Splitscreens, reframing, and roto in '1917' (5 replies)
Roger, because of 1917's effect as a seamless single take in real-time, did you pursue any 'editing within the frame' to make up for the inherent lack of coverage? Not just stitching the different shots together, but combining different takes of the same scene through seamless spiltscreens and rotoscoping?
You seem to think that we made a series of shots and stitched them together after the fact. In reality every section was thought out in detail and every 'blend' was shot with as much precision as possible. Yes, sometimes we used effects work to enlarge an outgoing frame so that it matched the incoming frame more exactly than it was possible in camera but these were, for the most part, minor adjustments. I say for the most part as there was one transition that needed a little more digital adjustment than all the others.
I should add that half way through the shoot we saw a 'cut' of what we had shot to that date and the transitions seemed seamless although they were all direct cuts at that point.
I've read about and watched BTS of the vast and painstaking planning that went into the production. To my mind these kind of editorial strategies don't preclude a similar level of precision in prep and the shoot. 'Parasite' is a recent example of a film that had very little coverage and so they did make extensive use of blending together different takes within a shot.
So in the final film of '1917,' the editorial 'selects' were a matter of choosing the best beginning-to-end take of a setup? For anything that didn't require too big a reset (in the way of physical effects, extras, etc.) was there anything like an average number of takes?
As far as I know there was only one blend made in the middle of what was the original full length take and that blend was made on an explosion. As the blend points were very specific, Sam would choose his preferred take and we would match to a specific frame for the continuation of the action. That required matching not only actors positions relative to the background and the frame size but also the speed of the movement of the actors and the camera. For these and other reasons using 'pieces' of one take, or shot, was not really an option.
We often shot a large number of takes, up to as many as 35 in one instance, but we likely averaged around 16. We rehearsed a large number of times, so that the technical details could be ironed out before the actors gave a full blooded performance or we blew away effects that took time to reset. For that long run along the trench line, for instance, we shot one take in the evening before the light failed, and a further three or four on the morning of the next day. For the final shot of the film we made, maybe, 25 takes but the first one is the one that is in the film. We might, and usually did, get a 'preferred' take early on but, as each shot was so hard to perfect, we usually went for a few more to see where things would go. I think that is the norm on any film.
"That required matching not only actors positions relative to the background and the frame size but also the speed of the movement of the actors and the camera. For these and other reasons using 'pieces' of one take, or shot, was not really an option."
I understand the need to have a clean frame in order to match one shot to the next, but what are some of the other reasons it wouldn't have been possible to employ split screen or roto outside of the 'end pieces' of a take?
Well, I wouldn't say that there weren't places were an end frame wasn't morphed into the next and there was some roto work as well but you still have to have the speed and size of end of one shot match the next. You can't roto out a background and roto in a new one just like that as you then have to backtrack that new background into the previous one. As I say, there were all sorts of subtle tricks used to smooth out o couple of blends but they were exceptions.