nailed the colour, but I hadn't looked very closely at it yet. I'm now starting to do so. It ain't as great as I thought, so I think I was getting overexcited when I said they might've gone back to a pre-grading master. Clearly they've just graded on top of the crap job the BFI did.

Just to really hammer that home, let me say before I show you these that I did not look at this frame from the Warner regrade until *after* I'd tried to fix it myself for a while. And yet you'll see that my own results are not a million miles away from the actual 2018 Warner disc, though I think my skin tone is a little better than theirs (and my background consequently *less* balanced, mine's more pushed into green). So yeah, I'm now pretty convinced that Warner just took the BFI regrade and tried to fix it as well as they could.

1993 JP Warner LD:


US Prime Video stream, 2002 Warner master:


2013 GB Lionsgate BD, 2007 BFI master:


@The Aluminum Falcon's regrade of the BFI master (N.B. single correction for whole film, not specifically for this one shot):


2018 US Warner BD, 2007 BFI master:


Current work in progress after a bit of fiddling in Resolve (2007 BFI version regraded):




2007 BFI master (the one I regraded) vs my current WIP:


2018 US regrade (which I didn't look at until after I did mine) vs my current WIP:


I don't doubt that somebody might be able to pull this closer to where it should be, but it just seems like half the colour information has been wiped out by the dark blue wash. In other scenes, there is little to no shadow detail lost, no black crush to speak of. In THIS sequence (the most problematic one imo), there is an absolute crapload of data just completely gone. I don't think building on top of the BFI master is a good solution.

Still, I think I at least got this one shot looking a bit better than previous versions of the BFI master (taking it in isolation; I haven't really checked the rest of the film with this single correction to see how other shots look).


-




Some conclusions drawn so far then...


BEST ENCODE OF THE 2002 WARNER MASTER

I'm saying it's the US streaming version, despite fairly low bitrate / file size (it would fit on a dual-layer DVD, despite being 1080p). Has not been cleaned up to the extent that the 2019 DE Studio Hamburg BD has, nor does it have as much of a distracting magenta push (though it still does have some magenta push).


BEST ENCODE OF THE 2007 BFI MASTER

Kind of a toss-up, they all have huge issues.

I'm sort of leaning toward the 2017 DE Anolis BD since it is structurally superior to the 2013 GB Lionsgate BD due to the fact that it includes 2 frames missing from 2013 BD, in both the 2007 and 2012 cuts of the film... but you need to remove the "full range" flag that's clearly not supposed to be there, otherwise the levels are knackered in the opposite direction the 2018 US Warner BD (washed out and greyish blacks, and so on).

The 2018 US Warner BD has the colours significantly improved, but they still aren't great, and they've introduced a new issue by crushing the hell out of shadow detail and blowing highlights whilst cranking the saturation way too much as well. One step forward, two steps back.


CORRECT COLOUR GRADING OF TITLE SEQUENCE

Will not really be determinable unless we scan a print that hasn't faded, so that almost certainly means a Technicolor imbibition print. But if I had to guess, I'd reckon that the more warm, browning grading of the 2007 restoration seen on the 2012 GB Lionsgate BD and 2017 DE Anolis BD are probably more correct to how it actually looked in theatres than the cooler (grey stone) grading of the 2018 US Warner BD.

Shot in the dark, did you try opening the clipped blu rays in Virtual Dub and interpreting as full range to see if the detail comes back? I think I experienced that a handful of times, tho I don't recall the exact place I experienced it. Basically it was encoded as full range YUV but the metadata had it as limited (as is the Blu Ray standard).

Also, see if you can find a 4K 10 bit version of either of the masters, for example from online streams. Sometimes those have more shadow detail. But of course that's assuming that the master wasn't busted to begin with.

(2021-01-27, 04:50 AM)TomArrow Wrote: Shot in the dark, did you try opening the clipped blu rays in Virtual Dub and interpreting as full range to see if the detail comes back? I think I experienced that a handful of times, tho I don't recall the exact place I experienced it. Basically it was encoded as full range YUV but the metadata had it as limited (as is the Blu Ray standard).

I did! It doesn't. I actually show what happens in the next comparison video, which I'll upload once my Vimeo 500 MB upload allowance resets in a few days or so (I'm skint, hahah).

That's exactly what Anolis did on their release except in reverse: it was encoded as limited but flagged as full. Whereas the Warner Archive Collection release is encoded as limited and the metadata says that it's limited, but it looks like it was actually full range clipped to limited, thus chopping off all the low- and high-frequency detail and compressing the colour so it all looks oversaturated. Desaturating the colour by resampling/decoding it as full range does actually improve it in colour terms imo but it also means that blacks turn grey (like the Anolis release is on the disc) except without restoring any of the actual blacks that appear to have been clipped off. You with me?

(2021-01-27, 04:50 AM)TomArrow Wrote: Also, see if you can find a 4K 10 bit version of either of the masters, for example from online streams. Sometimes those have more shadow detail. But of course that's assuming that the master wasn't busted to begin with.

I'm not great with streaming services because 1. I don't use them and 2. I'm not in the US so I can't even sign up to half of them without a US payment method and/or address. I did try to find it elsewhere but yeah, no dice.

The streaming master appears to be consistently using the 2002 Warner master (which is in some ways much better but in others definitely worse, e.g. cropping and it's got too much magenta and is way too bright/pale). As I mentioned, the Amazon version actually looks better to some extent than the Blu-ray releases of that master. Bitrate is between 10 and 14 Mbps roughly, without going into too much detail. But I've not found any yet that use any version of the BFI master, Warner's version or otherwise.

Here's the same frame from 3 different encodes of the Warner master, and 1 of the BFI master, to show you some of the differences between them in isolation. I am not going to tell you what the sources are. See if you can figure it out from what I've said about them so far! I'll include them below in the body of the post as well but you can probably compare them better by downloading them and flipping back and forth or by using @williarob's site: http://www.framecompare.com/image-compar...n/KGW6GNNX

Encode 1:


Encode 2:


Encode 3:


Encode 4:

Hmm... well at the risk of making a complete fool of myself (which I probably will)!

RANK:

1 (best): Encode 1
2: Encode 2
3: Encode 3

I'd guess that Encode 3 is from the latest German BD since it's DNR'ed a little, then Encode 2 is the earlier German BD from 2016 (the one that was 25fps). Encode 1 with slightly different framing is the Amazon rip?

Encode 1 and Encode 2 are pretty competitive but I think Encode 1 resolves the grain slightly better in some areas. If I'm right about my guesses, that'd be weird though because I assume the Amazon rip would be lower bitrate than the first German BD...

(2021-02-13, 04:53 PM)The Aluminum Falcon Wrote: Hmm... well at the risk of making a complete fool of myself (which I probably will)!

RANK:

1 (best): Encode 1
2: Encode 2
3: Encode 3

I'd guess that Encode 3 is from the latest German BD since it's DNR'ed a little, then Encode 2 is the earlier German BD from 2016 (the one that was 25fps). Encode 1 with slightly different framing is the Amazon rip?

Encode 1 and Encode 2 are pretty competitive but I think Encode 1 resolves the grain slightly better in some areas. If I'm right about my guesses, that'd be weird though because I assume the Amazon rip would be lower bitrate than the first German BD...

DINGDINGDINGDINGDING

100% correct.

The Amazon bitrate is indeed significantly lower than the first Studio Hamburg Blu-ray and yet it very much holds its own.

The framing diff seems to just be less black horizontal padding on left but I've not yet tried to work out which is more correct. Obviously there's not much in it, couple of pixels or something.