Not possible in digital ? (6 replies and 31 comments)
Hi master Roger,
I've heard a cinematographer I admire opinions that this scene in apocalypse now, where storaro captured Marlon Brando being revealed occasionally in light and obscured by pitch black shadow to Martin sheens character , couldn't be done as it is in the film if it would have been
Shot on any digital capture medium ,
Could you give your opinion on this , hoping also an opinion on the blacks and shadows rendering in film Vs digital,
The cinematographer said , if it were to be captured with digital, the shadows would have always captured lot of detail and the transition of that lighting from black to shadows to light wouldn't be same.
With HDR you can have a solid black. In fact, I add a little tone to it so that the shape of the screen is still apparent in the dark shots.
Thank you master Roger for replying ,
So without hdr it's not same as in film? Tone means here as of little of a particular hue or luminance point ?
You can achieve black shadows with digital. The rate of fall-off to black is determined by the gamma (contrast) in the shadow region which is controllable in digital color-correction.
Now in terms of contrast ratio of projected images onto a screen, standard digital projectors can't achieve what a film print does in terms of deep blacks but laser projection can.
Thank you master David for replying,
My favourite observation on understanding film Vs digital was one you said , "film has higher contrast in shadows but lower contrast in highlights , digital has higher contrast in highlights and lower contrast in shadows " I try and make this observation to practice when doing a color grading,but what I found was the latitude must be available From the sensor in the first place to get this to have a good result , right?
Do you need more detail / latitude in the shadows if your intent is to turn them black?
Color reversal (slide) film has limited latitude compared to either digital or color negative film -- there's no problem getting black shadows with slide film...
You need a wide dynamic range to match film's overexposure latitude in the highlights. In the shadows, if you want a quicker fall-off to black, you just need to light for darker shadows and to color-correct with enough contrast in the shadows, you don't need more latitude in the shadows if the goal is to make them fall-off faster to black.
Black is the absence of information, not the increase of information.
Thank you master David,
I was thinking about highlight latitude when saying film might have more latitude, the limited latitude I. Shadows of film over digital part I get .
Exposure, exposure, exposure. Detail at each end of the curve is controlled by exposure regardless of what is capturing the image.
Thank you master Roger,
But when faced with a high contrast scene like somebody in front of a strong sunset there is only limited choice in exposure right?
Also master Roger can you please mention which kind of image curve you like the most, like if you see an image and see that it has a smooth transition or not that smooth , there are varying degrees of smooth luminance and color transition possible from renders of different sensors right?
Do you prefer an s curve image or can you please say if you may , when you are presented with a flat image from a modern digital camera and you can only do curve color grading on it , which type of curve would you be pulling it to match the exposure you felt when capturing or say when you were choosing to test the digital camera image the first time , or is it some other process than a curve , I certainly don't know the way of conversion you do on the image, is it some kind of number conversion for each luminance range.
if you may , please let us know how to understand an image from a flat to rec 709 conversion and let it be natural , as you know there is lots of tools available for image manipulation these days and one often find themselves lost played by the tools options to be dictating their image , once a while.
Thank you .
I think Roger is saying that the main thing is lighting and exposure -- the details of the particular conversion LUT from log to display gamma (Rec.709 or P3) is not as important, you don't need a specialized LUT just to create images with good contrast and blacks.
Thank you master David,
What I really wanted to know wasn't much about Lut , but the preferred curve of an image , I'm sure that Roger and you , All great cinematographers can see an image for if it is overprocessed or not having a natural curve , so going back to the curious question , If you could only color grade from a flat log image using only curves adjustment , not RAW , which curve would you be pulling .
That's right. We always work in P3 but that is not essential. I suppose, you could work without a LUT and time your final images from the RAW file.
Thank you Master Roger,
If you may ,
Going back to the curious question, if you could only grade from a flat log image, not RAW, which curve would you be pulling ,
And can that curve render an image similar to the one in apocalypse now, or assassination of Jesse James train robbery scene shots. under same lighting conditions and same exposure.
What sort of answer are you expecting? Is there a number value to describe a curve (technically it is a gamma but even that value only describes a straight line, not a curved line) -- do you want a photograph of a waveform monitor where the camera is pointed at a grey scale so you can see the curve? Basically there are standard gamma conversions from ARRI Log-C to Rec.709 or P3 gamma that you work with as a starting point, then you adjust that in the shadows, midtones, highlights from there, maybe you save that adjustment to the standard as a new LUT for monitoring purposes on set.
