Indulging in a rare cross-post because this seems more relevant here than in the original thread it grew out of...

I watched the UK release of this a while back (distributed by Second Sight) and was relatively impressed overall except that the transfer suffers from a really distracting "bump" at nearly every hard cut in the entire film, which initially confused the hell out of me, but I've since decided is just an artefact of the physical cutting of the final IP at 16 mm size and is therefore inherent to the scan, though they could've easily fixed it by stabilising it digitally.

I learned afterwards that Arrow actually had their own restoration (courtesy of MoMA) but it was never released here, presumably because Second Sight has the rights. My experience of Arrow had been overwhelmingly positive so I ended up importing a copy. I had a vague intention to reinsert a missing distributor logo from the Second Sight version or the original US release of the same transfer, but I've been distracted with hundreds of other things and only just decided I wanted to do that now. But I think having compared the three releases, I've now changed my mind, at least for the time being.

See, in the interim, I've realised that Arrow's Achilles' heel more often than not seems to be colour grading accuracy. This makes sense to me, because they very often scan negatives, rather than IPs; I guess they're not necessarily aiming to recreate the original colours but rather just make them look subjectively "good". I feel like this would be more or less a non-issue if they graded to reliable colour references but from looking at this film and others they've released, I'm not convinced that they always do.

What I've ended up doing is lining up all three versions and comparing them, as I've done with some other things (including, recently, DRACULA). Look at the difference!


I've put together a fairly brief comparison video to show some of the main differences between these, with more comparisons of the colour grading as screenshots in a gallery.

720p screenshot comparison gallery is here: https://postimg.cc/gallery/qtfvqC1/df77b742

Interesting, I just was thinking about how the Arrow Video version needed the opening distributor logo reinserted. Interesting how even the first 2 BDs have different colors considering they are the same transfer overall.

Fascinatingly, I would actually have to go with the belief that—for once—Arrow might have the more correct colors. Now I can't confirm this since I haven't seen a print of this movie, but the blue push in some of those shots do at least seem more "correct".

It is entirely possible that perhaps this restoration did make use of some answer print as color reference or something since the restoration was by MOMA which is different than Arrow's usual go-tos for negative restorations (and it would make sense for the Museum Of Modern Art to have access to more film elements than just the negatives alone).

(2021-02-05, 02:11 AM)LucasGodzilla Wrote: Fascinatingly, I would actually have to go with the belief that—for once—Arrow might have the more correct colors. Now I can't confirm this since I haven't seen a print of this movie, but the blue push in some of those shots do at least seem more "correct".

It is entirely possible that perhaps this restoration did make use of some answer print as color reference or something since the restoration was by MOMA which is different than Arrow's usual go-tos for negative restorations (and it would make sense for the Museum Of Modern Art to have access to more film elements than just the negatives alone).

Could be, but look at that scene where they arrange their "date". What the hell is with the skin tones? It's like they were grading to a badly faded print or something. Odd.

Bear in mind that the earlier ones are IP scans. My impression is that the Something Weird one is more or less what the IP actually looked like when scanned, slightly faded to magenta. Then Second Sight gets it and tries to correct for that, but then has to accentuate red back out and slightly overshot in places (credits are REALLY red, like a showroom TV with the contrast cranked to eleventy-stupid). Then Arrow manages to secure rights to MoMA preservation of negative materials and in doing so they fix the characteristic "16 mm bump" I've now seen on other 16 mm shot films but, based on the evidence, either didn't worry too much about using actually graded colour references or used references that were flawed, ending up with extremely muted colours generally (credits text now basically brown instead of bold red) and moments that seem overwhelmed with blues that aren't seen on either of those earlier transfers.

The BFI and Warner both clearly had access to reference material for DRACULA and yet they both keep screwing that one up too so i wouldn't put too much faith in MoMA theoretically having good reference (therefore their grading should be accurate). I might be wrong but it makes sense to me that colour might be the one significant flaw in the Arrow version. Well...that, and the missing logo.

(2021-02-05, 05:10 AM)pipefan413 Wrote:

(2021-02-05, 02:11 AM)LucasGodzilla Wrote: Fascinatingly, I would actually have to go with the belief that—for once—Arrow might have the more correct colors. Now I can't confirm this since I haven't seen a print of this movie, but the blue push in some of those shots do at least seem more "correct".

It is entirely possible that perhaps this restoration did make use of some answer print as color reference or something since the restoration was by MOMA which is different than Arrow's usual go-tos for negative restorations (and it would make sense for the Museum Of Modern Art to have access to more film elements than just the negatives alone).

