Jump to content

Exposure in low light (Amateur who need to clarify)


Filip Aladdin

Recommended Posts

Hi all.

Im not that experienced so maybe this is a really dumb question, but Im confused about exposure in low light situations. 

Let me try and explain this easy.

Im usually navigating my exposure with False color on Alexa. And I was shooting a night scene where the actor is drawing on a "light-table" and the rest of the room is dark. Because it is night and she is drawing in a dark room. Nothing strange about that. (you can see on the first attached picture) I wanted most of the room to be dark and black in the end, after post. So I lit it like that. I watched false color and the parts that I wanted dark was purple and blue on false color. Fine I thought. But its noisy and grain in those parts. And im not really sure how to avoid this.

Should I have lit the whole room so the parts that I knew I wanted black wasnt black at the time of the shooting? And then make them black in post? Would that get rid of that noise because of the time of the filming it was information. If someone could explain to me the relations between on set exposure and after post, that would be awesome. But maybe its not that easy and im just to inexperienced. 

I wonder for my self when I watch "real" movies and night scenes where there are parts in the frame that is completely black and is totally clean and no noise. Did they film it like that when they exposed?  Were those parts "purple on false color" when they shot it. Or was it 20% and they fixed it In post. (picture 2) 

If I know I want something black in the end product should I film it that way with no information on the sensor or should I light it and in post make it darker. 

I hope you understand what im looking and sorry for not using the right technical words. 

Best, Filip 

 

 

 

lighttable.jpg

example 2.jpg

Link to comment
Share on other sites

  • Premium Member

Good question. It depends on: 

1. What EI or ISO setting did your shoot at? 

2. Did you shoot Log and use the standard Alexa Rec.709 LUT in post? Or did you do a color grade where you lifted the blacks? 

If you just used the standard Rec.709 LUT to convert your Log footage, then the noise should be normal for that EI. If it’s too noisy for your taste, then simply lower the EI and expose brighter to get less noise. 400 EI is generally very clean on Alexa even for very dark scenes. 

However, I notice there is no pure black in your frame grab, the darkest object is the letterbox and the shadows are higher than that. If you have lifted the blacks in post, then there will always be more visible noise. Noise hides in the shadows, so by lifting the blacks you are making it more visible. In that case, I would suggest adding more light into the shadows when shooting to get closer to the final image while viewing the standard Rec.709 LUT, so you don’t have to lift blacks in post.

Another option would be to create a lifted black Look using ARRI Look Creator that you can view in-camera to see something closer to the final intended look - this would give you the same amount of noise as lifting the blacks in post, but you can see it on the monitor/scopes and it should be easier to judge while shooting. 

  • Upvote 1
Link to comment
Share on other sites

Thanks for a fast answer Satsuki!

 I realise its a bad formulated question and the picture above is not relevant cuz its shot on a black magic 6k. And I don't event think its graded. 

But maybe this is a better formulated question:

What is the "right" way to do for the best end result with least amount of grain and noise in the blacks. If you know that you want crushed blacks in a frame. Do you crush them when u shoot it. 0% exposure on site. Or do you lift them and then crush them in post?

Link to comment
Share on other sites

  • Premium Member
1 hour ago, Filip Aladdin said:

Thanks for a fast answer Satsuki!

 I realise its a bad formulated question and the picture above is not relevant cuz its shot on a black magic 6k. And I don't event think its graded. 

But maybe this is a better formulated question:

What is the "right" way to do for the best end result with least amount of grain and noise in the blacks. If you know that you want crushed blacks in a frame. Do you crush them when u shoot it. 0% exposure on site. Or do you lift them and then crush them in post?

The advice is still the same with any camera, shoot exposure tests with the final LUT or grade applied to see how much noise (or film grain) is acceptable to you in the dark areas at different ISO levels.

Any time you have lifted blacks, like in your Blackmagic 6K frame, you’ll see a lot more noise in the shadows. So if you know you’ll be setting your blacks in post to 0 IRE, then 0% exposure should be fine. But it will depend on the specific camera, as some are noisier in the shadows than others and may need to be crushed more. 

Link to comment
Share on other sites

  • Premium Member

EFB6A611-C7DB-4650-BA90-A800CE4538AC.thumb.jpeg.356f2451698b0dd626831584c61fb487.jpeg
 

I just did this quickly on my phone so it’s a bit rough, but if you crush the blacks on any image you’ll see less noise. There’s no need to change lighting or exposure if this is the result you want. If you want more shadow detail like you had before, then it’s better to add more light into the shadows on set to keep that detail out of the noise floor. 

Link to comment
Share on other sites

  • Premium Member

There's no reason for blacks to be noisy. Black in digital is just zero signal.  Trouble is that with underexposed footage your shadow detail is often noisy and if you set the blacks to be pure black with no noise, then the shadows look too dark or look artificially crushed.

But you should make a scene as moody as you want. As Satsuki says, either you need to work at an ISO setting where the shadows aren't noisy and/or you need to work with a basic correction / LUT in mind where at least the blacks are darkened so that you will add more light for the level of shadow detail you want -- in other words, you are either using a lower ISO overall or you are essentially lowering the ISO for the shadows, either way you need more exposure to reduce noise. But shadows detail isn't the same thing as black, which mean no detail, it's black!