ARRI Log-C, which is what an Arriraw file is usually converted to for color-correction unless you choose to use the ACES system, was designed by ARRI to emulate the gamma curve of color negative film, so it can be color-corrected in a similar manner to working from a Cineon Log file from a scan of a negative. How much adjustment you want to make to the shadows, midtones, or highlights after that is a creative choice in the D.I. whether you shot film or digital. I think you are over-thinking this technical issue -- shoot the image as well as you can for the look you want and then color-correct it to finish that look -- stop worrying about whether the gamma curve exactly matches Vision-3 5219 or something like that. The point isn't to recreate film, the point is to shoot images that tell a story with the mood you want.
If you really want to study gamma curves, you'd have to start reading documents like this:
There are companies that promote LUT's to convert ARRI Log-C to Rec.709 with a closer match to color negative gamma curves, but remember that you could just do this in color-correction manually, the simplest thing would be to just shoot a side-by-side test with film and an Alexa:
I just want to add that I've never explored any "film look" LUT's, I just shoot on the Alexa and light and grade for the look I want, I'm not thinking of matching film exactly down to some scientific level. That would only be important if, like some VFX movies do, I had to shoot Alexa and 35mm film side-by-side for a scene and intercut them and have everything match. You might read this:
Thank you master David for the resources to read .
This thought of matching exactly to film came to mind because of the first question I've asked , A cinematographer I admire very much said that film still is the superior medium and for example he pointed to the scene in apocalypse now and said that kind of magical lighting is hard to achieve in digital and probabily not possible at all .
Thank you again for the resources to read that you pointed on gamma and curve.
Two great cinematographers can have opposite opinions on technical things and both can be right and wrong because we're talking about artists.
Personally it's a matter of degree, not a binary answer. Modern cinematography using digital cameras is perfectly capable of creating moody, high contrast images with deep blacks, rich color, etc. that is comparable to photography of the past. However, it is not EXACTLY the same as shooting on 1970's 100 ASA 5247 pushed one-stop, etc. It can get close with some work if necessary but you're still talking about photochemical subtractive-color technology rather than digital additive-color technology. The question is why that matters if you can create good images today, unless you are involved in something unique where you have to exactly match a film scan within the same scene.
Thank you master David ,
I get what you're trying to point out.
I've to read stuffs to understand this better.
There are many still photographers who have created similar images to the one you post and have done it using both film and digital. There is also a painter or two. To say it could only have been created on film is a stretch. As David says, what is most important is why you are creating such an image in the first place.
Thank you master Roger, could you mention the name of the still photographers please and another quick question,
If you had to shoot with a camera that has log capability like a Panasonic camera which has v log instead of arri log c , the sensor performs well without noise on ETTR (expose to the right ) exposure method
Would you still set up the proper exposure or do ETTR ,
I'm asking since you said exposure is such vital but the digital cameras , force the user sometimes to consider it like metadata , like do ETTR ON every shot and proper balancing later on coloring stage For a noiseless image. Doesn't that destroy the art of exposure.
I don't think the ETTR method works for cinematography because you have to match the noise level for the overall scene, not shot by shot. Let's say you have a master wide shot where one actor moves to a dim corner of the room and is 3-stops underexposed, and another actor is next to a bright desk lamp, and the third has sunlight hitting them and is 2-stops overexposed. You may pick an exposure that works for the wide shot but are you really going to overexpose (to just below the clip point) the actor who is supposed to look underexposed? Let's say the ETTR method means that the actor is 3-stops overexposed without any clipping -- then when you make them look 3-stops underexposed in post later, that's bringing down the actor by 6-stops compared to the coverage on the other actors in the scene. And the actor standing by the bright desklamp had to be, let's say, underexposed by 3-stops to hold detail in the bulb of the desk lamp and now in post you are having to brighten the actor to the level you actually wanted to expose them at. So that actor is now noisier than the actor you darkened by 6-stops who was standing in the shadows. None of that makes sense for cinematography where continuity in exposure and mood and contrast matters more than exposing to get maximum noise reduction on each set-up. That only makes sense for still photography or a one-shot scene. And even there, what happens when you move the camera and the actor is moving through different light levels? What happens if in the shot, the lights are supposed to dim? Are you going to slowly open up the f-stop to counteract the dimming effect so that the shot stays ETTR? And what about all the time you are now adding to color-correction in post because your exposures are all over the map and do not indicate the final look?
