Forum Replies Created
-
Replies
-
Normally I’d agree that there should just be one aspect ratio for a movie — I don’t recommend, for example, showing an old 1.37 Academy movie in any other aspect ratio.
This is a bit different for a couple of reasons — first is that IMAX traditionally (I’m talking about the original 15-perf 65mm format projected in special theaters) was always designed to be composed with an excess 1/3 space at the top, extra headroom, because the theater is designed for your eye to go to the middle of the very large and tall 1.43 screen. To some degree, the area at the top of the screen is like the sky above your head, extending to the edges of your vision unless you sit way back in the theater. Not that Nolan seems to compose his IMAX photography with that much headroom but on the other hand he composes with enough headroom that the blu-ray versions of “Oppenheimer”, “Dunkirk”, etc. show the IMAX scenes in full 1.78 : 1 HD cropped from 1.43 and the headroom looks fine.
Second, the Digital IMAX screens are 1.90 : 1, not 1.43 : 1 and more people are going to see that version rather than the 15P 70mm prints.
Third, Nolan likes to release also in 5-perf 70mm 2.20 : 1 and 4-perf 35mm 2.39 : 1 prints.
Clearly he is fine with shifting aspect ratios in the past although this movie was only shot in IMAX, not a mix, so he could have simplified things a bit and picked 1.85 for his non-1.43 IMAX showings as a second aspect ratio rather than 1.43, 1.90, 2.20, and 2.39 in this case. 1.85 is very close to 1.78 which is what the HD blu-ray versions would be anyway. The 5-perf 70mm could have been side-masked to 1.85. But I think he’s being a bit of a showman wanting to make each version special in the theaters and he seems fine with the reframing.
I would add that you can tell it is a very large soft light close to her because the shadows are almost horizontal so it is eye level and yet she is looking down and you can still see the light reflected in her eyes — which suggests a soft light that is almost her height.
A Rec.709/P3 display LUT on set works most of the time if you want straightforward, standard image contrast and saturation. Sure, you may prefer a personal LUT with a tweak towards slightly deeper blacks, etc. If you have a DIT on set, they can do adjustments like that using ASC CDL values on top of the basic LUT.
The problem becomes when you want a more stylized final look in terms of contrast, like the look of a higher-level ENR or skip-bleach print let’s say. Then it would be helpful to have a LUT with that look so your lighting is taking it into account (or a DIT that can adjust the LUT on set). That’s when it becomes more important to create a LUT in prep based on tests. Otherwise, if you can’t create that LUT, you may have to rely more on your light meter as a guide… but it does beg the question if that display image on set is inaccurate, what is your plan for dailies, which usually use the same LUT as on set? Or is this a one-off sort of thing like a 2-day commercial or short film where you color-correct the dailies after the shoot is over?
Ideally you are making the decision based on story reasons — does the background add any important narrative detail or set the appropriate mood for the performance to play against, or is it distracting to the story and performance (keep in mind that a hot window can also be distracting sometimes.) And if (a big “if”) realism is important for the story, does the balance between the interior and exterior feel believable, natural?
But the truth is that often you are also taking into account logistical issues. Will the real background keep changing throughout the day as you film coverage, creating matching issues? Is the background correct for the period of the story (maybe there is some anachronistic element that needs to be de-emphasized)? Is the background correct at all (maybe the location is a cheat for another location)? Can the background be shot at the correct time of day for the story or will you be filming past sunset into dusk for a day scene? How much control do you actually have over that background (maybe you’re in a skyscraper on the 50th floor and the windows are too large to gel)?
And then there are similar practical considerations. Will the background always stay in the background or will the actor move to the window and be followed by the camera, making the background more prominent?
Break it down…
Fall-off rate: get the light as close to the subject as possible because the fall-off rate becomes less steep the further away a light it (inverse square law).
Warmth: use a light that has a color temperature lower than what the camera is set for.
Softness: the larger the source relative to the subject, the softer the shadows. A bounce surface or a diffusion frame is the source in this case, not the light hitting it.
Besides those three things, you have angle of the light (in this case, from the side and slightly low) and the contrast (in this case, no fill and maybe negative fill as well, combined with the fast fall-off, combined with your camera’s gamma settings and black level for display/conversion from log/raw.)
It’s a cult film now, I’ve had many young people on set, in production offices, at trade shows, etc. come up to me and express their love of this film.
Unfortunately you’ve run into the major issue with tenting a location for filming night scenes during the day — when you need to put your lamps far away and high enough to get a good moonlight effect. Obviously the first solution — besides going back to shooting those shots at night with no tent — is to build the tent as large as possible!
Another possibility is to shoot day for night, perhaps even just blacking out the view through the windows with a wall of black but shooting light from over the top of the wall. You’d have to avoid the real sun coming through if you wanted more control, or use the sun as the moon. But inside you’d have a hard time balancing with dim sources like flashlights and candles… (and heavy ND gel on windows creates its own issues/problems.)
You could see if a mirror mounted at the top of the tent, if it is tall enough, could be used to bounce a source on the ground to get a better beam. You could also try using Lekos as your moonlight since they need less space to create a beam.
Experience sort of teaches you how, let’s say, a chest-up close-up is going to feel on a big screen as opposed to a neck-up close-up… but also most movies even today shoot tests in prep and watch the tests on a theater-sized screen in a D.I. suite so they can judge the results accurately, so there is some feedback from the start about screen sizes.
But the truth is that each movie sets up its own language in regard to shot size that the audience quickly gets used to. Occasionally when you shoot super-tight close-ups and see them later on the big screen, you sometimes feel that perhaps it’s too tight, too much detail, seeing every nose hair and pore… but hopefully you felt that way when filming the shot on set and had time to back off if necessary.
What is the alternative to “lighting to the display LUT”?
Because lighting by viewing a flat log image — i.e. no display gamma LUT applied — isn’t necessarily going to give you more accurate results. Log images usually place the highlights rather low in the IRE scale in order to hold bright detail in the overexposure range, so if you light by looking at a log image, you tend to overexpose it in order to make the whites look white and the highlights look bright enough. Even in the days of photochemical post, you lit negative film based on how it would look in a contact print, which is a form of a LUT, limiting the dynamic range to what the print stock can display, which is why you lit for that range, not for what the negative could capture.
For the most part, using a P3 theatrical display LUT is fine if the final result is both for a film-out and a DCP. The Print Emulation LUT is mainly just to make sure you don’t push colors in color-correction outside of the reproduction capability of Vision print stock, it’s not particularly a “print look” LUT, it’s more of a technical limit to color space. It would be hard to think of a lighting situation where you would need to work within that limit, you would just have to keep in mind in deeply colored lighting situations that the colors in a film-out to be printed will be limited by what the print stock can reproduce.
The differences become pretty minor other than output and spread once you put them through a diffusion frame.
The problem with anamorphic lenses is that if the squeeze factor is 2X and your final unsqueezed aspect ratio is 2.40 : 1, then the sensor area used is very squarish — 1.20 : 1. Since the LF sensor is 25mm tall rather than 18mm, if you record 4.5K 3:2 Open Gate, 3096 pixels tall, you gain some vertical area for stabilization (plus extra horizontal area that you won’t need as much for 2.40 anamorphic). I wouldn’t record 4.3K 16×9 — you need to record more height, not a 16×9 area.
