A production cel is a sheet of acetate with paint on one side and ink on the other. The paint goes on in flat fields by hand, which means it has thickness, and thickness means the edge of a colour field sits very slightly above the sheet. Stack three or four of those under a camera, light them, and you are photographing a small piece of physical sculpture. Dust lands on them. The registration pegs hold each layer in roughly, not exactly, the same place.
All of that is in the negative. For thirty-three years almost nobody has seen it, because every format the film reached — VHS, laserdisc, DVD, a 1080p transfer — threw away more detail than the cels contained.
Ninja Scroll goes back into cinemas this week with that ceiling lifted.
What is actually shipping
HIDIVE and Iconic Events Releasing are putting a new 4K restoration of Yoshiaki Kawajiri’s 1993 film into roughly 600 screens across the United States and Canada, on three dates: 4 October subtitled, 5 October in the English dub, and a 7 October encore running whichever format the individual theatre picked. A 4K UHD SteelBook follows in early 2027. The restoration had its world premiere at the Berlinale back in February.
The spec, as announced: a 4K scan of the original 35mm negative at 3840x2160, then frame-by-frame damage repair and colour correction, producing a new archival master. Sentai produced it. Kawajiri supervised.
Scanning the negative is the whole decision
Film duplication is generational. Camera negative to interpositive to internegative to release print, and each hop costs resolution and adds contrast. A lot of nineties anime reissues were pulled off whatever element was nearest to hand, usually an interpositive, because the negative was in a vault in Tokyo and getting at it is a legal and logistical problem rather than a technical one.
You can see the difference without any equipment. A scan from a print has blacks that have already been crushed once and highlights that have already clipped once, and no amount of grading recovers what the duplication stage removed. A scan from the negative starts with the full latitude the cinematographer exposed for.
So “original 35mm negative” in a press release is the line worth reading twice. It means somebody did the unglamorous half of the job.
The pass that ruins these
Here is where cel-animation restorations go wrong, and they have gone wrong a lot. A 4K scan of a 1993 negative arrives full of things that look like faults: film grain, dust that was sitting on the cel in 1993, faint seams where one acetate layer overlaps another, a shimmer along painted edges as registration drifts a fraction of a millimetre between frames.
The temptation is to clean all of it. Digital noise reduction will take the grain out, and it will also take out the finest variation in an inked line — the place where the brush ran dry, the slight swell at the end of a stroke. Push it and faces go waxy, linework thins toward nothing, and flat paint flattens further into something that looks like vector fill. The film stops looking photographed and starts looking generated.
Which is the same argument this blog had about a generative pass over finished art from the other direction. Inventing plausible detail and recovering real detail are different operations that produce superficially similar pictures, and the second one is harder, slower and worth more. A restoration’s grain is evidence. Scrub it and you have a guess.
Dust is the judgement call nobody agrees on. Some of it is genuinely a fault — a hair in the gate, a scratch from a projector. Some of it was on the cel when the camera fired, which makes it part of the photographed image, and a frame with no dust in it never existed. I have sat through enough before-and-after reels to know that the aggressive clean always looks better in a ten-second comparison and worse over ninety minutes.
Why Kawajiri’s name on it matters
Supervision by the original director is not a courtesy credit on a job like this. Three or four passes in a restoration are pure taste, and the only person who can settle them is the one who signed off the original answer print.
Kawajiri’s look is specific and fragile. He builds shots out of hard-edged shadow shapes — a face half in black, a figure read entirely as silhouette against a lit wall — with a limited palette and an enormous amount of the frame given over to near-black. That is a style that lives or dies on where the shadow detail sits. Lift the blacks in a grade and the shapes go soft and grey. Crush them to look punchy on a modern display and the shapes weld together into one mass, which reads as cheap rather than dramatic.
A colourist working alone optimises for a waveform monitor. A director who remembers which blacks were meant to hold a little detail and which were meant to be a hole optimises for the film.
If you want the physical version of this argument, the Ghost in the Shell retrospective made the case better than any transfer can: the piece on what those hand-painted backgrounds look like up close is about standing in front of the artwork and seeing the brush decisions. A good 4K scan is the nearest a cinema gets to that.
One number to be slightly suspicious of
3840x2160 is UHD. It is the resolution a television wants. A 35mm frame scanned properly supports more than that — 4096 wide is the standard digital cinema figure, and scanner operators will go to 6K or 8K on an archival pass precisely so that the master outlives the current display generation.
Announcing 3840x2160 suggests the delivery target was the consumer release as much as the theatrical run. That is not a scandal and it is a long way above anything this film has had before. It does mean the SteelBook in early 2027 is probably the full resolution of the project rather than a reduction from it.
What to watch for, if you go: the blacks in the interior scenes, and whether the line on a face holds its weight when the face is small in frame. Those two tell you how heavy the clean-up hand was. Everything else about this film has been settled for three decades.
Screen count, dates, formats, scan resolution, the 35mm negative, Sentai’s role and Kawajiri’s supervision from HIDIVE’s own announcement with the theatrical dates and exclusive still via Anime Corner; the Berlinale premiere from the February announcement of the project. The restoration’s handling of grain and dust is inferred from the trailer and published stills rather than from a graded master, and is flagged as such above.