In cinematography, it's better to test the camera for an ISO level that gives you a noise level you can live with, and then light and expose the scene for how you want it to look, and match the coverage so that the dailies when edited flow together more or less with the correct look, sort of the Gordon Willis ideal of shooting the scene so that it all can be printed at one set of printer lights with no corrections. Sure, occasionally the principles of ETTR and setting exposure below the clip point come into play in extreme contrast situations in uncontrollable situations or VFX shots where the VFX people want the plates to be exposed as fully as possible (like the DFN sequence in "Mad Max Fury Road" was).
Thank you master David,
Yes now I get the point, can you also say what is the davinci resolve equivalent of printer lights , is it offset .
I have no familiarity with that camera or work flow.
ETTR (Expose To The Right) is a digital still photography technique of setting the exposure so that on a histogram, everything is pushed to the right side (overexposure) but just below the clipping point, to minimize noise and maximize shadow detail. It was promoted as a technique in the early days of the Red camera when people were having noise problems but it has limited applications in cinematography unless you are both the cinematographer and the colorist and you're shooting things like stock footage, or VFX plates, etc. For narrative work in scenes with coverage, it doesn't really make sense.
ETTR also gets recommended by marketing types so that can say "sure, rate our camera at ISO 2000 -- as you as you overexpose everything to just below the clip point to keep the noise down!"
Oh yes master Roger I was mentioning what master David said , and got clarified , Thank you,
One more question on this post , if I may ,
Are your exposures spot on when you are doing digital capture , or do you make the shot slightly darker or brighter like incremental gamma , gain corrections like available in davinci resolve, asking because those small incremental steps are not like standardized like aperture , 2.8, 5.6 etc.. ?
And to cap of this post ,
Do you think digital cinematography had made the process more simple or complicated , after reading Steve yedlins film Vs digital debate , which master David had shared for reading , Steve yedlin mentions that it's the cinematographer that needs to update and understand the tool better and describes how he understood the film characteristics and added algorithms to replicate that film can be simulated exactly in digital.
Do you think
Digital cinematography made the process more complicated or not.
Because you've dealt best with both .
And do you think a per day shoot costs less on digital than on film in a full fledged production.
Considering the enormous data storage and else requirements..
Cinematography is as complicated as you want it to be. Richard Crudo once said to me that some DPs like peeling back the layers of the onion more than others.
Yiou can shoot digital like film: pick an ISO, pick a basic LUT, and then light and expose as best you can for the look you want, get it as close as you can, tweak in the final color-correction.
TIME is the biggest challenge when shooting, you do whatever works in the most efficient manner. So even IF you had a DIT/Colorist on set and a great monitor set-up in a dark tent, you don't want to spend a lot of time using DaVinci Resolve to finesse the look of each set-up.
Thank you master David,
This advice to keep it simple if you want is , what I'm taking , God bless you and Roger and all the great Artists who keep an unflinching eye on their art, I was a bit lost after seeing the interview of the cinematographer that I mentioned , that's why started this post, but learned a lot , yes
Master Roger it was a stretch to say that this kind of image is not possible in digital just watching 1917 for idk how many times I've seen it . Already watched it multiple times , should have taken notes , there are lot of scenes with the kind of blacks.
Thank you Master Roger and Master David , keep the unflinching eye and guide us through when we are lost , always greatful for this forum.
If I may add to this topic, I have a Sony camera that shoots LOG and I've followed the ETTR method numerous times with one result:
It's never a good idea for when you want to learn lighting, it basically gives you the whole scene and you're happy you've got everything until you hit Davinci Resolve, then you find out oh I need to hide this this and that, then you think why did I expose this part anyway?
Choices must be made you can't just record everything and decide later.
Maybe technically it works, but for the decision workflow of film making it does not. In my own personal experience of course!
I've shot one short film myself, I tried the LOG format when I was testing then I decided to shoot in standard Rec709 and do very little adjustments in post, some of them where contrast ratio adjustment and saturation. I was stunned by the results and by how powerful it is to take a decision on a take and stick with it. It makes you think more!
People are talking about the Good old days of film and how film can't be matched, I think maybe the decision making process is what cannot be matched.
Thanks sheriftolba , that's also my take away from this . Keep it simple if you want it to be .