Could be, but look at that scene where they arrange their "date". What the hell is with the skin tones? It's like they were grading to a badly faded print or something. Odd.

Bear in mind that the earlier ones are IP scans. My impression is that the Something Weird one is more or less what the IP actually looked like when scanned, slightly faded to magenta. Then Second Sight gets it and tries to correct for that, but then has to accentuate red back out and slightly overshot in places (credits are REALLY red, like a showroom TV with the contrast cranked to eleventy-stupid). Then Arrow manages to secure rights to MoMA preservation of negative materials and in doing so they fix the characteristic "16 mm bump" I've now seen on other 16 mm shot films but, based on the evidence, either didn't worry too much about using actually graded colour references or used references that were flawed, ending up with extremely muted colours generally (credits text now basically brown instead of bold red) and moments that seem overwhelmed with blues that aren't seen on either of those earlier transfers.

The BFI and Warner both clearly had access to reference material for DRACULA and yet they both keep screwing that one up too so i wouldn't put too much faith in MoMA theoretically having good reference (therefore their grading should be accurate). I might be wrong but it makes sense to me that colour might be the one significant flaw in the Arrow version. Well...that, and the missing logo.

I highly suspect that the terrible skin tones are mainly resultant of the original negatives being extremely faded. Judging from this clip from Arrow of outtake footage, the 16mm footage looks quite a fair bit faded.


So what I could possibly believe, was that somehow, the IP scan was simply less faded (possibly due to a mixture of storage and the positive film stock used) whilst the negatives were in a worse spot / condition by the time MOMA and Arrow got ahold of them.

Either that or perhaps it is as simple as an incompetent color grade job. I don't know. I wouldn't compare BFI nor Warner to MOMA since every company / studio is different in how they handle these sort of restorations though.

In regards to the prevalence of certain colors in certain shots and scenes being accurate or not is completely up in the air though. I would say an IP transfer should be a solid way of checking, but knowing some companies, it is possible that the photochemical color grading got "fixed" in the restoration. Unfortunately, there is no easy way of checking other than seeing SD transfers pre-dating any of these HD releases or getting ahold of an unfaded print of the movie SOMEHOW (I mean, it is theoretically possible given Kodak LPP became a thing the year this movie came out, so there may be the chance an LPP print is floating around out there).

(2021-02-05, 07:30 AM)LucasGodzilla Wrote: I highly suspect that the terrible skin tones are mainly resultant of the original negatives being extremely faded. Judging from this clip from Arrow of outtake footage, the 16mm footage looks quite a fair bit faded.

So what I could possibly believe, was that somehow, the IP scan was simply less faded (possibly due to a mixture of storage and the positive film stock used) whilst the negatives were in a worse spot / condition by the time MOMA and Arrow got ahold of them.

Aye, I reckon that's most likely it.

(2021-02-05, 07:30 AM)LucasGodzilla Wrote: Either that or perhaps it is as simple as an incompetent color grade job. I don't know. I wouldn't compare BFI nor Warner to MOMA since every company / studio is different in how they handle these sort of restorations though.

Obviously, but my point is that both MoMA and BFI *should* know what they're doing. BFI isn't some crappy Grindhouse print slinger, they're meant to be the UK's primary film archive / filmmaking initiative and in general have unprecedented access to reference materials compared to most. And yet...

So all I'm saying is just because MoMA *should* have loads of reference and *should* use it wisely, that doesn't mean that it's safe to assume they actually did. The evidence in the result suggests otherwise to me.

(2021-02-05, 07:30 AM)LucasGodzilla Wrote: In regards to the prevalence of certain colors in certain shots and scenes being accurate or not is completely up in the air though. I would say an IP transfer should be a solid way of checking, but knowing some companies, it is possible that the photochemical color grading got "fixed" in the restoration. Unfortunately, there is no easy way of checking other than seeing SD transfers pre-dating any of these HD releases or getting ahold of an unfaded print of the movie SOMEHOW (I mean, it is theoretically possible given Kodak LPP became a thing the year this movie came out, so there may be the chance an LPP print is floating around out there).

That's the trouble of course. I have no way to know if even Something Weird/Image also messed with the colour apart from intuition. But it does look, to me, like a more or less unf**kedwith scan of a somewhat red faded IP, so my previously given description of one potential chain of events wasn't completely guessed, I was basing it on what I saw.

Just like the BFI based their dark blue nightmare Dracula on "what they saw in the negative"... hahah.