BASKET CASE framing differences / re-stabilisation

[pipefan413]

Interesting! So the Image/Second Sight blu is mastered from a release print or IP that was a 35mm blowup of the 16mm source.

Well, apparently, no. It was supposedly from the negative (they talk about this on the bonus features of the Second Sight release). I have no idea why the Arrow release is missing the logo.

(...) Henenlotter and the Blu-ray producers went back to the original 16mm negatives and made the decision to present the film in its native aspect ratio, which restores quite a lot of the picture to the top and bottom of the frame.

https://www.blu-ray.com/movies/Basket-Ca...65/#Review

[SpaceBlackKnight]

Well, apparently, no. It was supposedly from the negative (they talk about this on the bonus features of the Second Sight release). I have no idea why the Arrow release is missing the logo.

(...) Henenlotter and the Blu-ray producers went back to the original 16mm negatives and made the decision to present the film in its native aspect ratio, which restores quite a lot of the picture to the top and bottom of the frame.

https://www.blu-ray.com/movies/Basket-Ca...65/#Review

For independent and non-studio productions, the OCN typically lacks original texts/credits as well as distributor and ratings logos. Those get added to internegs and release prints of various countries.

Much of the time, OCN restorations use the IP or Interneg for the original opening/credits and distributor logos. Sometimes, they utilize the OCN start to finish but also rebuild the credits/texts from scratch as the OCN typically lacks them, which explains the digitally redone credits on the 4k remaster of Beverly Hills Cop and some others.

[pipefan413]

Well, apparently, no. It was supposedly from the negative (they talk about this on the bonus features of the Second Sight release). I have no idea why the Arrow release is missing the logo.

(...) Henenlotter and the Blu-ray producers went back to the original 16mm negatives and made the decision to present the film in its native aspect ratio, which restores quite a lot of the picture to the top and bottom of the frame.

https://www.blu-ray.com/movies/Basket-Ca...65/#Review

For independent and non-studio productions, the OCN and sometimes the IP typically lack distributor and ratings logos. Those get added to internegs and release prints of various countries.

I guess the logo might be missing from the OCN but was reinstated from another element, and the Arrow release just didn't reinstate it?

Either way, hopefully easy enough to fix, although maybe not as straightforward as one would think because they seem to have subtly different framing / ratios. Might need to fire it through AviSynth to add borders or something to match the ratio, but I don't know if that would be really visually obvious because of the lack of grain etc. Possibly needs a grain plate or something too (over just the added black).

[crampedmisfit1990]

The Second Sight BD is the same remaster as the Something Weird BD (US), I was thinking... would you rather use that for this project? I ask because it seems to have better video than the Second Sight. I compared them years ago, as I own the US, UK and Arrow version on BD. If I recall correctly... the Second Sight had some artifacts the US one didn't.

[pipefan413]

The Second Sight BD is the same remaster as the Something Weird BD (US), I was thinking... would you rather use that as the base for this project? I ask because it seems to have better video than the Second Sight. I compared them years ago, as I own the US, UK and Arrow version on BD.

Oh, wait, really? I thought the Second Sight and Image / Something Weird releases were identical. Christ...

Well, in any case I would definitely be grateful for a chance to look at the Arrow one, but if you're saying the Image one is also different somehow then that might be worth a look too. I'm mostly interested to see if any of these don't have the irritating alignment issue on most hard cuts.

[crampedmisfit1990]

Oh sorry, I meant to imply the Image one has a better encode of the same transfer as the Second Sight.

But if I recall correctly... the Arrow doesn't have the alignment issues. Could be wrong tho. All I know is it looks 1000% better than the older BDs. I was blown away when I got it as I figured it looked as good as it ever would last time.

[pipefan413]

Oh sorry, I meant to imply the Image one has a better encode of the same transfer as the Second Sight.

But if I recall correctly... the Arrow doesn't have the alignment issues. Could be wrong tho. All I know is it looks 1000% better than the older BDs. I was blown away when I got it as I figured it looked as good as it ever would last time.

Aye, that's what I thought you were getting at: I just assumed they were the same encode. I suppose that's probably just because I'm getting used to seeing German distributors just lift existing encodes wholesale... guess that isn't common elsewhere.

[pipefan413]

Well, apparently, no. It was supposedly from the negative (they talk about this on the bonus features of the Second Sight release). I have no idea why the Arrow release is missing the logo.

