Shooting the Shit – the What, Where & How
This is a donated still from footage*. I got no information about where it’s from or the context, and it’s been stripped of metadata. I’m going to make a few assumptions here.
What: Intimate scene between a couple or people who are going to become a couple in either a film or TV show. There’s some intense eye contact and intense cleavage in this shot, and the lighting is dim and warm. The Production Design has warm, muted hues.
Where: Bedroom/hotel/motel, either a set or on location. It’s most likely on-location based on the Production Design and how the lighting is set up.
How: I can make a guess that there is a key light off to right of the frame, an overall fill, and a light set up to the left to look like the overhead light in the bedroom. Or, the bulbs have been switched out in the actual light fixtures to something more powerful and the right temperature. This is definitely not a mix of room lights with set lights because there is no colour/temperature mixing or variation in the breadth of the light (eg. a cheap, greenish fluorescent won’t reach as far or be as intense). A light on a nightstand (behind the male actor’s face) is either part of the lighting or made to appear as a source. There may be a small accent light or bounced light hitting the actress’ eyes as well.
All I know is that this was shot on a RED Epic, evidently underexposed, and in 4K resolution. This is framed pretty well with good placement of actors; there’s an effort to remove the bed post (I wish it wasn’t there at all) in the foreground by placing something over it that matched the colours in the Production Design. Something like a big, blurry, brown lump in the foreground would absolutely ruin this frame.
Now, we can tell this is underexposed…because it’s dark as shit…and the highlights are stuck in the mid values. Sometimes, footage looks strange because it’s shot in a certain camera profile that maximizes the amount of data the censor is in-taking. This is not the case. If I had the RED Raw file, I could use preset Looks like LogFilm to speed up the clean-up process. From my experience, it’s not really necessary because I can get a similar image on my own. However, it does help with roll-off of the blacks/whites. I am using Adobe Speedgrade and don’t have these though.
Note: Noise reduction comes before grading in the colouring process so that the footage is the cleanest possible. It doesn’t really matter in Adobe Speedgrade because of the layering process.
So, the very first step in dealing with underexposed footage/photos is to bring the data into the appropriate range before doing anything with the actual values:
- Shadows – increased Contrast and mostly brought everything down to reduce the crushing; the data range for the shadows is set lower so it is not affecting mid-tones.
- Mid-Tones – increased Offset and Gamma, and marginally increased Gain to brighten; added 0.02 Contrast.
- Highlights - Offset increased significantly, Gamma and Gain increased; the data range for highlights is adjusted to read more mid-tones to pull out highlights. Otherwise, in the data, this image technically has only mid-tones.
The reason the blacks are not in their fullest range is because this is a backtrack screenshot – the shadows are manipulated by a vignette I added next, and I had to turn them up again in this grade to keep everything from being crushed. As this is a warm image, the dominant channel is Red. The brightest specs of highlights max out in the Red channel.
A subtle darkening vignette is applied to the edge as a stylistic choice and to alter the composition of the image. As is, the room is unevenly lit in the image. The subject looks too dark despite proper lighting on her face. This is either at fault from the lighting or from a vignette happening inside of the lense. The darkness in the applied vignette balances out the image by redirecting attention to the actress’ face. It also brings down values to happy zero, with the adjustment of a PaletteCut layer.
Shitty Noise Reduction
Zooming in, there is colour noise mainly in the blue channel. This type of noise is a problem because it affects the colouring of the image directly – the pixels are literally the wrong colour and add a grain texture. Unfortunately, Adobe Speedgrade does not have any powerful noise reduction features. I think that’s a big deal for a colour grading program – no one wants to be moving back and forth between software for something this basic. What I ended up doing is creating a Secondary Grade layer that literally targets the blue in the pixels and replaces it with a neutral warm tone. It’s not as corrective as it needs to be, but it at least converts the noise into grain. A heavy fxDegrain layer is used to smooth out the grain texture to something less abrasive without losing sharpness.
I’d like to use this as an example to rant about why you still need to light your set properly and expose footage properly with our new technology. Yes, RAW allows for footage to be brought up in Post and the dynamic range and sensitivity of today’s cameras can be so amazing it seems you don’t need as much light. There is still losses, like this permanent damage to the footage, that can be avoided with proper exposure and enough light. There is still a limit in all censors where, at a certain point, blacks become noisy when trying to read very low light. It wouldn’t be a big deal if the colour noise was off in the background in a corner somewhere, but it’s RIGHT IN HER FACE. THE FACE! And elsewhere.
- Overall: -0.21 in temperature (cold)
- Shadows: +0.06 in temperature (warm), -0.05 in tint (green)
- Midtones & Highlights: +0.14-16 in temperature (warm), -0.06 in tint (green)
Removing warmth from the whole image and selectively adding it back in seems to have different (better) results than just manipulating it directly – in case that seems like a strange procedure. As shown in the last shot, the image is overly magenta. The lighting is warm, and of course I want to keep the warm tone, so I make sure the mid-tones and highlights are warm while being white balanced. This is the difference between creating a warm atmosphere and just throwing yellow-orange over the whole image.
- Layer targeting some of the excess magenta to get a better skin tone.
- Layer targeting the red in the flowers on the bed (to the left) to bring back some saturation in the colour.
Finally, I adjusted the overall saturation of the image to be 1.15.
That’s it. Not necessary: if I were to keep working on this, I might use masks and secondaries to highlight her eyes more, further reduce noise, and – if this is a short or static shot – maybe go as far as removing the bedpost.
*All donations will remain anonymous, unless specifically requested before or after.