The simpler thing would be to just use an ISO setting that gives you an acceptable noise level and then light it for the amount of mood you want, don't be afraid of shadows or blackness.

  • Like 1
  • Upvote 1
Link to comment
Share on other sites

Filip, I experience the same thing all the time. 

My final conclusion is to avoid having anything in your frame falling below 10% IRE (or 128) when shooting with any digital camera. Even if those areas are intended to be pure black in post. So expose with nothing purple in your false color monitor. And then in post bring them down from 128 to near 0. 

Another issue that I observe but cannot fully understand is that when you underexpose with any digital camera the noise is not limited to the deepest shadows. You see more noise in midtones and even highlights. Even though those areas were well exposes. It's some kind of a "leakage" effect. I don't really understand the science of that. But I see it all the time. Perhaps someone has an explanation. 

The point is underexposure yields very bad results in Digital (and so is overexposure, Haha). You need to get everything in the middle and then grade the contrasty look you want in post. 

Link to comment
Share on other sites

10 minutes ago, Raymond Zananiri said:

The point is underexposure yields very bad results in Digital (and so is overexposure, Haha). You need to get everything in the middle and then grade the contrasty look you want in post. 

This is simply not true, for all the reasons that Satsuki and David have explained above.

Link to comment
Share on other sites

6 minutes ago, Stuart Brereton said:

This is simply not true, for all the reasons that Satsuki and David have explained above.

I believe from David's statement that he validates my point. The issue is the next tonal value above pure black. What he calls "shadow details". Those are always rendered best when you overexpose slightly for pure blacks (above 10% IRE). The whole tonal gradation in the shadows will then become better, plus less noise in the overall image. 

Link to comment
Share on other sites

  • Premium Member

Not quite. It’s not about how you expose black, black is black. Some black areas can be a dozen stops below key.

And if you overexpose slightly overall then that’s the same thing as picking a lower ISO, either way you’re giving the sensor more light to reduce noise.

There’s no reason to avoid having areas of the image that fall near black, you don’t have to lift everything to a higher zone. It’s natural for parts of a frame to fall off. If the image has noise it’s because you’re working at too high an ISO. Or you are lifting your shadow detail up in correction to essentially a higher ISO in those areas. Or you are working from a log recording and not bringing the shadow detail and blacks down to normal Rec.709 gamma levels.

So you need to use a lower ISO or use a LUT with deeper blacks to keep the noise down (and then perhaps compensate when needed by adding more fill). Either way.  Or both. 

Link to comment
Share on other sites

  • Premium Member
1 hour ago, Raymond Zananiri said:

I believe from David's statement that he validates my point. The issue is the next tonal value above pure black. What he calls "shadow details". Those are always rendered best when you overexpose slightly for pure blacks (above 10% IRE). The whole tonal gradation in the shadows will then become better, plus less noise in the overall image. 

It’s not really true to say a digital image is always noisy in the shadows. How much noise you get in the lower values really depends on where you place your middle grey exposure. 

It is true that quite a few digital cameras can be fairly noisy in the lower stops if you’re shooting in log at the manufacturer’s recommended base ISO value, because they are biasing their dynamic range in log towards capturing more highlights. Hence the middle grey tends to exposed much lower than for Rec.709, (38-42 IRE range) and the shadows fall into the noise floor. 

However, since most digital cameras remap middle grey based on the camera’s ISO or EI setting, you can counteract this by lowering the ISO value and increasing the exposure to compensate. This requires some testing in advance to discover what EI/ISO value produces an acceptable amount of noise in the shadows for you.


TL;DR. Use a lower EI/ISO value and compensate with more exposure for less noise in the shadows. 

  • Upvote 1
Link to comment
Share on other sites

6 minutes ago, Satsuki Murashige said:

It’s not really true to say a digital image is always noisy in the shadows.

I'm not saying that at all.

What I'm saying is that the bigger portion of the image that falls below IRE 0 when exposing, the lower the quality of the image in the remaining, adequately exposed, portions of the image. I really don't know why that is. But it is my experience as such. 

That is the whole argument behind the "exposing to the right" method. Push everything up as much as possible (short of clipping the highlights) and you will get a better overall image (once graded accordingly in post) than exposing correctly for middle grey. 

Now a scene in bright sunlight would have unavoidable patches of deep, underexposed blacks. That usually is not a problem. 

But, with scenes such the first one above, with large areas of the frame very close to pure black and a desk light in the middle, results with digital are never truly satisfying (unless you fill the shadows with lights and then bring them down later in post).

That is just my experience. Maybe I'm doing it all wrong ?

Link to comment
Share on other sites

  • Premium Member
26 minutes ago, Raymond Zananiri said:

I'm not saying that at all.

What I'm saying is that the bigger portion of the image that falls below IRE 0 when exposing, the lower the quality of the image in the remaining, adequately exposed, portions of the image. I really don't know why that is. But it is my experience as such. 

That is the whole argument behind the "exposing to the right" method. Push everything up as much as possible (short of clipping the highlights) and you will get a better overall image (once graded accordingly in post) than exposing correctly for middle grey. 