(...) Henenlotter and the Blu-ray producers went back to the original 16mm negatives and made the decision to present the film in its native aspect ratio, which restores quite a lot of the picture to the top and bottom of the frame.

https://www.blu-ray.com/movies/Basket-Ca...65/#Review

For independent and non-studio productions, the OCN and sometimes the IP typically lack distributor and ratings logos. Those get added to internegs and release prints of various countries.

I guess the logo might be missing from the OCN but was reinstated from another element, and the Arrow release just didn't reinstate it?

Either way, hopefully easy enough to fix, although maybe not as straightforward as one would think because they seem to have subtly different framing / ratios. Might need to fire it through AviSynth to add borders or something to match the ratio, but I don't know if that would be really visually obvious because of the lack of grain etc. Possibly needs a grain plate or something too (over just the added black).

Oh, well, whaddayaknow... apparently the reviewer got it wrong. I'm sure I remembered them saying the same thing in the bonus features but I guess I misremembered.

[pipefan413]

I'm starting to wonder if perhaps a lot of my questions about both Basket Case and Dog Soldiers come back to a physical limitation of the 16 mm negative elements... both of these films are shot on 16 mm and both of them have issues with hard cuts "bumping" the frame at the cut. What I'm wondering is if this might be something to do with the much narrower spaces in between frames on 16 mm stock vs the bigger gaps between frames on 35 mm.

I was under the impression (I think mostly from the blu-ray.com review) that the Image Entertainment / Something Weird restoration was done from a scan of the original 16 mm negative, but apparently that isn't the case and it was actually from the 35 mm blowup interpositive. That would make more sense, given that the image is visibly less sharp than the later MOMA / Arrow restoration. The director is quoted in the blu-ray.com forum post as saying the following back in 2011 re. the Image restoration:

Found most of the 16mm negative and an incomplete 16mm print which have a different color timing to the night scenes which somehow got lost when the film was blown up to 35mm. Though we’re using the 35mm inter-positive for the transfer, our point of reference is the 16mm original. The result is that this version of BC is going to look a bit different than what we’re used to, but it will be as close as possible to the original 16mm version I shot way back when.

[...] A side-by-side comparison [of the original 16mm negative and the 35mm blowup] showed that the 35mm blowup was just as sharp. In fact, I couldn't tell the difference. However, as was the custom way back when, the 16mm negative was A&B roll which meant it would cost twice as much to transfer it and then it would have to be edited together. Worse, there were a couple of gaps in the negative for footage used in the sequels. It just made more sense to use the 35mm for the transfer but the 16mm for reference.

(source: blu-ray.com forum)

The reference to the 16 mm negative being "A and B roll" is explained over on Wikipedia as follows...

35 mm film was wide enough to hide splices, but 16 mm film revealed the splices as flaws in the picture. To avoid this problem, the intended shots were spliced to opaque black leader, with the black leader hiding the splice. Two sequences of shots were assembled, the odd-numbered shots on the A-roll, and the even-numbered shots on the B-roll, such that all of the shots on one roll were matched by black leader on the other roll, in a checkerboard pattern (an alternate name for the process was "checkerboard printing".) Unexposed 16 mm raw print stock was exposed twice, once to the A-roll, then it was exposed again to the B-roll.

(source: Wikipedia page on B-roll)

... which got me thinking about whether this might explain why most of the hard cuts bounce the frame. What I'm wondering is if it's something to do with the splices (which presumably is how a hard cut would present itself on a negative?) not quite being perfectly lined up. That said, it's weird that it seems to follow exactly the same pattern every time, with the frames just before and after the cut being misaligned, then the 2nd or 3rd frame after the cut being misaligned in the opposite direction, then the 4th or so being in the same alignment as the rest of the shot. I wonder if it's something to do with the sprocket holes not quite exactly lining up due to tiny ~1mm or so misalignment of splices so that the film stock isn't properly stabilised when passing through the scanner/projector? I'd love to hear from somebody with more actual experience of working with film, I'm only guessing here.

