I find Sam Mendes to be a frustrating director far more often than not - he lets his actors get away with a whole lot of very easy choices, and he gravitates towards projects with the most desperately banal bourgeois middlebrow sensibilities - but if one of the signs of a great film director is to surround themselves with top-tier collaborators and trust them to do top-tier work, there aren't too many better. It's not news that he's spent his entire career in movies working with the finest cinematographers, or that as a direct result, the best thing about every single one of his films is how it was shot, regardless of whether there is anything else to recommend it (as there is very often not). 1917, his eighth feature, and the first for which he has received a writing credit, alongside Krysty Wilson-Cairns (it is based in part on stories he heard from his grandather about the First World War), is maybe the apotheosis of this tendency. To say that it is shot well isn't the half of it. Being well-shot is the film's entire raison d'Γͺtre, the beginning and end of its aesthetic program. And this is true even though it's a craft masterpiece: pretty much every major aspect of filmmaking (production design, music, sound mixing, sound editing, makeup, visual effects, costuming, and in that order) is represented by some of the best work of 2019. But cinematography is sort of the film's thing. Certainly, that is its thing more than telling a gripping story about engaging characters is.

Not that the story is bad. It's perfectly serviceable: in a trench somewhere in northern France in April 1917, Lance-Corporal Blake (Dean-Charles Chapman) is asked to pick a fellow soldier to report with him to General Erinmore (Colin Firth). He selects Lance Corporal Schofield (George MacKay), and the two then receive their orders: traveling nine miles across territory that may or may not be crawling with Germans, to provide a certain Colonel Mackenzie with new orders calling off his can't-miss assault on the German lines the following morning. For what looks like a perfect opening is, in fact, a trap, and if Mackenzie goes through with it, more than a thousand men are going to be killed almost instantly, among them Blake's brother. With that, the two men are off through one of the most notoriously deadly landscapes of the 20th Century.

The gimmick here - and I am sorry to say, but "gimmick" is unquestionably the word for it - is that all of this is presented in what appears to be two uninterrupted long takes (not one, as some people would have it; there's a "passing out from exhaustion" moment a little bit past the film's halfway point that breaks it into unequal halves). Editor Lee Smith, Christopher Nolan's guy, has stitched the constituent parts together with perfectly invisible seams, so even if you can tell where the illusion is happening if you decide that your only goal is to spot it (there are a handful of oh-so-convenient passages into jet black, fast pans, and the like), it's hard to deny that it's an awfully convincing illusion.

The reason for doing this - beyond "because it's cool" - is to drop the viewer right into the middle of the stress and panic of a war zone, emphasising through visual continuity how it feels to be always in the middle of a deathly, miserable present, where war just is; we're living through every jolt exactly as Blake and Schofield do. The mildly annoying thing is that the experience of watching the film isn't really all that different from the experience of reading about it or hearing it discussed - knowing that there is a war film that uses a fake single take to subjectively place us into the war zone gives you most of what actually watching 1917 gives you. And, obviously, the only way the experiment works is if they committed to it all the way, but like a lot of experiments, once you "get it", you can't help but wonder if maybe they could stop experimenting. One thing that Thomas Newman's amazing, punctuative score (if it's not necessarily the best work of his career, it's at least the best work he's done since WALLΒ·E in 2008) and Dennis Gassner's slightly heightened, apocalyptic production design both prove is that you can get to the same emotional highs and extreme subjectivity with much less splashy razzle-dazzle; that is, you can use cinema to show what the two men are thinking and feeling, and 1917 is already using cinema to do this very well. The long take style ends up feeling more like a straitjacket than an opportunity, and for ever moment that absolutely sings because of it - a plane engine slowly rising in the sound mix, followed by the plane showing up in the distance, followed by the plane crashing feet away from the soldiers; a late sequence in which a trench slowly appears like a white, chalky gash in the ground, and then suddenly engulfs the whole frame as the camera spins around to find itself closed in by the unhealthy pallor of the trench walls and the men within it - there's at least one moment where it's just sort of grinding along as it must, because that's what this film is. I don't even know that 1917 gets as much aesthetic mileage out of its long take as Mendes's last film, the James Bond picture Spectre, did using the same trick in its pre-title scene.

Still and all, even if the film would perhaps be better off with some cutting to help guide and shape it, the fact remains that the long take is done very well. And in particular, it provides a hell of a challenge for the great Roger Deakins, who has now fully redeemed himself for being attached to The Goldfinch earlier this year. Lighting and shooting this film must have had an astronomically high degree of difficulty: besides the obvious matter of working with the blocking and having to move almost constantly, there's the matter of how to create compositions within that movement, so that at the right moments we're in close-up, or wide shot; we're in front of the characters or behind them; we're aware of the world or we're aware of the people within it. Plus, the camera keeps going in and out of doors, as well as making it through a sunrise; while I'm sure that these are particular places where the long take was faked, at least sometimes, just the fact of shooting from a dark tunnel into a bright sunlight exterior, and using almost nothing but natural light to achieve it, is already more than we'd wish on anybody who wasn't a bona fide god of cinema. I don't know if it's the most beautiful cinematography of 2019 - though it is certainly in the running, capturing something poetic and rich even in the (literally) darkest moments - a scene of a cemetery lit by the flicker distant explosions is a perfect combination of the sublime and the hellish - but it's definitely in the conversation. And even if it doesn't necessarily look the best, it almost certainly is the most technically accomplished.

And that's kind of 1917 in a nutshell: you praise it by talking about its technique, because it's pretty hollow beyond that. This has been the limitation of Sam Mendes's career since it started: his films are impeccable and emotionally inert. It's for this reason that he proved to be such an inspired choice to direct James Bond films, but it's not necessarily what we hope to see in every project, especially not one that claims to be making the emotional appeals that this does. Here's the most succinct way I can think of to describe my experience with the film: at one point, a man is stabbed in the gut and we watch him die, and the whole time this happened, I was thinking (quite impressed) about how they showed his face getting greyer and greyer as he bled out, in real time. The notion that I might be sad or upset never crossed my mind. And hey, if anybody's going to be happy with a film because it has exquisite craft, it's me. But I really don't think that's what 1917 set out to do.