Now a scene in bright sunlight would have unavoidable patches of deep, underexposed blacks. That usually is not a problem. 

But, with scenes such the first one above, with large areas of the frame very close to pure black and a desk light in the middle, results with digital are never truly satisfying (unless you fill the shadows with lights and then bring them down later in post).

That is just my experience. Maybe I'm doing it all wrong ?


Yes, but: 

1. What camera are you using in this situation?

2. What EI/ISO setting?

3. Are your shooting log, raw, or Rec.709?

4. If shooting log or raw, are you viewing a Rec.709 LUT on the monitor? 

5. Where is your middle grey reading on the waveform?

6. In post when the footage looks noisy, are you viewing log footage with the same Rec.709 LUT applied? Or are you grading the footage? 

These things all make a difference.

In general, I agree with you that lower values on all digital cameras will be ‘thinner’ and noisier than the midtones and highlights. That’s just how linear encoding works, each increasing stop of light has twice as much data. So between pure black and Zone 1 (first hint of visible shadow detail), there are only 2 bits to represent those tones of grey. I’m sure a more technical person could explain in detail how that works. 

And yes, exposing to the right and bringing down in post does work to mitigate those issues, though you are now introducing other issues such as consistency of exposure between shots in a sequence. Generally, lowering the ISO and increasing exposure does the same thing while maintaining exposure consistency, which is what some of us are recommending.

But really, if you are trying to get an image like this: 

#1.

4 hours ago, Satsuki Murashige said:

EFB6A611-C7DB-4650-BA90-A800CE4538AC.thumb.jpeg.356f2451698b0dd626831584c61fb487.jpeg

 

Then you don’t really need more exposure than this to achieve clean shadows: 

#2. 

On 12/16/2020 at 9:26 AM, Filip Aladdin said:

lighttable.jpg


On the other hand, if you want to achieve a final look like #2, then you need to start with a bit more exposure. In this case, I don’t think ETTR will help because if you increase overall exposure any more, then the practicals will blow out. So you need to add more local fill light to crush down. But not that much more, just a little bit. At least, that’s been my experience.

Link to comment
Share on other sites

Very valid point Satsuki:

I'm assuming of course that we're talking about log footage and that the waveform in the monitor is based on the log curve (even if a lut is applied in the monitor for viewing).

I'm intrigued however as to why you and David bring the ISO setting up as that critical of a factor. Maybe I'm wrong, but for me ISO setting in raw recording is quite irrelevant. It is a tool for exposure for sure. But, the camera will always record in its native ISO and the different settings are only for monitoring. For me, what I see on the waveform monitor is the key. And, while different ISO settings would result in different waveforms, the clipping points remain the same for all settings, and that's what counts. I always think of digital (and I don't think I'm alone on this) as "stay away from either clipping points as far as possible and don't worry about where middle grey is, as all this would be sorted in post anyway". This is the assumption I operate on and would love to hear from you why it's not right. You seem to value being accurate on middle grey. Is it about consistency from shot to shot? My camera by the way is a Sony F65.

Edited by Raymond Zananiri
Link to comment
Share on other sites

  • Premium Member

Even if the ISO setting is metadata, it affects your exposure decisions and thus the amount of light the sensor gets. So ultimately it makes no difference if you are recording raw or log in this case, generally you have the same exposure information to work with in color-correction (however, color temperature is usually not baked into raw.)

The problem with basing exposure shot by shot on the clip point, like a still photographer might using the ETTR method, is that cinematography is not the art of single-shot scenes, we shoot multiple shots for a sequence and we don't expose each one individually for optimal noise level.  So continuity of look within a sequence trumps maximizing exposure decisions to reduce noise.

You could have a wide master where one person is standing in an area that is two or three stops under key and someone else is under a light that puts them one-stop over key. Plus you may have a bright window or practical in the frame.  When going into coverage, you don't re-expose each set-up differently based on where the clip point is, otherwise the the coverage of the person in the dim underexposed light, if they had no whites in the frame, might end up being exposed two stops or more over key (just below any clipping) for maximum noise reduction. On the monitor it would look ridiculous and you'd be forced to explain to the director why their dark scene looks so bright. And then each set-up would vary in noise floor because you'd then in post have to recorrect everything to match the levels in the wide master. 

Hence why choosing a working ISO that gives you an acceptable noise level and then lighting each set-up for the mood and look you want is the more common method in cinematography.  

Otherwise you get into a situation where each set-up is exposed for technical reasons rather than artistic reasons and then needs shot-by-shot color-correction to make it look the way it is supposed to, which is often not possible for dailies.

Link to comment
Share on other sites

  • Premium Member

What David said. ?

The ISO setting in most digital cameras acts as a preview LUT, so by itself of course it does nothing to affect exposure on the sensor. What it does do is place your shadows lower, so if you want more shadow detail then you are forced to add more light, encouraging you to keep important detail out of the noise floor. Of course, if you don’t need to see any detail in the deep shadows, then you don’t need to change anything at all and your shadows will remain clean. Hope that makes sense.

  • Like 1
Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...