Anyway, since the older 2011 Image Entertainment scan is apparently sourced off a 35 mm interpositive blown up from the 16 mm negative, and the native aspect ratio on a 35 mm blowup is 1.33:1, based on the result I'm thinking that what must've happened here is that they preserved pretty much the entire height of the frames on that source, leaving no remaining vertical space to crop out in order to stabilise the misaligned frames by shifting the final framed image up or down by a few pixels to compensate. It's been very conservatively matted down to what appears to be a slightly odd ratio of ~1.3352:1 (1442 x 1080) on the 2011 (Image) and 2012 (Second Sight) BDs:



The Arrow image on the other hand appears to be slightly wider, pretty much exactly 1.37:1, and from this particular comparison looks like it manages that by preserving more of the picture both on the sides (especially the right) and on the top:



I suppose logically, this makes sense, since the overall native frame shape on a 35 mm film stock that the 2011 master was scanned from would be about 1.33:1 but with rounded corners that would need to be cropped out for a squared-off final Blu-ray Disc image (with the intended final projection resolution more commonly being 1.85:1, as Basket Case was indeed originally shown in theatres). Presumably the additional picture visible on the edges of the 1.37:1 16 mm negative would have been lost when creating the 35 mm blowup interpositive. The MOMA/Arrow restoration was instead scanned from the original negative elements, so had access to the additional picture area around the edges and presumably did the same and just cropped out the round corners (rather than matting to 1.85:1 as the original theatrical presentation was).

Anyway, this is good news for the overall picture, but bad news for the issue I described trying to eradicate: the Arrow release also has the really noticeable frame misalignment on hard cuts that you get on the earlier scan, presumably because although it does seem to have access to more vertical resolution, they've understandably chosen to preserve the maximum possible amount of picture rather than using the extra vertical resolution to manually crop every frame down and realign the hard cuts / splices. This is a pain in my arse, but does mean that I can take the Arrow release and crop in just enough to get the vertical space to fix the alignment issue without losing a huge amount of picture in the process (I'm guessing without actually testing it yet that the result would be similar framing to the Image release, though I'd probably stick to 1.37:1 to avoid cropping too much horizontally). Alternatively, I could do a "theatrical reconstruction" which goes back to the old 1.85:1 matte, I guess.

Something that's led me to the conclusions I have is that I've noticed the same issue (though not quite to the same extent) with another film I'm working on: the Koch Media release of Dog Soldiers. Basket Case and Dog Soldiers have something very important in common: they were both shot on a custom version of 16 mm film stock called Super 16.

Here's what normal 16 mm film and Super 16 look like, side by side:



To explain that in words, the difference is that Super 16 stock only has one set of perforations running up one side, rather than the usual symmetrical pair of single perforations running up both sides. The additional space recovered by getting rid of one set of perforations is used to record additional picture resolution.

By contrast, the much maligned Scream Factory 35 mm blowup print scan of Dog Soldiers crops out a chunk of both the right and top edges of the image for a final, decidedly non-theatrical aspect ratio of 1.78:1 (16:9), but it doesn't seem to suffer from this misalignment issue on the hard cuts, so I assume they must have used the extra vertical space gained by cropping in to adjust the vertical alignment. My guess is that unlike Basket Case (which evidently has a 35 mm blowup interpositive left at the maximum possible 1.33:1 aspect ratio on the film stock), Dog Soldiers may have potentially been hard-matted down in camera to 1.85:1 which would mean that no additional vertical resolution would have been available with which to stabilise the image, but cropping in to 1.78:1 made stabilisation possible. On the other hand, the newer Koch Media release that was scanned primarily from the actual Super 16 negative seems NOT to have been properly stabilised on the hard cuts / splices, presumably because they just matted it down to 1.85:1 at the exact centre of a 1.37:1 image. What that means for Dog Soldiers is that I can't re-stabilise the Koch release like I can the Basket Case release by Arrow without losing picture information that was intended to be seen in the first place, because Arrow's Basket Case is pretty much fully open matte but Koch's Dog Soldiers is matted to 1.85:1 (either in camera or at some stage later in the process), which is the intended projection aspect ratio. I'd need to crop the edges off in order to both keep the correct 1.85:1 ratio and vertically stabilise the picture.

So to summarise all that, here's my (highly unqualified and conjecture based) theory:

1. Since it appears to be known that both Basket Case and Dog Soldiers were shot on Super 16 stock, which has very little space between frames with which to disguise and align splices, I think that perhaps even the tiniest of errors (maybe less than 1 mm) might have resulted in the frame - and perhaps more importantly, the perforations / sprocket holes - being ever so slightly misaligned.
   
   
2. Since Super 16 stock also has only one set of perforations / sprocket holes running up one side of the image (instead of the more common symmetrical pair of one set up each side of the image), the aforementioned tiny inaccuracies with splices may have potentially resulted in a more noticeable amount of pulling / warping as the slightly misaligned perforations pass through the projector, or in this case, the scanner. This assumes the scanner even uses the sprocket holes, and I know some don't (using a sort of belt system instead of pulling the stock through the gate by its perforations), but it would certainly seem to be a logical explanation of what I'm seeing here with both of these films. Theoretically, the stabilisation done during the restoration process would mean that the sprocket holes would appear more or less aligned but the actual scanned frame of picture would be a few pixels shifted up or down, as we're seeing here. Also, unlike 35 mm stock, which is generally 4-perf (4 on each side for a total of 8 per frame), 16mm being smaller means it's only normally 1 perf up each side, meaning that Super 16 is only 1-perf on the one side... so any stabilisation done in software that looks at sprocket holes would only have 1 point of reference for every frame on a Super 16 negative, compared to 8 for every frame with 35 mm. Although, that wouldn't really explain why the same instability is visible on a scan that's ostensibly been done from a 35 mm blowup IP (the 2011 Basket Case scan).
   

EDIT: A friend pointed out that they were seeing the exact same frame instability at hard cuts in the mid-90s BBC TV miniseries version of Pride And Prejudice, which was shot in... you guessed it... Super 16. So I think I must be at least vaguely onto something here, even if I'm not quite getting it totally right.

[pipefan413]

GOOD NEWS: after checking several known problem sequences, the US-exclusive Arrow release *does not* appear to suffer from the jumpy cut problem I described. It's stable!

What I am thinking now is that the issue may have been introduced when transferring the OCN to IP, since the IP has the issue but the OCN apparently doesn't. I say that because at least so far, it would appear that Arrow (or rather, MOMA) has now simply stabilised the image by cropping in; they seem to have retained pretty much the entire picture off the OCN, which means that it must have been stable on the source in the first place, unlike the IP if the Image / Second Sight scan is anything to go by.

Now, this basically makes my originally intended project completely unnecessary, but I am nonetheless going to try a couple of different approaches for re-adding that logo at the start, which I'm very keen to do. After looking at the Arrow video, I'm a bit worried about re-encoding it given the massive amount of grain (I don't know how well that'll hold up to a re-encode, basically) so I might not do it the most obvious way by just encoding the whole lot to a video master grade format, editing them together and encoding back to AVC again. Depending on things like the locations of keyframes, I can potentially work around it by doing something slightly unusual like just splitting the video and audio at the same frames, then having the missing logo from either the Image or Second Sight release in its own separate (trimmed out but not re-encoded) video to add to a playlist, which makes it play before the (also trimmed but not re-encoded) Arrow video with the additional MOMA stuff at the start lopped off. That would be relatively easy to do, just some calculations and working out where keyframes lie in each video. That however means having to rely on playlists rather than a single MKV container (since I won't be able to join two MKV files from different sources with different encoding) so I'd probably be best to do it in a format that allows for it, like a BD folder structure / ISO or maybe AVCHD (as opposed to two separate MKV files with some separate playlist file you'd have to use).

Additionally, the framing of the Image / Second Sight release is different, so I'm not simply going to crop that off and stick it into the playlist structure because that would irritate me. That will have to be re-encoded, and the frame expanded slightly to achieve the exact same aspect ratio as the rest of the video it's to be appended to. I will most likely attempt this in AviSynth+ to start with, although depending on the results that may not be my final method (if you can see obvious borders that way then I'll need to try something more advanced).

ASIDE: I had a look at Dog Soldiers again as well, and I'm rather confused to discover that I can't find (so far anyway) any of those jumpy hard cut moments that are so incredibly obvious in the Image / Second Sight Basket Case. But I'm certain I saw claims that they were present on another forum, I didn't dream that; I guess I may have misread or misremembered and they were actually meaning the print scan (Scream Factory) but I was pretty certain they were complaining about the Koch, which is canned off OCN the same as Arrow Basket Case. Will update as I learn more and check through the files. For now, back to Snowpiercer!